• Site location - and necessary local valuation of industrial sites.

• Proximity - to services and infrastructure (and available capacity)

• The purchaser - whether it's an owned and operated site by big tech, or venture capital funded and leased.

• The required investment to make it suitable for purpose.

Each of those are broad titles, but there's much more nuance in the detail of each; one company may be seeking a hyperscale site, which would require a huge amount of infrastructure upgrades; another may be looking at a site for a more simplistic conventional compute expansion. The same site will be worth hugely different sums to each party.

For valuers, this is a bit of a nightmare. Presently, most values are assessed through comparative methodology - finding out what other sites sold for the locality, and adjusting the value based on that. But this still introduces too much room for error and variance in valuation - a similar sized site in the same zip code area, but with ten times the available power capacity is worth a hugely different sum.

I believe there's a better way of valuing data center sites - and we can do it mathematically, formulaically, and algorithmically. If we distill it down, a site's suitability is basically the combination of:

• Site location

• Power capacity and proximity

• Connection speed and latency factors

• Proximity to water bodies and mains water pressure

The suitability of a site depends on whether each of the above fits in alignment with the purchaser's requirements for their hardware operation. And the more aligned it is, the more valuable it is to a purchaser.

The strange thing about the existing methodology is that it applies a qualitative valuation approach to factors which all have quantitative metrics attached to them. For instance, the available existing power capacity to a site will be a specific, certain amount of MW (or GW) - and yet it's only ever used as a loose indicator to inform value, without ever being tied to it.

We've got other valuation methodologies that use quantitative data to inform the capital value - income capitalization, cost approach, residual and profits methods; so we can use those as a basis to coming to a solution to a more rigid approach to determine value.

Tying all of the main site metrics together can give a scoring of a site's suitability; and then that score can be tied to the value of other local generic industrial real estate to give a specific value.

The approach: constructing a Data Center Valuation Suitability Index (DCVSI)

To come to a suitable figure for assessing the suitability and value, we need to factor in the following attributes to enable a fair assessment between sites:

Hereafter, it's quite a simple process:

1. We gather the specific values for those key attributes (power, power proximity, fiber connectivity, proximity and water pressure, site size)

2. Assign relative weights to them based on importance

3. Normalize the values so they can be compared and combined

4. Produce a Valuation Suitability Index (VSI) that can be multiplied by a market price baseline to give a value.

Scoring and Weighting the Attributes

My method is creating a straightforward normalized score for each attribute - scored on on a 0–10 scale; then we can assign weights based on importance.

Suggested example weightings (these are adjustable based on your client preferences; but ultimately most are held in a relatively similar weight):

• Power capacity: 35%

• Fiber connectivity: 25%

• Cooling water proximity: 15%

• Mains water pressure: 10%

• Site size: 10%

• Proximity to power source: 5% (admittedly, this is already somewhat captured in power capacity, but still relevant)

VSI Components and Attribute Formulas

The VSI is a formula that brings these factors together into a single, comparable score from 0 to 10. To construct it, we'll firstly gather the algebraic components of each attribute:

1. Power Capacity (P) — Weight: 35%

The single biggest constraint for a hyperscale deployment. We score this on available MW against a hyperscale target, capped at 10.

2. Fiber Connectivity (F) — Weight: 25%

Speed and redundancy drive uptime and latency. Scored on available Gbps plus a redundancy bonus.

3. Cooling Water Proximity (CW) — Weight: 15%

If you’re near a large body of water, cooling becomes easier and cheaper.

4. Mains Water Pressure (WP) — Weight: 10%

Good mechanical mains water pressure means less pumping cost for mechanical cooling, and improved flow and circulation.

5. Site Size (S) — Weight: 10%

Bigger sites will inevitably provides more options for growth and expansion.

6. Proximity to Power Source (PP) — Weight: 5%

If a site is closer to a substation, it will translate into faster deployment and less setup and fit out cost.

The VSI Formula

We bring the above together into a full VSI formula to give our score out of 10.

VSI to Value

Once we've scored a site, it's just a case of multiplying the site size by the average local market price per acre for industrial properties, and then by the weighted VSI function. This will look like:

Putting it all together: a worked example

We'll use the suggested earlier weightings for our example:

• Power capacity, Ps: 35%

• Fiber connectivity, Fs: 25%

• Cooling water proximity, CWs: 15%

• Mains water pressure, WPs: 10%

• Site size, Ss: 10%

• Proximity to power source, PPs: 5%

Firstly, we'll define what our (or our client's) requirements are for their site:

Next, we'll gather the metrics for the site we're assessing:

Now, we'll score each attribute using our above formulas, utilizing the data from the above tables:

And now, we can use those values in our main VSI formula, to give our weighted VSI calculation:

And finally, we can use our weighted VSI figure to inform our value:

Let's say our site is a 40 acre site, with a value of $500,000 per acre.

So our final figures:

• VSI: 8.3/10

• Adjusted Site Value: $36.6M (vs $20M base land value)

And that's the Data Center VSI Methodology

It's a much more refined model than comparable methods, and has benefits extending far beyond the valuation of sites for conveyancing and acquisition; this approach lets developers, investors, and municipalities speak a common language:

• Developers can prioritize high-impact sites.

• Investors and purchasers can compare apples to apples across regions.

• Cities and local governments can identify infrastructure upgrades that directly boost land value.

TH

Share

See other real estate posts

Find me on Linkedin

You may also find these of interest:

The cooling system is one of critical importance; and a significant contributor in the capital and operational expense in data center construction and management. One of the issues which the cooling system must contend with is that of thermal spikes during periods of peak computation. Suddenly, a system is required to spool up to run at maximum, or even supraoptimal, capacity. Such a draw on the electrical distribution can additionally increase risks of circuit overload, throttling and even cause the power to trip.

In order to stabilize the ambient temperature - and subsequently the demand on the HVAC or cooling systems - the easiest method is to provide a heatsink to absorb excess enery in periods of spiked compute.

A novel approach could be the introduction and integration of phase-change foundations: a concept where the building’s concrete foundations are embedded with phase-change materials (PCMs) that act effectively as a giant thermal battery, absorbing excess heat during peak computation, and releasing it during cooler periods or when energy is more scarce.

Phase-Change Materials: A solid base for foundational integration

Phase-change materials (PCMs) absorb and release large quantities of latent heat when they undergo a phase transition — typically between solid and liquid phase. During this transition, they store heat at nearly constant temperatures.

In data center applications, this, in theory, will translate into:

• Peak loads cause PCMs to melt, absorbing thermal spikes passively.

• Cooler, off-peak periods allow PCMs to solidify, releasing heat slowly and evenly.

• This thermal cycle flattens demand on mechanical cooling systems and acts as a thermal buffer integrated directly into the building.

  

Eutectic Metal Alloys could be a particularly promising phase-change solution

While many PCMs operate in the 20-40°C range, eutectic metal alloys present an exciting class of materials for higher-temperature thermal buffering. These alloys are mixtures of metals that melt and solidify at a single, sharp eutectic point, making them highly stable for repeated phase transitions.

There are a handful of relatively well-researched and documented EMAs that could be suitable:

It's quite likely that a hybridized approach may be adopted; EMAs could attract a relatively high cost component, and therefore their placement should bve specific to the direct environment.

Other suitable phase-change materials

• Paraffin Waxes [Melting Point: 20-60°C] -Stable and cheap; but low conductivity and flammable.

• Salt Hydrates [Melting Point: 28-35°C] - High heat storage; but a high risk of phase separation in exposure to higher temperatures.

• Bio-based Fatty Acids [Melting point varies on fatty-acids used] - A renewable and more sustainable solution; but limited thermal range, variable purity and more prone to degradation over time.

This is to suggest that parts of the structure which has less exposure to heat could easily substitute EMAs for more economic phase-change materials with lower melting points, such as paraffin waxes (this would likely have to be encapsulated within the concrete structure). EMAs with appropriate melting points (Bismuth-Tin, Bi-SN; or Indium-Tin, In-Sn) could be embedded on zones of the building with a higher exposure to heat generation, such as the main server storage or generator rooms.

Note: Lead-based alloys may face restrictions in certain regions due to RoHS compliance. Alternatives like Bi–Sn or In–Sn are preferred for environmental safety.

Construction integration and deployment

Eutectic alloys are not typically mixed into concrete directly, mainly due to corrosion and chemical incompatibility; but are suggested to be integrated in a variety of ways:

• Encapsulated in high-conductivity shells (e.g., stainless steel, aluminum, or graphite composites)

• Arranged as modular thermal cartridges or embedded tubes within the foundation or under raised floors

These encapsulated PCM modules are integrated into the slab (on-grade), or potentially into wall cores or cavities. Active airflow or fluid loops can assist in heat exchange with the surrounding structure.

Impacts on ambient temperature control in practice

• High Thermal Density - With latent heat capacities an order of magnitude higher than concrete alone, eutectic metal alloys store more energy in less space than alternative PCMs.

• Thermal Lag - By absorbing high-temperature spikes (~120-150°C) from equipment exhausts, they reduce the thermal gradient reaching HVAC systems.

• Temperature Stability - Creates a buffer zone for ambient air, reducing fluctuations by up to 5-8°C, easing strain on cooling units.

• Redundancy Value: In the event of HVAC failure, alloy-PCM foundations can delay critical overheating by approximately 15–30 minutes, enabling safer shutdowns and failovers.

Indicative figures, efficiency and examples

Latent heat storage:

• Paraffin wax PCM: ~200 kJ/kg

• Salt hydrates: 250-300 kJ/kg

• Eutectic metals: 40-100 kJ/kg, but with 3-5x higher density

Storage energy density:

• Bismuth-Tin (Bi–Sn) EMA ~475 MJ/m³ (vs. paraffin wax ~140 MJ/m³)

Thermal conductivity:

• Bi–Sn: ~16-25 W/m·K (vs. paraffin wax: ~0.2 W/m·K)

• In practice, this means that a Bismuth-Tin (Bi-Sn) EMA would allow a faster response to rapid load changes.

Benefits and Drawbacks

The novel foun dational integration of phase-change materials addresses:

• Flattening out peak thermal loads - Eutectic metals excel over other PCMs at heat absorption - ideally suited to buffering high-temperature exhaust outputs from dense compute clusters

• Reduction in cooling costs: The thermal 'lag' created by the absoprtion and slow release of energy reduces peak HVAC draw

• Structural synergy: Makes building mass dual-purpose - support and energy storage

• Potential for energy recycling: Released heat could be redirected or used to drive thermoelectric generators.

This is likely not a silver bullet, but forming part of a wider HVAC and mechanical cooling solution. Drawbacks are generally minimal and generally surmountable, but should be considered:

• Material compatibility - Direct embedding in concrete is not viable for metal (and many other) PCMs - typically requiring encapsulation to be successfully deployed with no adverse chemical or corrosive effects.

• Cost - Bi-Sn or In-Sn alloys are expensive; economic return on investment will depend predominantly on life cycle energy savings.

• Design complexity - Adds modeling and maintenance requirements

• Weight - High density adds structural load; must be factored into foundation design

Embedding thermal memory into the fabric of data center structures could transform how we design and operate at scale. Data centers that stabilize their own climate, intelligently and passively, could usher in a new generation of thermally aware architecture - where energy isn’t just consumed, but stored and recycled.

TH

Share

See more Engineering posts

Find me on Linkedin

You may also find these of interest:

Whilst there could be an argument to simply wait for these developments, there's as much an argument to attempt to innovate around them. I've explored the idea of a novel wafer server design to overcome the inherent inefficiencies of architecturing servers into a conventional rack, but there's another provocative approach we could attempt.

This article introduces a provocative new approach: restructuring data center server layouts into three-dimensional lattice frameworks, inspired by atomic geometries found in crystalline solids.

This concept isn’t just aesthetic; it suggests a reshape of thermal dynamics within the data center; refines network topologies and space optimization; and can be mathematically derived and optimized.

The outcome of the initial workings shows incredible promise - indicating 30% in energy savings and power requirements, with less latency and hardware; whilst maintaining the same compute capacity.

A diamond in the rough: the crystalline data center design

Rather than arranging servers in flat rows or vertical stacks, conceptualize them suspended in a 3D geometric lattice - such as a face-centered cubic (FCC) or diamond cubic structure. Each node (server) occupies a mathematically optimized “atom-like” position.

Key characteristics:

• Uniform distribution - Servers are evenly spaced in all directions, preventing hotspots.

• Atomic-level precision - Server placement based on lattice constants, akin to atomic spacing in solid-state physics. Algorithmic control of each server’s position to balance thermal load, compute density, and cabling length.

• Tunneling airflow paths - Air circulates omnidirectionally—natural convection is enhanced due to the geometric openness.

• Optimized cabling - Shorter, modular interconnections via lattice-aligned optical pathways.

• Further Refinements and Technologies could be integrated for Non-invasive Diagnostics - such as Acoustic Resonance Monitoring. This will be briefly explored later as a mechanism.

• 

For this use case, it must be considered that there shoudl be some consideration towards a more suitable material; I've already outlined that much of the material selection in current server hardware is not hugely optimized for it's use case; and much of the ancillary power consumption and cooling is simply overcoming the design flaws as a result of using low-quality materials.

To support this design physically and thermally, the first choice would be a Carbon Fiber-Infused Aluminum Alloy;

1. Strength-to-weight ratio allows vertical expansion. Important for designs that will have to support loads of other adjoining racks.

2. Thermal conductivity ensures heat dispersion across struts.

3. EM shielding properties reduce cross-talk in dense configurations.

Alternatives could include a titanium honeycomb mesh (lightweight, high thermal resistance); an Aerogel-supported steel scaffolding (Excellent insulation but likely a high cost) or 3D-printed polymer-metal hybrids (Customizable with greater degree of design geometry, but still ultimately limited by load-bearing capability).

Down to the Maths: A Proposed Optimization Formula for Network & Thermal Efficiency

Let:

Suggested Optimization Function:

Where:

• The first term encourages high airflow and low temperature.

• The second penalizes long cable runs.

• λ is a tunable constant to balance between thermal and network priorities.

Worked Example

Benefits

1. Thermal Efficiency

Natural convection paths reduce fan load; distributed heat zones eliminate hotspots.

2. Reduced Network Bottlenecks

Peer-to-peer mesh topology reduced network bottlenecks through hardware such as switches; also results in shorter average cable lengths, minimizing latency.

3. Space Utilization

Vertical stacking within spherical or hexagonal enclosures, utilizes more available eaves height; could reduce footprint of overall architecture by up to 40%.

4. Modular Scalability

Easy expansion by adding nodes in lattice-compatible vectors.

5. Advanced Monitoring

Acoustic diagnostics catch issues early with minimal disruption.

6. Architectural Freedom

Could remove the need for traditional top-of-rack switches in favor of direct optical interlinks.

Drawbacks & Challenges

1. Maintenance Complexity

Replacing a faulty node inside a dense lattice is labor-intensive. Potentially achievable with robotic assistance to extract servers. Failsafe modular access system may be required; but overall design ensures that issues are not as critical, as traffic is easily routed around any fault.

2. Cooling Design Overhaul

Traditional Computer Room Air Conditioning systems aren’t optimized for spherical or cubic layouts; and therefore would require custom airflow engineering or possibly immersion cooling integration.

3. Power Delivery

Complex internal routing needed for power cabling. There's a chance that high-density arrangements may induce parasitic losses (defined as any energy consumed or lost by components of the system that do not directly contribute to the net output or desired function) - unless carefully managed.

4. Higher Initial Cost

Custom materials and design engineering increase upfront costs; but modularity ultimately will quickly reduce manufacturing costs, owing to repeatability of design with universal applicability.

Putting it all together: The bottom line on cooling performance & power demand

Cooling Load Example:

• Traditional 1MW data center = 350kW cooling load

• Lattice model (same compute) = ~250kW

Assuming these figures, this results in a potential 28.6% reduction in cooling energy requirements over a traditional layout for the equivalent compute capacity, and over 60% less cabling per node.

Could this replace traditional architecture?

Probably not yet; the design concepts are strong enough, but they'll need to be tested and refined. There's plenty of angles for solutions, but this will be dependent on equivalent development in the cooling methodology and power distribution. Not all existing infrastrcture could be converted to this architecture readily, but new developments could incorporate it much more easily.

The lattice model would likely not negate network switches entirely, but may reduce switch hops and minimize intra-cluster switching load—especially with integrated optical mesh backbones. Ultimately the mesh configuration assists in a more dynamic and efficient routing of traffic around the system, and there's less chances of hardware such as network switches to act as bottlenecks.

Additional innovations: Acoustic resonance monitoring for non-invasive diagnostics.

In crystalline materials, phonons - the quantized vibrations of atoms - carry both heat and sound through the structure. By drawing inspiration from this principle, acoustic resonance monitoring in a lattice-structured data center could unlock continuous, non-contact diagnostics of physical and thermal states.

How It Works

Each server node in the lattice acts as both a vibration source and a resonance receptor. By embedding small, low-power ultrasonic transducers and microelectromechanical system (MEMS) microphones into the structural struts or adjacent to the server housings, a distributed acoustic field could be created across the lattice.

Servers or embedded emitters periodically emit controlled sound pulses in the 15–40 kHz range. These propagate through the air and frame, reflecting and resonating in predictable patterns. The resulting signals are captured by strategically placed receivers.

Sensor & Software Integration

• Sensors: Piezoelectric Ultrasonic Transducers, MEMS Microphones; or Laser Doppler Vibrometers for advanced surface vibration detection

• Monitoring Software: Real-time DSP Engine for Fast Fourier Transform (FFT) analysis, resonance fingerprinting and anomaly detection via AI pattern matching

• On-Premises Orchestration Software: Historical acoustic signature logging, heatmap generation (correlation with thermal, power, or vibration metrics), with an API integration with data center infrastructure management platforms.

Example Use Case: Fan Degradation Detection

1. Under normal operation, fan A emits a characteristic acoustic fingerprint (stable frequency peaks like somewhere between around 16.7 kHz and 19.4 kHz from research).

2. Over time, bearing wear subtly shifts the frequency profile.

3. MEMS sensors detect this deviation before physical failure, long before a temperature spike is observed.

4. System flags early maintenance opportunity in data center infrastructure management dashboard.

Other Applications:

• Detecting cable tension anomalies (indicative of movement or thermal expansion)

• Structural resonance shifts caused by overheating nodes

• Cross-link correlation to detect air blockage or changes in airflow pathways

Benefits of Acoustic Monitoring in Lattice Architecture

• Non-intrusive - No physical contact or internal node access required

• Continuous - Always-on background diagnostics with negligible power draw

• Predictive - Enables proactive maintenance scheduling and thermal rebalancing

• Complementary - Works alongside thermal sensors and telemetry to cross-validate anomalies

• Cost-effective - Uses inexpensive, scalable components

Utilizing such data for management efficiency

• Dynamic load shifting - If resonance shifts correlate with mechanical strain or cooling inefficiency, workloads can be shifted automatically to cooler or more stable zones.

• Cooling strategy adaptation - Resonance data highlights air flow stagnation or vibration-induced inefficiencies, triggering fan ramp-ups or duct reconfiguration.

• Infrastructure aging models - Long-term signal degradation is used to predict structural fatigue or material wear, prompting modular replacements before failures.

This lattice approach draws from physics, material science, and systems engineering to break away from the legacy flat-pack model. While challenges evidently remain, the potential for radical energy and space efficiency gains, coupled with improved fault detection and modularity, make it a compelling principle that should be considered further.

TH

Share

See more Engineering Posts

Find me on Linkedin

You may also find these of interest:

We're all familiar by this point with the challenge of heat within data centers - how to manage it, dissipate it, use it, and even extinguish it. Most solutions are active, and usually require extensive HVAC or liquid cooling systems, requiring ludicrous amounts of energy.

There's a passive approach though, and conceptually it exists in cacti, camels and pine cones. It's the concept of dynamic heat regulation - and how these organisms use their cellular structures to open and close pores to control transpiration based on their temperature exposure.

Is there a way that we could turn the skin of the data center - the walls of the building themselves - to become a passive cooling system.

This novel concept explores the application of porous, ceramic foam walls with embedded smart phase-change materials (PCMs), hybridized with aerogel insulation and clad in adaptive facades - manufactured into prefabricated wall panels.

Together, this layered system could enable self-regulating, passive cooling for data centers—minimizing reliance on active HVAC systems, cutting energy costs, and redefining what's possible in data center design.

Reflecting on the Convection: Core Design Concept

1. Variable-Porosity Ceramic Foam

Engineered ceramic foams (such as silicon carbide or alumina) provide a structural and breathable base layer. These foams can be manufactured with graded porosity to promote natural convection airflow when internal temperatures rise, and restrict it when cooling is not needed.

By designing these foams with temperature-responsive porosity, it's possible to create structures that modulate airflow based on thermal gradients.

• In cool conditions: Pores remain partially closed to conserve internal temperature.

• In hot conditions: Pores open to allow convection and facilitate natural ventilation.

This mimics how certain desert plants or animal skins regulate heat exchange passively.

Some suggested materials options:

• Alumina (Al₂O₃) - high thermal resistance, stable up to 1500°C - good for general use.

• Silicon Carbide (SiC) - superior thermal conductivity, strong structural stability - best for high performance walls.

• Zirconia (ZrO₂) - low thermal conductivity, high strength - useful in low-temperature environments

Recommendation: Silicon Carbide foam, due to its durability, heat conduction, and compatibility with phase-change material embedding. However, it will have to be combined with a hybridized construction technique and adaptive facades to respond to humidity within the outdoor environments.

2. Embedded Smart Phase-Change Materials (PCMs)

Phase-change materials embedded into the foam matrix absorb and release heat at specific thresholds, smoothing temperature spikes:

• Daytime heat: PCM melts, absorbing excess heat without raising internal temperatures.

• Nighttime cooling: PCM solidifies, releasing stored heat gently, extending the thermal stability window.

Standard PCMs buffer heat by melting and solidifying near target operating temperatures (e.g., 24–28°C). In this system, microencapsulated paraffin or salt hydrates are infused within the foam matrix.

The innovation lies in smart PCM integration:

• Tiny temperature and state-of-phase sensors are embedded within the PCM capsules.

• These provide real-time thermal profiling, enabling building automation systems to adapt in response.

• Combined with IoT and AI systems, the walls become part of an intelligent feedback loop—predictively managing cooling loads.

Some suggested materials options:

• Paraffin Wax; transitioning phase at around 25–30°C, with a latent heat of 200kJ/kg. Low cost, stable and easy integration; but not the most sustainable solution.

• Salt Hydrates (e.g., CaCl₂·6H₂O); transitioning phase at around 29°C, with a latent heat of 190-220kJ/kg. Higher heat transfer but can be corrosive. I'd have concerns absout ambient gaseous salt ions reacting with metallic server hardware, which could increase corrosion rates.

• Bio-based fatty acids; transitioning phase at around 20–28°C, with a latent heat of 180kJ/kg. This is the most sustainable solution, and is also biodegrable - but as a result, also relatively unstable.

Recommendation: Paraffin wax, for its safety, tunable temperature window, and ease of microencapsulation.

3. Aerogel Hybridization

Ceramic foam is structurally advantageous, but lacks the ultra-low thermal conductivity needed for complete insulation. To address this:

• Silica-based aerogel blankets are proposed to be laminated behind the ceramic foam layer.

• These aerogels have thermal conductivities as low as 0.015 W/m·K—lower than any other solid material.

• They serve as a thermal break, keeping exterior heat from penetrating while allowing the foam-PCM layer to handle transient heat spikes.

4. Adaptive Facade Integration

The outer layer of the wall system can further enhance performance through adaptive facade technology. Options include:

• Thermochromic coatings working via a color/pigment change with temperature to reflect more heat - enables passive solar control.

• Phase-change shutters with louvers that open/close based on expansion of PCM wax - creating a mechanical-free control of airflow/light

• Electrochromic glass qwith electrically controlled tinting glass; could easily integrate into other building automation.

• Bio-mimetic skins (e.g., ETFE cushions with embedded actuators) with a dynamic envelope that expands/contracts with conditions - would enable an architectural flexibility and climate responsiveness.

Recommendation: Use thermochromic ceramic coatings on the exterior skin to modulate solar heat gain without moving parts, supplemented by mechanical-free PCM-based louvers for passive ventilation when required.

The Good, the Bad and the Ugly

• ✅ Passive, self-regulating cooling

• ✅ Low maintenance, with few to no moving parts

• ✅ IoT-enabled thermal intelligence, capable of real-time monitoring and dynamic adjustmebt

• ✅ Fire-resistant, chemically stable materials

• ✅ Reduces reliance on high-coefficient of performance HVAC systems

• ✅ Modular for retrofits and edge deployments

• ⚠️ Higher initial capital expenditure due to advanced materials; but arguably a quick return from savings in energy bills

• ⚠️ Moisture sensitivity of PCM and aerogels (requires robust sealing). If parrafin is not used as the PCM, some others would require encapsulation prior to embedding.

• ⚠️ Structural fragility of ceramic foam (requires composite frame reinforcement, but this could be fabricated from recyclable materials from conventional constructions.). Panels could not be utilized for load bearing

• ⚠️ Sensor calibration & long-term stability must be validated

• ⛔️ Parrafin PCMs arguably not the most sustainable material choice, but still the best for this application. Further PCMs could be developed for this application.

Manufacturing and Cost Profiling

Fabrication Process

1. Foam Casting: Silicon carbide foam panels formed via direct foaming or polymer sponge replication.

2. PCM Microencapsulation & Infusion: PCMs embedded through vacuum impregnation or mixed into slurry pre-firing.

3. Aerogel Layer Integration: Laminate aerogel insulation sheets to the back of foam panel.

4. Smart PCM Sensors: Add microthermal sensors at encapsulation or during post-processing.

5. Facade Layer Assembly: Add thermochromic ceramic coating or prefabricated dynamic cladding.

6. Final Encapsulation & Framing: Seal with breathable, waterproof membranes and mount into modular frames.

Indicative Bill of Materials [assuming a 1 m² Panel]:

• Silicon Carbide Foam (10 cm) - $100

• Microencapsulated PCM (2 kg) - $40

• Smart Thermal Sensors (2–4 units) - $25

• Aerogel Blanket (1 cm) - $50

• Thermochromic Facade Coating - $20

• Lightweight Frame (Al/Composite) - $20

• Assembly + QA - $30

Total Bill of Materials assumes an approximate cost of around $285/m²

Advanced Design Techniques and Further Refinements

A slower manufacturing technique, but the benefits of an additive process - 3D-Printed Graded Porosity; to create customized porosity layers per panel section for directional airflow. Possible suitable for some sections of some walls where intake or exhaust flow is of more critical importance.

Utilising Finite Element Analysis, Digital Twins + Thermal Modeling to simulate and validate passive performance before deployment.

Turning Passive Walls into Active Participants

This concept is proposed to be a systematic-level rethink of the data center envelope. It's accepted that ceramic foam walls embedded with smart PCMs is not going to solely solve the cooling requirements for data centers, but could assist in being part of the solution for lower humidity environments.

As we approach the limits of active cooling and carbon budgets, concept materials like these offer a quiet revolution. Individually a small contributor, but collectively improving the performance and sustainability of our new essential infrastructure.

TH

Share

See more Engineering Posts

Find me on Linkedin

You may also find these of interest:

Most modern data centers provision twice the capacity they need to ensure business continuity. One set of servers sits idle or “warm,” consuming power for the rare moment it’s needed. The expense and energy comsumption is the collateral damage for uptime; but in the era of climate consciousness, perhaps we should be aiming to move towards a design which can account for both resilience and responsibility.

The Concept: Carbon Flow Failover

What if your disaster recovery site wasn’t just the nearest, but the greenest?

Carbon Flow Failover is an concept that uses real-time carbon intensity data to influence where workloads fail over during an outage or surge.

Instead of defaulting to a fixed disaster recovery site, the system dynamically chooses the location with the lowest environmental impact at that moment.

This is a great time to implement such a protocol, as developments in the infrastructure management are starting to align with the corporate and social responsibilities for data centers.

• Energy grids are no longer static. Renewable penetration varies hourly and regionally.

• Infrastructure is programmable. With cloud-native tooling, dynamic provisioning is the new norm.

• Sustainability is business critical. Carbon flow failover supports corporate environmental, social and governance goals; and can reduce scope 2 emissions (indirect electricity use).

How It Works

1. Carbon Signal Ingestion - continuously ingest real-time carbon intensity data for each region/data center:

• Sources: WattTime, ElectricityMap, or local grid operators.

• Metrics include: gCO₂eq/kWh, marginal emissions, and renewable % of grid mix.

2. Site Readiness Registry - Maintain a dynamic inventory of standby or disaster recovery-capable sites with:

• Live health status (infrastructure, connectivity, network latency)

• Available failover capacity

• Local carbon intensity score

3. Failover Decision Engine - A policy-based decision engine evaluates:

• Calculates a composite sustainability-weighted score for each failover site.

• Selects the site with lowest net impact within SLA bounds.

4. Workload Migration / Activation - Data Redundancy workloads are spun up or scaled in the selected green region [Tools: Kubernetes federation, HashiCorp Nomad, OpenShift, etc]. DNS/load balancer redirects traffic accordingly.

5. Reporting & Audit - System logs the failover action, location, carbon delta, and time; and subsequently generates reports for ESG compliance or carbon accounting.

A suggested implementation stack

There's plenty of integrations that could work; and will more likely depend on the individual systems, firmware and softweare being utilized within the infrastructure itself; but some suggestions could be:

• Carbon Data Feeds - WattTime API, ElectricityMap API, NREL open datasets

• Monitoring/Health Checks - Prometheus, Grafana, Datadog

• Decision Engine - Custom Python/Go service, Terraform policies, Open Policy Agent

• Workload Orchestration - Kubernetes multi-cluster, Nomad, ArgoCD

• Failover Routing - Cloudflare, NS1, AWS Global Accelerator, Akamai

• Audit/Compliance - ESG tools (Watershed, Persefoni), carbon APIs, ledger logging

Superior Sustainability Scoring - Benefits at a Glance

Traditional Failover

• Always-on, static location

• Ignores carbon impact

• May waste energy 24/7

• Single failover path

• Focus: uptime only

Carbon Flow Failover

• Adaptive location based on real-time data

• Optimized for green energy availability

• Cold/warm Data Redundancy can be spun up dynamically

• Multiple flexible Data Redundancy destinations

• Focus: uptime + sustainability

What’s been done, and the enhancement opportunity

While carbon-aware scheduling has gained momentum in cloud computing (see Google’s work on Carbon-Aware Compute, or Microsoft’s work on carbon-intelligent load shifting), applying this explicitly to failover and redundancy systems is still nascent and largely untapped.

This is a space ripe for innovation and policy shaping, although many of the solutions would be transferable from existing systems and architecture through integrating API protocols.

The above methodology addresses some of the fundamental gaps in the current models operating by Google and Microsoft:

• Google’s Carbon-Aware Compute shifts jobs to cleaner regions during low-priority batch processing.

Carbon Flow Failover activates on-demand in response to outages — it includes emergency resilience, not just scheduled workloads.

• Microsoft’s Carbon-Intelligent Load Shifting focuses on batch jobs in flexible time windows.

I'm proposing multi-region active-passive failover optimized for latency, sustainability and Service Level Agreements, useful for production systems.

• Google and Microsoft are both focusing on internal cloud provider operations.

This proposed model is infrastructure agnostic: applicable to any enterprise or hybrid cloud with multi-site Data Redundancy.

• Current methods use basic regional CO₂ averages.

This model proposes the use of marginal emissions, time-of-day signals, and forecasted grid carbon intensity. [API Integration to WattTime API, ElectricityMap API, NREL open datasets]

• In the Google/Microsoft systems, sustainability is only considered post-deployment.

This model suggests making it part of the failover decision tree itself — sustainability-aware resilience, not just compute.

Conceptual Growth and Development Potential of Carbon Flow Failover

• Forecast-Based Carbon Pre-Warming - Use machine learning to pre-warm standby sites before a predicted high-carbon period, shifting workloads proactively.

• Carbon Budget-Aware Disaster Recovery Policies - Allow organizations to set annual/monthly carbon budgets for failover; failover happens only if the action stays within the carbon envelope.

• Smart Disaster Recovery Escalation Protocols - In cases of multi-site failure, the system can choose partial failover to multiple greener sites rather than one dirty but close site.

• Carbon-Aware Service Level Agreement Tiers - Offer users the ability to select Data Redundancy policies based on lowest latency, lowest cost, or lowest carbon.

Carbon Flow failover is perhaps the next frontier in sustainable cloud infrastructure. It's the convergence of DevOps, sustainability, and intelligent systems - effectively turning disaster recovery from a reactive necessity into a strategic lever for environmental stewardship.

TH

Share

See more Engineering Posts

Find me on Linkedin

You may also find these of interest:

In traditional infrastructure design, redundancy means duplication. One server fails, another takes over; or one data center goes offline, another picks up the load.

It's reliable, but it's also inefficient; with most data centers typically running with between 40–60% of their capacity unused, as a large proportion of the infrastructure and architecture is just dedicated to backup scenarios.

If we think about redundancy strategies as an insurance policy - the current state is madness:

Redundancy is just a digital insurance policy that currently costs far too much; and realistically pays out far too little to justify that expense.

I have no clue on the exact figures for this, but it's safe to say that if 50% of the hardware infrastructure cost of nearly all data centers is purely for redundancy protocols, then it must be billions of dollars invested in idle hardware; megawatts of unused energy; and square kilometers of underutilized rack space.

The smart loading systems do aim to use some of this capacity during periods of intense operations; but it's still not optimal - such capacity is only monetized to a partial degree - and likely never enough to justify the cost of the capital outlay. Existing infrastructure nearly has to work for double the rate of return to cover the costs of it.

We need to come up with a method of monetizing the redundancy architecture better - it needs to be elastic, flexible, and pre-determined. Short term and affordable for a customer; but generating an additional revenue stream for the operator.

Today's concept is what I'm dubbing as an Elastic Failover Exchange. This could also be thought as 'Redundancy as a Marketplace' - and is effectively commoditizing redundancy capacity to be bought for short term periods.

Redundancy as A Liquid Resource

So what's the use case for this? Consider the example:

• Company A has spare compute capacity in Singapore.

• Company B, in the same region, wants to run disaster recovery drills or prepare for the typhoon season, where they could be a potential outage.

• Instead of over-provisioning their own infrastructure, Company B rents failover capacity from Company A — just for a week, or a day.

This is the basis for the Elastic Failover Exchange:

A system where failover capacity is bought, sold, rented, or shared — not statically owned.

This is a concept which is already relatively well tried and tested in other domains - and infact, can create a staple income stream once the market is matured and solidified.

• Electricity markets: Providers buy/sell surplus power, especially in grids with renewables.

• Spot cloud pricing: Cloud providers monetize unused capacity dynamically.

• Bandwidth peering: Network capacity is traded between ISPs to ensure balance and efficiency.

In theory, there's no reason why we can't compute redundancy.

The Mechanisms and Monetization

1. A Redundancy Exchange Platform, selling Smart Contracts or SLAs.

• A digital marketplace (for instance, easily integrated into AWS Marketplace or Meta could be positioned to operate a similar service from their experience in marketplace also);data center operators and providers list available "redundancy slots".

• Tenants purchase redundancy as a time-bound service with defined SLA. Redundancy commitments are governed by enforceable contracts.

• This would be very similar to spot instances, but for failover infrastructure with guarantees.

2. Mechanised and Orchestrated by APIs

• Failover orchestration systems (like Kubernetes, Nomad, or OpenShift) connect to the marketplace to claim and activate backup infrastructure dynamically.

3. Real-Time Monitoring + Monetization

• Usage-based billing triggers only when a failover event happens.

• Idle time can be billed at a discount — or zero, depending on the model.

The Win-Win Window of Opportunity

M<onetization windows of time assists both startup businesses and data center operators - slots are only sold when capacity exists. In terms of implementation, a team far more intelligent than me with software could probably get this implemented and rolled out fairly rapidly, for little to no cost to operators with significant software skills in the workforce.

• Cloud-native architectures: Stateless, containerized apps can be spun up anywhere.

• Observability tooling: We can measure failover success and SLA compliance in real time.

• Sustainability goals: This model reduces carbon and energy waste by maximizing active utilization.

• Economic pressure: Enterprises want resilience without doubling their cost base.

Providers and Operators (sellers) - Monetize idle infrastructure instead of wasting it.

Buyers (businesses/tenants) - Get access to data redundancy capacity without having to undergo any capital expenditure.

Cloud & Colocation Operators - Have an opportunity to add a new service tier, positioning failover as a revenue stream.

Sustainability teams - Reduce footprint by minimizing unused hardware.

Addressing the Challenges - Not Perfect, but Pretty Close

There are a handful of challenges that would need to be realistically considered, but nothing which would be hugely difficult to overcome - and there's likely already existing systems which could be replicated to address them:

1. Trust & SLA Guarantees - What if failover capacity isn’t available when needed?

Likely overcome by the use of verified contracts, auditing, and automated checks (blockchain, escro systems, etc.).

2. Latency / Data Locality - Some failover needs ultra-low latency or regional compliance.

This is where the beauty of the marketplace can come in. Buyers and sellers within jurisdictional or regulatory boundaries can be matched algorithmically. Local redundancy markets could form within metros or countries.

3. Security & Tenant Isolation

The platform must enforce strict multi-tenancy and container/network isolation, like any public cloud. This should be a basic service component which nearly all operators or providers will be doing already as part of their other services.

Some Potential Use Cases, and What It Enables

Use cases could include:

• Startup clusters buying DR capacity from hyperscale-friendly regions.

• Colocation facilities sharing excess with partner sites.

• Edge infrastructure providers offering redundancy to local governments or health systems.

• Multi-cloud failover without vendor lock-in.

and in turn, a successful implementation would result in:

• A circular economy for compute.

• More democratic access to disaster recovery — especially for small and medium enterprises.

• Reduced global energy waste.

• Better ROI on infrastructure investment.

Redundancy doesn’t have to be static, nor wasteful. It can be a shared, traded, and intelligent layer of infrastructure — where resilience is a product, not just a cost.

The cloud taught us that compute should be elastic, and so there's no reason why we can't make resilience and redundancy elastic too.

TH

Share

View more Business articles

Find me on Linkedin

You may also find these of interest:

It's a great film about overcoming business challenges and spotting market gaps.

McDonalds really won over their competition for two reasons; the first is standardization of their products - everyone who ordered knew it would always look, taste and cost the same as it did last time. Whether you order your food in Alaska or Arkansas, this is still the case - you always know exactly what you're going to get.

But their biggest innovation was the creation of true 'fast' food. The McDonald brothers were the first to apply first-order manufacturing and production line principles into a commercial kitchen to create a level of efficiency that simply no other restaurant could begin to compete with.

There's a great scene in 'The Founder', where they show how the McDonald brothers conceptualized the redesign of the kitchens for maximum efficiency. They chalked out the kitchen footprint on a tennis court, and gathered the employees to enact their roles in the kitchen.

They iteratively refined the layout until they came up with the perfect system - with minimal time in moving from process to process; eliminating double handling or backtracking; and ensuring that there were no employee crossovers or bottlenecks.

Their first attempt was rather incredible in comparison to the status quo at the time; but the concept has been further iterated and latest technology leveraged to continue their reign as pioneers of efficiency. Automation, pre-prepared and packaged ingredients, custom engineered catering equipment and top ergonomic and anthropologic commissioned research has continued to develop their process efficiency even more.

Although some of their solutions are indeed specific to the catering industry; there are certain principles which their kitchens are based off:

• Standardization: McDonald's process is highly standardized, with each task and workflow carefully mapped out to ensure the most efficient use of time and resources.

• Training and Process Protocols: Every employee is trained to perform tasks with maximum consistency.

• Lean Operations: The elimination of waste and constant drive to streamline processes, with a robust supply-chain and pre-packaged products where appropriate. No movement is unnecessary, and no time is wasted.

• Technology Integration: The use of technology to track, manage, and optimize performance (e.g., timers, sensors to assist in monitoring for quality control).

• Speed Balanced with Quality Control: The balance between delivering a fast product without sacrificing quality.

These principles are largely transferrable to many applications; but for this example, I'm proposing a novel take on how to apply them in the context of operating theater design and fit-out.

The direct correlation: the McSurgical approach

Aiming to draw parallels as far as practicable with the above principles leads to create an initial brief for redesign concepts:

• Standardized Surgical Processes: Just as McDonald’s has precise workflows for each menu item, hospitals can develop highly standardized procedures for operating theaters. It's accepted that some operations will have a degree of variability - as ultimately, every patient is different - but there are certainly many procedures that are similar enough to standardize for a large part. Surgical kits could be implemented that include only essential tools, pre-organized, sterilized and packaged for specific surgeries, reducing time spent searching and procuring equipment. Once spent, they can be repackaged into disposable containers for processing, incineration or sterilization by other staff.

• Training and Protocols for Surgical Teams: Streamlining surgeon, nurse, and support staff training using clear protocols, much like McDonald's staff training on consistency and efficiency. Modular training programs have been utilized in many industries as a way of developing training for staff, and would equally work as well in this scenario - although it's noted that this is not a core component of this brief, so won't be focused on much further.

• Lean Principles for Operating Theaters: Reducing unnecessary movement or steps by designing operating theaters with efficient layouts and tools that are always in the right place. The idea is to eliminate “waste,” such as unnecessary tool sterilization cycles, or inefficient use of resources.

• Technology to Improve Workflow: Just as McDonald's uses point-of-sale systems to track orders, operating rooms can integrate software to track surgeries in real-time—monitoring instruments, team members, and progress. Robotics and automated solutions could serve to benefit this eventually, but there will be greater leverage from maximizing lean principles to start with - technology could deliver better efficiency, but it's implementation will come at a larger cost.

Theater Adaptations - some novel design attributes

Centered and Modular - The Overhead Gantry Concept

Although ceiling-mounted fixtures are commonplace in operating theatres, there is a more universal solution capable of integrating increased modularity, ease of handling, and assistance for surgeons, staff, and patients to develop greater ergonomics and efficiency measures.

This will be similar to the solutions used in industrial manufacturing facilities for the use of overhead gantry structures which will feature rails capable of moving bearing-mounted fixtures.

Fixtures which are permanent and commonplace throughout all operations, such as:

• Lighting

• Diagnostics & Vital Signs Monitoring

• Waste management (such as suction machines)

These can be mounted on the gantry system combined with articulating and actuating arms. These accessories can be rearranged in an infinite amount of configurations to suit the individual surgeon, the procedure, and patient welfare.

Additionally, certain fixtures could also be installed on the gantry system combined with motorised units to create an automated sweep function 360 degrees around the room; this will come in particularly useful for sterilization processes, which will be detailed in sections below.

Waste Disposal and Management

One efficiency measure that appears to be regularly overlooked within hospital environments, (admittedly, sometimes owing to the existing infrastructure and build design), is the waste disposal and management seen within operating theaters.

Which still can regularly rely on conventional lined bins where waste is segregated in the theatre itself before being bagged and taken elsewhere for processing.

Staff skilled in operational techniques are not best suited to carrying out basic sanitisation and waste segregation.

Therefore, similar to many other manufacturing processes, the waste disposal and management should be dealt with by a system that simply extracts all waste from the first instance where it can be separated and managed via automation screening or other manual separation by specialist hygiene professionals.

How this translates in reality is a centralised waste chute in the centre of the operating theatre. The all-waste can be quickly transferred into an underground conveyor system or chute system that will transfer the waste to a centralised processing location where instruments, metals, and other incinerable waste are collected via magnetization or detection. Other forms of hazardous waste and other medical waste can be dealt with by alternative means.

Fundamentally, this is ensuring that all time within the operating theatre is dedicated to practises pertaining to operating itself and no other ancillary operations such as cleaning or waste disposal. Are absorbing the time of the skilled members of staff.

This approach also ensures that there is minimum time and exposure to shared atmosphere between contaminated instruments or tissues and the operating theatre environment, ensuring that all potential contaminants are continuously being evacuated from the sterile environment. This practice will also assist in reducing sterilization cycle time between operations, potentially enabling operations to be carried out nearly back-to-back if required.

Sterilization and Infection Control - in-procedure and between procedures

It is recognised that no single fumigation or sterilization method is 100% effective. Therefore, a number of different approaches using the most common technologies would assist in serving the most efficient type of sterilization.

Certain fixtures could be installed on the Gantry system combined with motorised units to create an automated sweep function 360 degrees around the room, fixtures could even be multi-purpose - such as using the rear concave side of the surgical light to contain UV-C or Blue Light Panels. The concave outer will assist in serving to provide a greater spread and exposure to the light.

Floor and eaves vents are to integrate a full HVAC system, combined with a full fumigation system. This will enable the most effective longer cycle sterilization when time permits, but also features shorter cycle solutions between operations.

In practice, the most effective sterilization system will be an automated approach, using lighting panels and the HVAC/fumigation system in tandem:

• Continuous sterilization where possible through the use of blue lighting panels overhead of surfaces, mounted as full ceiling tiles. There will also be Hydroxyl generators to run continuously throughout the room, in loop with continuous extraction and air treatment. Both processes are deemed to be safe amongst humans and equipment.

• Between surgery - a 360 degree UV-C light sweep from a gantry mounted panel or fixture, which could be carried out in short cycles between surgeries. To improve efficiency even more, modular pre-sterilized carts with all equipment required are prepared in advance, and wheeled in as and when required.

• Overnight and periods of dormancy - Hydrogen Peroxide Vapor and/or Formaldehyde Fumigation - typically longer cycles, creating a more sterile environment, and unlike the lighting based sterilizers - is not limited to line of sight exposure.

Finally, consideration could be given to a one-way system for all staff and patients to take:

• An entrance where all equipment and staff that enter through it must be of a sterile nature prior to entering the theatre, to ensure that the sterilisation status is maintained and contamination only occurs at minimal levels and when absolutely necessary.

• An exit for the patient following surgery back onto the ward.

• Additionally, an ancillary exit for all of the staff, trolleys, storage equipment, operating tables etc to exit through to go elsewhere for sterilization.

In order to maintain the efficiencies of the operating theatre, new sterile equipment is simply wheeled in as modular units through the first door.

Ergonomics and Anthropological Considerations

The most recent data on ergonomics and effects of colour on neuroscience indicate that hues of blues and greens assist in maintaining an environment of calm and peace. This is typically seen amongst many of the existing medical equipment, but the blue panel ceiling will assist in developing this further. As a result, walls can be kept white in colour, gaining illumination hues from the blue light panel ceiling.

A direct crossover from the McDonald's model would suggest a redesign of the operating theatre with clear zoning and workflow paths to ensure that equipment, instruments, and staff are positioned to minimise any unnecessary movement.

Specifically, this will encompass a circular or U-shaped layout where the surgeon's main area, the scrub nurse station, and surgical assistance tools are within arm's reach for all parties, minimising the need for staff to move around the room during procedures. Finally, 360-degree screens mounted at eye level throughout the entire room will also assist in giving clear visual representation of vital signs throughout the entirety of the surgery to all parties.

In reality, this will transpire as effectively as smart operating theatre design, very much akin to smart home where lighting, temperature, and air circulation are controlled with a single interface.

There are plenty of other optimisations that could inevitably assist in refining any of these ideas further. The implementation of AI and other technological advancements will assist in improving efficiency to an even larger degree. However, in the interests of maintaining some brevity for this discussion, those concepts can be explored later in other publications.

It does serve an interesting concept however - often seeking inspiration internally to a problem simply results in an echo chamber; whereas looking to other industries as to how they overcame their challenges can often serve as a better starting point.

TH

Share

See more medical posts

Find me on Linkedin

You may also find these of interest:

Could it be argued that cloud evangelism is starting to show its cracks? It seems that everyday there's a fresh report of a data breach. And increasingly, there are plenty of businesses that are migrating away from the cloud and back to local, on-site server hardware.

Cloud computing does have a number of huge benefits; and really points to the underlying reason for its popularity over recent years. A robust set of architecture that has been relatively cost effective with lower management obligations; global access; disaster recovery; but predominantly for ease of scaling.

But it's never been the perfect solution for all businesses though, and particularly in the case of financial institutions. They absolutely have a requirement for scalable storage, but it's always been a balance of providing that in conjunction with mitigating the main drawbacks:

• Latency matters — in trading, milliseconds can mean millions (or even billions).

• Cybersecurity is local — risks compound when you're sharing infrastructure with a thousand strangers. Colocation in data centers has mitigated some of this risk; but it's never going to be eliminated.

• Regulation is growing teeth — and it's biting down hard on sovereignty, auditability, and compliance.

We’re entering an era of data nationalism and zero-trust by design, where the idea of running sensitive workloads in some abstract, amorphous “region” doesn’t sit right with boards or regulators. Equally, ditching the cloud entirely isn’t viable either - there are too many dependencies.

But there's a solution: what if data center operators stopped trying to pull banks into the cloud - and instead brought the cloud to the bank?

Data (city) centers - prime real estate for free

Site acquisition within inner-city areas remains a continuous challenge for modern data center operators. With most primary real estate in major cities such as New York, Seattle, San Francisco, Miami, and others often exchanging for extremely significant premiums that can only be genuinely justified and afforded by businesses making substantial, continuous, and stable profits.

As a result, many directly central inner-city areas - although possessing a large demand for data center infrastructure - often rely on sites up to thousands of miles away, owing to the uneconomic rental (and operational) overheads from running an inner-city facility.

There is significant mileage to be gained from rolling out more urban data center infrastructure though:

1. Deeper enterprise foothold. Some suggest that cloud growth is starting to plateau, and the next frontier is more likely to be industry-specific infrastructure.

2. Localized infrastructure has near-zero latency. Inner city clients such as institutions carrying out high frequency trading would pay a premium to have a lower latency than their competitors.

3. Future-proofing the cloud narrative. By embedding inside regulated institutions, data center operators avoid any potential regulatory whiplash coming for public cloud platforms. It becomes part of the critical infrastructure fabric, not an external dependency.

The billion dollar idea: cloud infrastructure baked into bank vaults

This idea is a win-win if it can be pulled off: data center operators - Google, AWS, Equinix, Iron Mountain or Meta - launch specialist divisions to build embedded cloud infrastructure directly inside financial institutions. Private, secure, fully-managed micro-regions installed within the walls of a trading firm, investment bank, or clearing house.

As more office workers have shifted to more remote or hybrid working patterns, there's likely an abundance of square footage of space available to be repurposed - and still likely enough room for the plant, cooling and operational infrastructure required to run it.

I imagine this more as a a bespoke infrastructure cell, run by hyperscalers but physically and logically isolated to serve a single tenant. The data center operators would have a specialist team or division with the security clearance to operate and manage these systems exclusively.

Effectively cloud-grade tooling at zero latency for the financial institutions. Increased interconnectivity and infrastructure for data center operators. A win-win situation, which also provides a natural way out of the binary cloud/on-premises debate too.

This could be a complete new category of cloud computing

A specialized team builds modular, scalable infrastructure nodes that live inside the walls (and under the policies) of the institutions they serve. Cloud-native tooling, AI-powered observability, API-driven provisioning - but with the physical security and audit trails you only get from an on-site infrastructure model. Infrastructure that’s sovereign, secure, and smart, without being centralized.

This is arguably less about building the cloud where it should be, and more about building it where it needs to be.

TH

Share

Find me on Linkedin

See other Engineering Posts

You may also find these of interest:

For those unfamiliar, Bloomberg operates a subscription model for each trader within a financial institution, which gives access to industry news, market prices, data insights, as well as a trading platform operated by their own proprietary hardware, infamously named the Bloomberg terminal. It's an iconic industry standard, and one that has never changed.

The Bloomberg Terminal's continued use throughout modern day financial institutions is one of general compatibility - as it is utilised by all of the financial institutions, it remains an important staple part of all operations. Nothing performs as well as the hardware, and no other hardware integrates as well with the software.

It's the same playbook that Apple have created with their hardware and software pairings; and Blackmagic Design with their daVinci products also with their video editing software and controller.

A Chilling Truth: Cold Chain Logistics is Currently Crap

One of the fastest developing sectors in pharmaceuticals is the the rise in the prevalence of biogens and personalized medications. This is arguably fast approaching the peak of medical and pharmacological technology, but it's being let down by the simplest factor - logistics.

The issue with these medications, is that they often require to be kept at a very stable temperature - which can be down to as a cold as -150 degrees celsius for cryogenic chilling; but as a bare minimum, many require a stable temperature of around 2 to 8 degrees celsius.

There's a few reasons for the piss-poor performance, but it can all be distilled down to a simple fact - noone owns the process. And as such, there's no accountability for when it doesn't go to plan.

(which, conversely, seems to be most of the time).

1. Firstly, there can be a complex chain of custody. Sometimes this involves many different logistics partners. Many use completely different systems. Some good, some bad. Some are fine, but badly operated; some are bad, but operators just have to do their best.

2. This means that as soon as there's a gap in the tracking, there's a chance of mishandling - for products to exceed their safe temperature range, damage to product or packaging etc.

3. As with much of logistics, the real weak point is the last mile delivery. Everything from bicycle courier through to HGVs can be utilized. The product can travel thousands of miles in perfect condition, and then is ruined at the last stage. Infrastructure is often lacking when delivering to more remote areas - making transit times and methods very unpredictable.

4. Very few logistics partners have dedicated solutions for these purposes. Temperature regulated storage can exist, but it's a little rudimentary - the reality is that medicinal products only form one tiny part of their total packages they will handle; and their vehicles will be built to suit the widest specification of packages, and not tailor made to the transportation of medical goods. About the best solution that's currently mustered is an insulated coolbox, usually ABS shelled with expanded polystyrene cores.

There's other more macro problems with this practice also:

• Complexity from Personalized Medicine: The rise of biologics, gene therapies, and patient-specific treatments requires smaller, more frequent, and highly specialized shipments. These products often have unique storage needs (including ultra-cold or cryogenic temperatures) and short shelf lives, complicating logistics and increasing the need for precision and speed.

• Regulatory Compliance: Stricter global regulations demand robust tracking, serialization, and data integrity throughout the supply chain. Meeting these requirements is resource-intensive, particularly as regulations evolve and differ across markets.

• Visibility and Data Fragmentation: Ensuring real-time visibility of temperature, location, and chain of custody is challenging, especially when using multiple logistics partners and crossing borders. Data silos and fragmented systems make it difficult to monitor and respond to issues promptly.

• Sustainability and Rising Costs: Environmental regulations and the push for sustainability are driving up costs. There is pressure to adopt eco-friendly packaging and reduce energy use, but these solutions can be expensive and technologically demanding.

• Packaging Optimization: Advanced packaging is needed to maintain temperature stability during long or unpredictable transit times. This includes insulated containers, gel packs, and smart packaging with temperature sensors. However, optimizing these solutions for cost, weight, and sustainability is a complex balancing act.

• Expansion into Emerging Markets: As pharma companies reach into new regions, they face challenges transporting temperature-sensitive products over long distances, often through areas with poor infrastructure and unpredictable customs processes.

A cold-chain gold rush; and why medical device companies have the perfect shovels

There's an opportunity to simply specialize in cold-chain medical logistics, and the bar is currently set very low to do it well - with a little effort, and a deliberate attempt to get this implemented, the market can be totally sewn up before competitors even have a chance to begin assessing it's addressability.

So why medical device manufacturers?

• There's a Cashflow Gap - Pharmaceutical developments in arteas such as diabetes means that some divisions of medical device companies have been quietly shut down. Something needs to fill the hole left from that lost revenue.

• This requires a hardware solution, and they're already pros at it - It's a simple product development, and noone is better set up to handle the challenges, select the best materials and practices to get this designed and operational. It can be manufactured in their existing facilities, using their existing equipment - and so can be scaled as linearly as required as the business and network develops.

• As the hardware is proprietary, it's software system can be integrated into the MedDev company's existing system. It's a simple GPS tracker and in-case sensors with cellular connectivity, implemented into the logstics system with a simple API.

• The only gap is the logistics side, but this could be completed with a simple acquisition of the right company. A perfect acquisition could be a division of Brinks or Axiom - these companies regularly carry out the ultra-secure logistics for transferring cash and valuables between banks. They're used to uninterrupted chain of custody solutions; transport utilizing secure specialist vehicles that are fully tracked and monitored from end-to-end; and delivering with a precision with effectively zero margin for error. All of the existing skills and systems could nearly be entirely replicated and it would carry a perfect success.

• Medical Device companies already have the existing contact and customer base within the medical industry - everyone from hospitals and clinics through to pharmacies and individual practitioners. It should be an easy sell - they're already a trusted partner, and offering a specialized service that has been lacking for a serious amount of time.

• And finally - operating just one specialist logistics domain enables a potential monopolization of the market for this. Existing logistics partners are too overleveraged in their other operations to implement this with any form of agility. The MedDev company will own and operate the specialist transport containers, which will protect them from competition - other companies can implement the transit, but none are as well positioned to design, develop and manufacture the storage containers.

Pull this off, and you've just created the Bloomberg of cold-chain logistics. Hardware, software and distribution cornered; and then it's just a case of maintaining and expanding however far you wish to go.

Holes in the hardware

Insulated containers are being used currently, but they're a little basic - alot of them passively cooled with an ice block nestled in a expanded polystyrene container with an outer plastic shell; the most complex is usually a mini-battery powered fridge.

The good news is that there's an opportunity to develop it leaps and bounds, and much of the technology already exists - little development is needed, it's more just a case of packaging existing solutions correctly.

• I'd suggest a modular solution - three compartments, individually sealed - a battery and component base, (with all of the technology and main systems installed); cryogenic container in the middle, cooled via a closed loop of liquid nitrogen. And finally, a refrigerated container on the top, to enable a lesser temperature to be maintained in a separate sealed

• The majority of the weight will be at the bottom with the batteries and refrigeration components, so the case will naturally always store this way - and reduced chance of the case ever tipping over or becoming disturbed in transit.

• Although this is mainly cooled via active systems, the whole exterior is finned so the entire case continuously acts as a passive heatsink. A metal outer finned shell will assist in this through material properties.

• A visual indication screen will assist in monitoring for the courier - with audible and visible alerts and warning lights showing any signs for concern or attention. The case will have an integrated GPS system, and cellular or satellite connection. As the whole outer is to be metal fins, these can be wired in to act as a whole antenna for these systems, resulting in absolute minimum chance of it losing signal etc.

  

Specifics for the software, please

1. Trusted Real-Time Chain-of-Custody Platform (Digital Custody Transfer)

There's no universal, immutable, and simple digital platform for logging chain-of-custody handovers in real time, especially one usable by low-tech field workers. This addresses:

• Lack of trust or traceability during transport (handoffs between logistics providers, drivers, hubs).

• Manual paper logs still common in many markets.

• Gaps in audit trails during temperature excursions or delivery delays.

Solutions could include:

• A blockchain-based or tamper-proof digital logbook for handoffs, using mobile/NFC/Bluetooth validation.

• Each participant (driver, warehouse, pharmacist) confirms custody and condition with a tap or scan.

• Simplifies GDP (Good Distribution Practice) compliance.

2. Predictive Thermal Degradation Analytics (Beyond Temp Monitoring)

Existing trackers tell you if a product exceeded temperature thresholds, but don’t predict or quantify the impact on drug stability or efficacy. This could address:

• Uncertainty about whether to discard a product after an excursion.

• Need to minimize wastage and insurance claims while ensuring safety.

Solutions could include:

• AI/ML models that consider time, temperature curves, and medication type to estimate remaining potency or shelf life.

• Mobile app or API integrated with existing temperature data loggers.

• Could guide pharmacists on accept/reject decisions and improve incident response.

3. Micro Cold Chain for Last-Mile Transport in Rural/Underserved Areas

Many developing or rural areas lack reliable last-mile cold chain infrastructure. Vaccines or biologics can commonly spoil in the last few miles (bike courier, rural clinic, etc.), and temperature control is often out of the sender's hands during final delivery. A proprietary hardware would address most of these issues; but additional solutions could include:

• Peer-to-peer or local courier optimization software, integrated with a compliance and monitoring app.

• Backup solar-powered or phase-change-material-based containers for low-resource settings and markets.

4. Simplified Cold Chain Data Standardization and Interoperability Layer

There’s no easy way to standardize and share cold chain data across stakeholders using different hardware and platforms. Manufacturers, third party logistics providers, warehouses, hospitals often use different systems. As a result, data silos and there's a lack of a universal data schema for cold chain status. The easiest solution for this would be:

• Lightweight middleware or API layer that normalizes data from disparate IoT sources (data loggers, GPS, ERP); creates a real-time dashboard for all stakeholders, but should enable a “plug-and-play” integration with minimal effort.

5. Environmental Condition Forecasting & Routing AI

Few products combine weather, route, and vehicle conditions to preemptively adjust logistics in real time. Most logistics providers operate static routing, which can't adapt to sudden environmental risks. Integrating this functionality would place increased priority on product integrity, which is core for vital medications.

This could be implemented by utilising AI to route in real-time and provide packaging recommendations based on current and forecasted temperature/ humidity, traffic conditions and types of product being transported.

And that's it.

Simple hardware combined with a relatively straightforward software. A market which is a serious roadblock in modern pharmaceutical manufacturing and distribution, and slightly too specialist to be offered by existing logistics providers.

This is a ripe opportunity for the right person; companies with large cash reserves or access to capital should consider moving with agility, and the market could be cornered relatively quickly.

In addition, there's additional opportunity for distribution of blood, breast milk and organ and tissue transplants between medical facilities also if the tracking and traceability protocols are refined sufficiently to meet regulatory compliance.

Although this has the potential to be a gold mine for the company that corners this market, there's a more important outcome from this -less missed opportunities and treatments, and increase in patient welfare and probability of successful outcomes.

TH

Share

Find me on Linkedin

You may also find these of interest:

There's one sector of property that is showing increasing promise despite its challenges: the investment into data centers. These facilities were once solely invested in by tech companies or cloud providers, whereas now venture capital firms and REITs are seeking an increasing amount of this square footage to form a key part of their portfolios.

I'll state the obvious though. A data centre is not a factory. It is not an office, and it's quite unlike any other form of commercial real estate. Carrying more complexity, demand, and competition than some other asset classes. Specialist in many regards and to a large degree in some aspects. Not as specialist as many other infrastructure installations, such as power distribution, renewables installations, or oil and gas facilities.

So, for traditional investors looking to dip their toe into this new world, the market appears to have stabilised sufficiently to begin to give a broad overview of the types of deals and negotiations that often take place for new and existing sites. These are a different beast to conventional commercial and industrial property however, and there's certain considerations and differences that can hugely affect a property's viability and value.

Broadly speaking, there's three main threads that run through all aspects to inform the property deals - power distribution, cooling capacity and connectivity.

The money, the makers and the mechanics

There is currently an unprecedented demand for good data centre sites. These facilities were traditionally once largely funded by major tech companies (what would have once been referred to as FAANG - Facebook, Amazon, Apple, Netflix, Google - companies). They would have secured the sites, constructed the data centers, and operated them as the freeholder thereafter.

After the first few operational years in the early 2000s, there was a marked shift towards venture capital firms providing the funding to construct these properties which are then leased to the above tech companies; new tech companies (Oracle, Salesforce etc) or increasingly specialist data centre operators (Equinix, Iron Mountain etc). Many sites would require investment into increasing infrastructure capacity also. Even legacy sites could now require additional investment into increasing capacity again, as AI applications require much more power and cooling.

This is to enable the larger tech companies to utilize more of their capital in the increasingly costly fit-outs of these properties - the server fit outs, cooling solutions and networking infrastructure will be of the absolute bleeding edge of the current technology, in order to provide a little future-proofing to their investment.

Leases appear to be typically offered for 15-20 years. Rent/rates review clauses were quite variable on older leases; but owing to the current demand for these sites, it's expected for more frequent review clauses to be more commonplace.

Additional conditional review clauses can also be implemented in the event of a phased infrastructure investment plan that may not be fully operational at the time the lease commences.

Reutilization at renewal

A recent report on the churn rate of existing legacy data centres that have been under lease for the past 15 to 20 years has shown that typically, on average, around 86% of sites have been re-utilised at the end of an existing term for a similar use.

This is predominantly due to the site's relative interconnectivity between other infrastructure and other data centres - factor that is of more critical importance to the overall operations than building and hardware itself.

Again, on average, rent/rates appear to have increased by up to 22% on renewal leases. However, it should be recognised that sites with a greater power capacity (even if that capacity is not already realised or operational) are now starting to attract much more significant premiums owing to a low number of sites that would be suitable for AI hyperscale data centers rather than the traditional compute types.

The general consensus within the industry throughout the late 2000s was the expectation that rents would decrease by around 15% upon the renewal terms, as it was expected that these properties would have more limited use, and may not necessarily be reutilized at the end of their existing terms. Legacy sites were once thought to become obsolete in favour of more modern infrastructure, but they do appear to have a intrinsic benefit where a lower latency is required for certain applications - it’s more achievable for those targets to be served by local facilities rather than trying to force more through a centralized hyperscale solution.

These sites could be utilized for modern applications, but will likely involve end users having to place more reliance on inference - using data locally stored on devices to inform responses to queries, with lesser requirement to route through data centers and cloud facilities.

It should also be expected that the premiums such sites are currently attracting will inevitably likely soften as the overall network of data center infrastructure spreads out wider into a more even and uniform distribution; and especially if inference applications increase in use.

A concept you'll be used to...Location is everything, but not necessarily for the same reasons.

Traditional real estate is dependent on location; most commonly for desirable areas with strong demand for rental, good transport connections, local prosperity, centralised locations, distribution networks, or otherwise.

Whilst arguably these considerations are factors in the context of data center sites, there are really four parts of the location in particular that have significant sway on the value and the variance of demand a certain site may receive.

1. Ashborne, Virginia

Likely not the one you thought would be key, but Ashborne - located in Virginia - is currently the hot spot for the US data center development. This was one of the first critical areas where significant infrastructure investment was made in the early 2000s. Currently, it's estimated that around 70% of all internet traffic in the US passes through data centers located here.

Significant investment continues to be made in this area, and as a result, sites are trading for a significant premium far beyond and in excess of any other sites in the country.

2. Power - Grid Infrastructure, Distribution Networks and Generation

There are some variants in quoted figures in terms of power requirements for data centres, but loosely speaking:

• It is thought that a typical compute data centre can use an average 36MW of power, easily stretching up to 60MW of power.

• However, in comparison, an AI hyperscale data centre can use anywhere between 100 to 300MW of power for the equivalent size of facility. T

• o put this into normalised terms, 100 MW of power is enough to power around 100,000 homes in any one region.

Therefore, the infrastructure required to run these facilities are significant; and as a result the number of areas which currently have the ready capacity, distribution and existing grid infrastructure to serve these facilities are relatively slim. Those areas that do possess such capacity and distribution will carry a significant premium over the existing.

The power requirements can be mitigated to a partial degree by an adjustment in cooling methodology adopted within the facility. However, whether liquid or air-cooled solutions are chosen, either will have a significant power component contributing significantly to the draw of the facility. For context, most facilities typically employ a hybrid system with a combination of liquid and air-cooled solutions.

3. Cooling - Water, Geology, Altitude and Climate

A concept that has been widely written about is the challenge of implementing cooling within modern data center developments; as these facilities have significant requirements to maintain a stable temperature. The hardware and associated infrastructure required to run it are renowned for producing significant amounts of heat - all of which which needs to be dealt with, processed, cooled and re-circulated either in liquid or gaseous form as a continuous loop as these facilities operate - 24 hours a day, 7 days a week, and 365 days of the year.

The cooling system can simply be an extension of the power requirement, especially in the event of the most basic air-cooled solution. However, more commonly, capacity tends to be met by leveraging the data center site itself. The access and proximity to bodies of water - be that open forms in rivers or lakes, proximity to sea/ocean water or access to subterranean aquifers - can assist with site selection criteria. For instance, sites located in areas with colder climates will obviously have less requirement for ancillary cooling solutions over those which are located in warmer temperate climates. If you were selecting for cooling ability alone, you would consider favouring Alaska over Arizona.

A lesser consideration - but one nonetheless - is a site's altitude. For a site located in a higher altitude above sea level, the ambient air may be considerably thinner than in other parts of the country. Additionally, climate and altitude can inform the natural humidity of the air which can have a huge impact on its suitability for air-cooled solutions within a data centre facility. In subtropical and humid environments in particular, any air that is taken in usually must have a large amount of moisture extracted before it may be used for air cooling purposes to avoid condensation build-up and to reduce the air density around the hardware within the facility.

A process such as that inevitably places additional strain on... you guessed it... power requirements.

This is just one example of the co-dependence of many of these factors, which are currently creating significant challenges for suitable site acquisition for major companies.

4. Connectivity - Latency, Inference, Transfer Rates and Compatibility

The argument for connectivity remains a continuous battle. The critical goal is to reduce latency as much as practicable to win the business of specific clients who demand a premium service. For example, Financial Institutions executing stock market trades, Defence Contractors, as well as Emergency Service Communication Channels and other core operators within a modern domestic society.

As mentioned previously, latency can be overcome to a partial degree by increasing inference on a user device level basis to ease reliance on the required data centre infrastructure.

Thereafter, within the data center facility, transfer rates and compatibility are the two key factors, although it should be regarded that this is less of an issue for the landlord or investor and more of an issue for the tenant or data center operator. Systems in most modern day installations are currently fibre optic based, and switch architecture is regularly updated as part of the ongoing maintenance and refinement of data center systems by operators.

On a more macro level, insofar as the investor is concerned, connectivity as a topic pertains more to a certain facility's proximity to other facilities by the same operator which could make a certain site a more valuable addition to one operator rather than another. As operators begin to roll out and create a network with a more uniform distribution across a country or continent, this will become a lesser factor. This sector may develop in time to be more akin to practises adopted by mobile and cell operators by sharing physical facility and infrastructure space with their own equipment.

Connectivity is arguably one of the lesser considerations for the time being and is relatively easily updated and overcome in comparison to some of the other factors. However, it can remain as a roadblock on some sites which would otherwise be regarded as particularly suitable. As wireless and satellite communication is developed further, sites deemed to be unacceptable on the basis of connectivity alone may soon become a more economic proposition for investors. Therefore, keeping a finger on the pulse of the developments within that space would be a wise move going forward.

Location sorted? Onto the Lease Structure

It is difficult to realistically give a broad brush approach, when so many sites operating in so many jurisdictions will ultimately have a huge variance in contextual factors. However, some first-order principles can be used to form the basis for leases. These are not designed necessarily to be exhaustive but perhaps serve as a good starting point to give illustration as to what parallels or differences may be drawn to deals which conventional real estate investors are more akin to.

• It is typically going to be a long lease term of 15-20 years in order to enable a realistic return on investment to be realised. As mentioned above, rent or rate reviews throughout that term will likely be structured at relatively regular intervals (every 3-5 years, to market rate or conditional on operational metric) within the current market. This may adjust as more sites become operational going forward.

• Of all the above factors, the power bank and grid distribution infrastructure will likely remain one of the most important factors going forward. Even if capacity is not currently required, it will nearly certainly be required in the near future. Sufficient funds should be planned to continuously upgrade infrastructure within the locality as required, and such infrastructure should be written as terms within the lease to enable increased rates to be charged upon completion or phased rollout.

• The investment in infrastructure will represent probably the largest cost for the investor, and therefore the maintenance of this infrastructure is critical. Certain provisions should be expressly written into leases as to the roles and responsibilities of such maintenance.

• An unconventional clause, but a consideration in light of expansion and connectivity; trigger clauses drafted to enact increased rates should operators open a number of sites in close proximity to a certain facility. Long term, this could soften the value or demand for the investor's facility - as competitors will not want to exist as a sole facility in the center of another dominant operator in a certain area. As a result of this, a reduced market demand will enable a softer lease renewal price going forward. Contractual provisions may be considered to be drafted in order to recognise this.

This is an exciting sector that will soon develop into a level of maturity that the remainder of industrial property has undergone. Unlike conventional infrastructure arrangements, which typically see a more specialized market, the requirement and prevalence of data centres forming part of community infrastructure assets going forward is more likely to see a more standardised deal structure, more akin to traditional office, retail, and manufacturing space.

Until this point, investors should be prepared to consider all opportunities potentially containing a range of vernacular agreements specific to site and context. And be willing and open to take such decisions based on the merits of each individual proposal.

As with many opportunities in business, favour fortunes the brave. Those who took the brave leap in the early 2000s have ended up becoming some of the most powerful and influential companies in human existence. Equivalent opportunities exist for the wealth and facilitators for these businesses also.

TH

Share

See more business posts

Find me on Linkedin

You may also find these of interest:

With a limited life span, concrete’s structural integrity exponentially decreases as soon as water penetrates the surface and cracks begin to form. In this situation, the only method to fully regain the original strength is to remove, break, and relay the concrete. There will always be a shelf life to concrete, but bacteria is touted as being the saviour for extending such shelf life. This technology enables the use of bacteria to produce calcium carbonate to fill cracks as they appear, to halt the degradation process and effectively create ‘self-healing’ properties.

This is a critical issue - despite many technological innovations, we are still yet to come up with a material that’s understood as well as conventional concrete - and can routinely match its versatility, strength and durability for relatively low build cost.

Particularly in the instance of structural concrete; the material is required to retain its mechanical strength during the whole life of the structure; but often has a high tendency to form cracks allowing aggressive chemicals to penetrate the structure at varying intervals throughout it’s life. This will result major durability problems such as alkali aggregate reactions, sulfate attacks, and steel corrosion.

A need was highlighted for future concrete iterations will require higher durability, and this led to the development of bioconcrete to satisfy socioeconomic needs with the lowest possible environmental impact.

Bacteria: A biological band-aid and self-healing saviour?

Concrete is effectively a mixture of cement, water and aggregate. Distill the concept down down even further, the cement is a processed form of Calcium carbonate (CaCO3), which can be extracted from limestone, chalk or marble after being mined or quarried.

As a natural product, research has developed which enables bacteria to be used to produce calcium carbonate. In the context of a concrete installation, this means that the product is able to effectively self-heal - with bacteria producing additional calcium carbonate to fill cracks as they form throughout the installation’s life.

Bacteria are able to produce calcium carbonate via biomineralization in the presence of a calcium source. This filling of cracks helps prevent the penetration of water and other harmful substances which can lead to shrinkages within the concrete matrix which eventually lead to the formation of defects. Bioconcrete has been shown to have greater durability and compressive strength than standard concrete which results in a longer operating lifetime also.

Bacteria are able to produce calcium carbonate by means of intracellular and extracellular precipitation as a result of a number of metabolic pathways. Urea hydrolysis is the most favourable pathway due to its high yield of calcium carbonate precipitation in comparison to the other metabolic pathways. This method of self-healing concrete reduces CO2 emissions as maintenance and new concrete production is significantly minimised; processes that are also very cost effective.

Therefore, bioconcrete doesn’t just offer a more environmentally friendly alternative, it also provides an economic relief.

The sensitive setback to self-healing bacteria; and alternative solid solutions

Bacteria are very sensitive to their environment and so require optimal conditions to provide effective concrete self-healing results, this can be due to factors such as pH, temperature and water availability. These conditions are simple to optimize in laboratory settings, but difficult to environmentally overcome during real world applications - where this concrete could be used in off-shore oil rigs, or in other harsh climates and exposures.

There are other methods of crack filling in concrete; most fall under the category of autogenous healing. This method relies on the constituents of the cementious matrix to provide self-healing properties, for example, the swelling of hydrated cement paste along the crack walls due to water adsorption, or the blocking of cracks due to debris in the water and fine concrete particles due to the cracking process.

Another approach is to use blast furnace slag rather than cement clinkers, an intermediate product in concrete production. These slag particles can remain unreacted for several years after casting and so their eventual reaction has the potential for self-healing of concrete structures.

However, these techniques of autogenous healing are only efficient for small crack sizes and are difficult to predict and so are unreliable. So although these could be future alternatives, no approach appears to be the definitively best route - and all should be explored as part of further developments.

This is an exciting development in materials science; not only replicating all of the intrinsic benefits of a widely used and common material, but increasing it’s effectiveness, sustainability and durability. Once refined, this technology could form the basis to carry out retroactive repairs to existing concrete installations as part of preventative maintenance also.

TH

Share

See other engineering posts

Find me on Linkedin

You may find these of interest:

A 1911 ruling in the Supreme Court eventually split Standard Oil into 34 separate companies (some of which became today’s equivalents ExxonMobil, Chevron, etc.).

The modern approach is more one that Amazon has managed to create, but is alot more discreet in it’s practice. The character of Rockefeller would have never have worked under the Amazon structure; he would have never entertained a buyout acquisition or merger - there was only one transaction worthwhile to him; and that was to acquire your business wholly and exclusively - by any means necessary. Amazon has managed to achieve a similar level of success and expansion, but has done so in a more competitive market; and although many of the investments have paid off, rarely have they escaped from having to pay the market rate to acquire businesses.

Synergy in Energy

An even more modern approach is one of more of a temporary synergistic partnership between companies - for a short-term gain in order to achieve a certain project goal or a one-off investment, whilst still retaining the flexibility to tread your own path or to break away from the agreement in pursuit of something greater that aligns more with what you want to achieve.

Rockefeller is an example of how a monopoly can be temporarily leveraged to extract as much value from it as possible before it is scissioned, whilst still retaining much of the return generated with little diminishment of individual brand or company value. Even when Standard Oil was order to split, there still remained a conceptual inertia to the market share that had already been captured; that success was nearly largely guaranteed.

So, is there another modern day partnership that could leverage the same amount of benefit to both the wider economy and the general consumer whilst developing a remarkable presence within the market, potentially eclipsing that which either have achieved already?

The modern day Rockefeller is undoubtedly ExxonMobil in the USA, with the largest land holdings and extraction licences spanning millions of square miles across the nation, operationally dominant and yielding efficiency and scale.

The other half of our modern day partnership may not be one that instantaneously comes to mind, but ultimately assists in providing a similar solution - albeit a different point in the process. These are my thoughts on how a partner, such as Kawasaki Heavy Industries could become the secret weapon in developing rapid infrastructure across conventional power sectors within the USA. Accelerating development, investment, local prosperity and unfathomable profits for both parties.

ExxonMobil's part would not necessarily be the supply of a raw commodity; but rather the waste product it produces. Look across the night sky of an oil field and you'll see it lit up with flames. These are gas flares as a result of the extraction process to avoid unnecessary pressure building up. The methane gas needs to be released and instead of just a simple atmospheric discharge for safety concerns, the methane is burned off and ignited, creating balls of flames that often run for hours at a time. As long as Exxon Mobil continues to extract their predominant product, their waste product could be utilised. There are some instances where, in some oil fields, depending on proximity to other populations, the methane gas can be collected and often recycled in other industrial processes, sometimes to run some processes within petrol refineries or other local uses. However, there are still many instances where locations do not necessarily create a viable economic proposition to collect and extract as required.

Where does Kawasaki come into all of this?

When it comes to the gas turbine market within the US, (the critical equipment utilised to convert the raw gas into useful electricity), there are only a handful of major players that typically seem to dominate the majority of the US market. GE Vernova, Siemens Energy, and Mitsubishi Electric operate a large proportion of the gas turbine market in the USA. Demand is typically for large-scale turbines to be housed in major industrial power generation facilities serving entire districts, regions, cities, or other densely populated areas.

Meanwhile in Japan, Kawasaki Heavy Industries is thought to have an excess of over 70% of the market share for gas turbines. However, their specialism is not in the production of large turbines, but more in small to medium scale. In contrast, their US presence is relatively minimal, other than a partnership with Cummins Engines currently. But market share is distinctly less through their US subsidiary, typically as a result of the US market not demanding smaller turbine sizes overall and their current restraints not enabling them to develop and produce a large-scale turbine.

The Opportunity:

But here's where the synergistic partnership lies: If we look at what we lack, what we require, what we've got, and the resources to deliver that:

1. We have a requirement for upgrading infrastructure throughout the entirety of the USA. Continuous interconnectivity required to be developed throughout all regions, rural and urban.

2. We have a requirement for electricity on all fronts: Capacity, distribution, and scalability

3. We have a waste product that is produced readily, continuously, but typically in small batches, as and when operational.

4. We have a company producing high-quality, trusted, and proven equipment successful in other markets directly suited and applicable to be adapted and implemented to utilise this waste product.

5. We have a business with a low market share in the US, with a potential to develop a strong foothold within the US market; in addition to contributing towards the US economy by expanding infrastructure, connectivity, and environmental sustainability of their operations.

The Long Play

The current bottlenecks in the current gas recycling efforts is typically the lack of pipeline networks between oilfields; and currently the inevitable challenge of processing and transferring a colourless, odourless product is just that of storage in sealed containers. Although this does save the waste product, it still exists in a form hich has a limited use case in its raw format, and it remains useless until it incurs further distribution costs in getting the stored raw product to a processing facility.

There exists an opportunity to develop a variant of portable, potentially modular solution which can be transported to site - a concept that Kawasaki have pioneered already with other models in their turbine range.

With basic distribution equipment - lorries, semi-trucks or others - temporary installations can made whilst fields are operational, and electrical grid structure slowly built out to increase distribution of the generated power. Once those oil fields are decommissioned or deemed to have dried past an economic extraction - there now exists new grid infrastructure to enable these areas to be developed at a later date.

With sufficient power distribution already existing in place, the turbines are relocated to an area where they would generate more efficiently. The revenues generated from the sale or use of the generated electricity from the waste product would eventually assist, if so desired, in funding a permanent installation to centrally distribute power between expanses of oil fields through other conventional or renewable technologies.

The build out of grid structure throughout sparse regions will enable development to occur within these areas that were once deemed to be unsuitable for development.

Sublet arrangements would enable additional returns to be generated from land which had no further economic value to ExxonMobil. Kawasaki would control the distribution of electricity throughout the area, giving monopoly on effectively the prosperity of the area, which will remain dependent on the continuous supply of electricity in order for modern economies to survive and thrive.

Even if not populated at this stage, there will be scope to enable the installation of other necessary infrastructure deemed to be unsafe, unsightly or unsavoury to be in proximity to populations - such as data centres, manufacturing processes and facilities for chemicals or otherwise.

Though the tale of Rockefeller is well told and fully understood by this point, our regulatory authorities would not typically enable such things to occur to the same degree in today's modern world.

However, there is an inevitability that comes when operators who are providing commodities of such significant value and essentialism exist within the economic domain. With limited government resources to invest in this equipment themselves, these operators must be given sufficient rein to use their wealth to their overall population's advantage.

Rockefeller may not be necessarily reincarnated as ExxonMobil and Kawasaki but there are certainly benefits that could exist from synergistic partnerships between industries. I believe now is the time to take a borderless approach between industries, expertise, value, and resources in order to develop the physical world at the rate that humanity demands it.

Of all the factors to attempt to retain as we navigate into a new revolution of the way that we live and interact with modern technologies, human capital needs to be preserved urgently and with more consideration than at present.

We require titans of industry to do this, and must accept that although some will undoubtedly profit to a significant degree from such developments, they also facilitate a different type of profit in the lives of the wider economy in terms of resources available to them, purpose, ability, skills, quality of life, and otherwise.

Rockefeller was widely regarded as a titans of industry for creating monopolies, and they were also regarded as incredible philanthropists, supporting gifting much of their wealth back into the system that helped them create it. We should not be shy from enabling companies to prosper, even to a large degree. We should be enabling it, but perhaps just a little cautious as to who we allow to benefit from it.

TH

Share

See more business posts

Find me on Linkedin

You may find these of interest:

In Greek mythology, he was a sea god, known for his ability to shape-shift. Anyone who delves deeper into the challenges data center operators face at present, there is always one roadblock which seems to be insurmountable:

Time.

• We can come up with clever engineering solutions for the power bottleneck; SMRs, gas turbine efficiencies, renewables.

• We can come up with novel ideas to improve the cooling; hybrid solutions, glycol-based liquid cooling, supercritical fluids.

• And we can also solve the regulation and site strategy problems; through careful consideration, liaison with stakeholders and many other workarounds.

The problem is never the capacity or the capability - the problem always remains the speed at which that can be realised.

Ultimately, the timeline for delivery cannot be realistically compressed any more than it is at present. The rollout of even the most creative of solutions is often dependent on many other industries, infrastructure, and factors requiring significant planning and expense before those parts can be implemented.

I have full confidence that through collaboration and innovation, all of the major hurdles the industry currently faces regarding power, cooling, and site selection will continue to be overcome permanently. However, unfortunately for operators, the reality is that demand exists presently for these facilities and will remain so for the foreseeable future.

So today's question: If the timeline for new facility rollout is the main problem, is there a method to which we can convert existing Cloud or Compute data centers to be suitable for running AI tasks?

Like Protheus, my solution involves a bit of shape-shifting; utilizing the core existing infrastructure that’s already there, but implementing a protocol into it, which I’ll loosely refer to as the AI Migration Acceleration Kit (AIMAK).

AIMAK - The elevator pitch

If there's a challenge that is universally accepted with converting existing cloud or compute hardware to be suitable for running AI tasks - it's the practise of converting CPU-centric racks to to run GPU-centric tasks.

This is not designed to replace the hyperscale AI facilities that are already being planned; but to assist in providing more bandwidth for simpler AI tasks requiring less power; it is accepted that than intense image or video rendering will still remain unlikely for this application - but such traffic can be routed to the already operational facilities designed for AI.

My AIMAK solution features three core components to be implemented properly:

1. Hardware: Modular AI Accelerator Nodes

• Plug-and-Play GPU Expansion Blades or PCIe Add-on Cards with:

  • NVIDIA L4, A2, or AMD MI300-class GPUs (low power, high AI performance)

  • NVMe storage for model caching.

  • High-bandwidth PCIe Gen4/5 or Infiniband/Ethernet (100G+) connectivity.

• These blades connect into standard servers or 1U/2U chassis.

• Optional: FPGA-based acceleration (for low-latency AI inference).

2. Software: AI Virtualization & Orchestration Layer

• Lightweight hypervisor or container orchestration (Kubernetes, K3s, or Nomad).

• AI workload scheduler that detects available acceleration and dispatches accordingly.

• GPU virtualization (NVIDIA vGPU, AMD MxGPU) to allow multi-tenancy on each GPU blade.

• Monitoring, usage metering, and fallback to CPU when necessary.

3. Network Fabric Upgrades (If required)

• Implement RoCE or Infiniband-compatible fabrics as needed for high-speed interconnects (depends on AI workload demands).

Why this approach?

Simply put: it’s about the best performance we can squeeze for the lowest cost. But a few extra considerations:

• It’s fine to cost a bit, but shouldn’t be alot - this is designed to be a temporary upgrade to help meet the demand until more suitable permanent facilities are constructed.

• It needs to be quick - if it takes forever to get it implemented, then frankly we all may as well wait until the permanent facilities are constructed. So it needs to be able to be rolled out as fast as people can fit GPUs in a server rack and reboot them.

• It needs to work within the constraints of the existing infrastructure - which means all of the existing power capacity and cooling performance needs to be able to serve this use. If everything needs to be upgraded, then again - we may as well wait for the permanent facility.

• AI traffic is becoming more simplistic - as people shift away from Google and onto ChatGPT for search enquiries - this architecture will still be adequate for LLM inference (e.g., OpenAI, Meta LLaMA-based models), computer vision at scale (retail, surveillance, manufacturing), speech recognition and translation, and even lightweight training for edge AI models.

• Supports multi-cloud or hybrid environments

• Scalable and reversible

I’m sold. Hit me with the hardware design

Any of the above GPU designs could be easily adjusted to suit this use case better at the production stage. A few ideas:

Half-Height Modular Heat Sink Kit

Redesign cooling with a swappable passive heat sink kit that fits both:

• Standard 2U servers with existing airflow (high-density fins).

• Edge/1U deployments (compact design).

This increases flexibility without needing SKU splits; reduces field engineering time.

NVMe Storage on Board

Many legacy servers have limited PCIe lanes or slow storage, so adding a high speed M.2 NVMe slot is the fastest storage we could deploy for the least amount of GPU PCB real-estate

This will also create a local cache for AI models, embeddings, or datasets; useful for inference workloads where disk I/O is the bottleneck.

Front IO Bracket for QuickSwap

Backplane access in some traditional racks can be hard; so a redesign of the front IO will help with access - support field maintenance without full server teardown.

It doesn’t have to be anything too extensive - redesign PCIe bracket to allow front-facing access to status LEDs, reset, or firmware update port (e.g., USB-C):

What the above should achieve is:

• More efficient resource usage in multi-app environments.

• Lower risk of power/cooling overrun in legacy data centers.

• Faster time-to-inference with preloaded models or local caching.

• Field-friendly maintenance for temporary deployments.

And that shouldn’t add more than probably about 5% to the typical bill of materials per unit.

Firm up that firmware outline for me…

Not my area of expertise (at all), and I may have this completely wrong. But loosely, my thoughts are:

Multi-Instance GPU (MIG)-Lite

Introduce a simplified MIG-Lite feature via firmware update, for better utilization in retrofitted cloud servers where no single app maxes out the GPU:

• Allow hardware-level partitioning into smaller inference instances (e.g., 2 or 4 logical GPUs).

• Ideal for small batch AI inference in multi-tenant environments.

• Reduces the risk of idle GPU capacity when workloads are light.

Dynamic Power Profiles

Add firmware-level adaptive power and thermal management profiles optimized for constrained power/cooling in older racks. Many legacy racks were not designed for AI-class thermal loads—this lets the GPU fit better within existing envelopes:

• “Legacy Rack Mode” (e.g., power cap at 60–70W instead of 72W peak).

• Aggressive clock gating when workloads are sparse.

Boot-time Resource Negotiation

Develop firmware that can negotiate GPU availability at boot based on system-level BIOS calls, to ensure safe and automated integration without manual intervention:

• Automatically disable/scale down GPU if power or cooling is marginal.

• Report availability to the host OS via SMBIOS or ACPI tables.

Pre-installed Model Support

Add secure firmware-level pre-loading of common open-source AI models (ONNX, TensorRT) into onboard memory to speed up deployment for temporary AI use cases without sophisticated DevOps.:

• Reduce cold-start times for inference workloads.

• Optional encrypted model store in firmware.

And there’s my take on an AIMAK.

Utilizing everything you’ve got, adding a few cheap items, and you’ve acquired a new AI data center to offer your customers who are chomping at the bit for increased capacity.

TH

Share

Find me on Linkedin

See more engineering posts

You may also find these of interest:

It was quite a simple premise. Their research suggested that many continued with local data storage as a result of the challenges and difficulty in migrating data over to the cloud, requiring long uploads through slow internet connections, large amounts of data, or lack of staff, knowledge, time, or resources.

Their solution was a series of devices that were shipped to the customer. Data could be transferred onto the device physically through high-speed networking or high-speed transfer cables before being shipped back to Amazon. Data was then processed by AWS employees, and could be transferred into data center servers ready to be accessed by the customer in the cloud.

The project continues after ten years of operation, though it has been noted that demand has continued to slow. The customer base dwindles, as a result of many already migrating into cloud computing; the Snow data migration method is a one-time sale opportunity - and in the last 10 years, internet speeds have increased sufficiently that some have simply managed to upload their data themselves also.

Whilst the small, portable devices such as the Snowball remain part of the offering, their largest device - the snowmobile - was retired around April 2024. It has been rumoured that it was never actually used, apart from moving some of Amazon’s own data between it’s servers in 2017. Even if that’s the case, it provided quite the marketing spectacle when an 18-wheeler was driven on stage during the announcement in 2016 - and probably did more to generate sales from free advertising from the world’s press that it’s redundancy never particularly mattered.

Whilst the snowmobile has been retired, there could be a bigger opportunity for it to have a second life. Not as a data migration tool; but maybe as the data center itself.

The holy grail for nearly all manufacturing is modularity - enabling infinite expansion quickly and easily - uniform, linear scaling to adjust to growing demand. It’s precisely the reason that shipping containers find themselves being used for a myriad of uses other than simply getting stuff from A to B - standardized size, weight and specification. Stackable, durable and compatible in a variety of applications.

I’m suggesting the snowmobile could form as the basis for the first modular approach for a datacenter. Instead of migration hardware - it’s fitted out with the server racks, cooling systems and power connectivity it needs to run already; and additional units can be stacked and linked in series or parallel to scale as required. You could end up with a data center of 1 shipping container, of 50 or more depending on your requirements.

There’s far more focus on permanent conventional data center construction methodologies than there is for a modular approach; but the reality is that progress in expansion is rather hamstrung for the time being.

The hype is bigger than the ability to hyperscale

If there is a data center buzzword in 2025, it is hyperscale. Hyperscale is a lovely idea in theory, but for a few reasons does not necessarily work in practice for every application, in every site, in every location, at all times.

The hyperscale dream is often just that; in an ideal world hyperscale would be rolled out as the new standard for all data center construction. However, the infrastructure required to operate a hyperscale data center significantly lags behind the technological ability to implement it:

• Archaic power grid infrastructure often dating back to close to the 1950s

• Remote areas that have poor, or little to no connectivity - power, data or both.

• And finally, areas with significant water shortage excessively warm climates/drought conditions can often pose far greater problems.

The rollout of hyperscale data centers is impacted far more on the above, rather than the operator's ability to build a bigger building with more servers in it.

Hyperscale installations are a solution, but not a universal one.

AI’s hungry for hyperscale only

The sheer compute and infrastructure requirements to run an array of equipment for the purposes of even a basic AI system far exceeds the capacity of traditional compute data centers. This is to suggest - if a hyperscale installation is being pursued, it’s nearly certainly for AI purposes - as it can’t be deployed anywhere else.

However, this doesn’t automatically mean that the demand for traditional and conventional cloud and compute data centers is any lesser, or even static. The requirements for storage appears to continue to increase - and will effectively continue to do so as proportional to the reliance and development we invest into the digital economy.

Addressing the infrastructure bottleneck, and why it’s the anchor to real progress

Demand for all data centers are going to continue to increase exponentially - as a result of the rise and advent of AI, as well as a development of all the additional new digital content being produced. The demand exists for construction of many new data centers, but it will have to be accepted that not all of those can be hyperscale. Some of those can't even be medium scale; and in some cases, that may even have to be the smallest of scales - as a result of the bottlenecks created by the archaic and underinvested infrastructure.

The speed of which infrastructure can be invested in and improved is relatively slow and often dependent on public money raised by tax systems in order to fund it. This places a limit on the amount that can be upgraded at any one time, in addition to the bottleneck of skill sets required to implement such huge projects globally.

Therefore, the reality is that not all data centers can be constructed at the hyperscale level instantaneously; but we should continue as much uniform rollout of conventional facilities as much as practicable. The modular approach is probably the wisest one, whereby small-scale data centers can begin to be rolled out with expansion and capacity slowly increasing as infrastructure is upgraded and invested in over time.

The modular elevator pitch

An entire post could be written on the design cues with regards to the specification for a modular data center unit equivalent in size to that that could be installed within a shipping container. Here are the key arguments:

• A tried and tested, proven, and durable outer shell capable of withstanding most environments they are subjected to in a large proportion of the world with relatively temperate climates.

• Sufficient room for the storage of a decent amount of compute power

• Sufficient additional space for a hybrid cooling setup .Predominant reliance on glycol-based direct-to-chip cooling. And stealing Microsoft's homework from Project Natick through filling the sealed chambers with nitrogen at higher pressure in order to create a stable and thermally efficient environment for equipment to operate in.

• A simplified manufacturing procedure capable of being constructed on a production line much akin to traditional manufacturing systems. Utilizing nearly all standardized parts available off the shelf readily.

• Easily and rapidly deployed. Often classed as temporary structures, potentially, in which case would not attract the same amount of regulation, planning, or permit requirements that a more permanent installation would attract - even as a temporary stop-gap until all such infrastructure is invested in.

Of course, infinitely scalable - with units being able to be adaptable to the site and deployed in series to meet demand.

Why AWS over the other operators?

The model is capable of being adopted by any of the major operators. But I’ll state the obvious - AWS is part of Amazon. I’m aware there’s Google, Microsoft, Apple, Meta and many other wealthy players with the funds to explore this idea. But only Amazon has the customer base and the expertise to deploy this with relatively little friction:

• An already very established and rapidly expanding logistics business, capable of easily integrating shipping container distribution.

• Significant experience in manufacturing and physical product development in-house.

• The full existing skillset to integrate the hardware into the AWS platform from many remote sites.

• Ability to deploy next generation connectivity, through satellite communications - building on SpaceX’s Starlink concept and implemented through Blue Origin.

• Millions of square feet of globally distributed industrial real estate, capable of housing installations permanently or temporarily, with much of the bare minimum of existing infrastructure present to run the installations already - providing an effective turnkey solution to deployment.

The combination of the above means that if Amazon proceed with this approach with a little effort, they would be largely untouchable - as no other competitor has this level of resource available to them; and this is a relatively easy vertical integration for them.

Owing to this factor, the remaining larger competitors will be doubling down on AI infrastructure, as their skillsets and resources are far better aligned with the development and deployment of software and other digital tools. This leaves a ripe opportunity for Amazon to corner the compute data center market - and after making significant progress, there’ll likely be a ripened opportunity to acquire some of the smaller operators in this space also to secure more real estate infrastructure.

If this premise holds any truth, it could spell equal parts of promise and problem. A improved infrastructure which can expand organically over time will only serve all of us better - but there stands potential for a monopoly to be created in the cloud computing space if it’s not monitored sensitively.

Modular data centers - as a concept - is a robust solution. However, as with all modern developments in technology, but must ensure that it’s entrusted and implemented within a free market in order to protect the wider population.

TH

Share

Find me on Linkedin

See more business posts

You may also find these of interest:

The benefits of vertical integration are two-fold:

1. Reducing the business's operating costs by purchasing enterprises which represent a large proportion of their overheads. As a method of saving costs or potentially generating additional revenue by selling such services to other operators.

2. To generate an additional income stream exclusively by adding another line of services or products, to be eventually sold to the same end consumer.

The vertical integration in the context of data center portfolios could encompass a number of enterprises, predominantly to reduce the operating operational overheads and generate supply of key infrastructure components.

As data center portfolios expand, there's an increasing argument for merge or acquisition of manufacturing businesses; either for key components, or maintenance consumables. This enables a reinforced protection again any potential future price hikes or impacts from increased demand causing supply shortages.

The two key obvious sectors would be cooling and refrigeration for consumables; and also power generation and connectivity for construction and fit-out.

I’ll cover my take on the 5 which makes the most practical sense as things stand at present.

• Glycol Production and Reprocessing - Liquid and hybrid cooling continues to pervade as the main methods of cooling within data centers of varying sizes, owing to the liquid’s additional heat capacity (compared to air) to produce a more efficient cooling solution. At present, the most favored and stable solution for coolants are glycol-based liquids. Such coolants must be selected carefully to ensure specific properties to avoid damage to very costly equipment and in addition to ensuring optimal efficiency from the product.

  Glycol has naturally entropic properties; which effectively means it begins degrading soon after it is produced. It’s degradation is relatively slow, but fluids will have to be exchanged numerous times throughout a data center's life. This means a significant opportunity lies in the production, distribution, and thereafter the recycling of glycol based equivalent solutions.

  The Dow Inc Heat Transfer Fuels Division would be a logical acquisition for a key data center operator; not only to ensure the protection of supply and quality of the product themselves but the additional benefits. They will also able to monetise this by selling to other smaller data center operators.

• Fan, Compressor, Pump & Heat Exchanger Manufacturing - Not necessarily a consumable but a critical part of all data center construction. These are the key components involved within the liquid and air cooling system. These components form nearly all parts of refrigeration heat transfer and distribution systems for nearly all cooling solutions incorporated within modern data centers. The acquisition of such companies will not only enable supply to be created but designs refined to be more optimized towards the applications required.

  I would suggest that the true opportunity exists in leveraging significant intellectual property generated through other unrelated industries to apply it to a solution more applicable to data center design. For instance, computer fan manufacturer of Noctua would be a great addition to further the advancement of cooling technology and application - few other companies have invested so much into air cooling computer components, and many designs could easily be scaled to suit larger airflows. Koch’s expertise in the manufacture of heat exchangers could refine the solution to a more specific degree also.

  Again, successful deployment could easily result in additional monetization by distributing these products to other operators or industries.

• Supercritical CO2 - What is thought to be the replacement for glycol-based coolants - the utilization of supercritical CO2 as a more efficient cooling strategy. Research into this field is still in its relative infancy, but solutions have indeed been applied with successful results, and continuous refinement to processes shows a relatively stable promise for this to be incorporated as a wider adoption.

  Some factors - such as compressing the supercritical CO2 prior to heating - results in a far more efficient solution; and this can be incorporated as part of a closed loop system that effectively feeds itself. The costs of wide implementation of this system makes it largely prohibitive compared to glycol-based solutions at the moment, but this could be the perfect time to invest before this technology becomes widespread - enabling monetization to occur and scale throughout the adoption phases and entire product lifetime. Key players to invest in or merge with could be Chart Industries or Infinity Turbine.

• Turbine Fabrication - It’s regularly stated that the power constrictions are the main issues with regards to data center expansion. Strangely, the issue is not necessarily generating the power, as many solutions do exist which could help to support any requirements in that regard. However, there appears to be a significant bottleneck within the fabrication and manufacture of turbines (specifically gas turbines) for power generation.

  Although this may not end up technically being a merger or acquisition of the entire company, there could be a fair argument for data center operators to invest directly into GE Vernova, Siemens or Mitsubishi, to facilitate additional production lines for turbine manufacture or fabrication. This is to enable a continuous and steady supply these critical components, to avoid any bottlenecks in construction timescales or constrictions on expansion.

  Wait times for turbine production from GE Venova are thought to be around four years with the lead time increasing continuously; so this is a problem that will not be overcome without some form of wider capital intervention.

• Fiber Optics - Finally, fiber optic cabling is utilised throughout nearly all modern data center designs, and currently the technology has more than sufficient quality and speed to satisfy current and future technological requirements. No only will an abundant supply be required for all future constructions, there will likely be a requirement for upgrade of legacy facilities cabling - modernizing to fiber optic from copper found in traditional RJ45 connections.

  This is an essential and widely required component, which would generate both significant overhead reduction, as well as the potential for additional revenue streams.

As the technology develops, there could be more other promising sectors - such as geological IP and geothermal installers and operators; silica and phenol producers/processors (key components of chipsets); and incentive to maintain a highly vested interest in semi-conductor manufacturers - predominantly TSMC, Intel, AMD, Nvidia and ASML.

The premise of a vertical integration is a sensible route which many industries have successfully implemented, and there could be numerous routes to incorporate additional enterprises into data center operators existing businesses.

TH

Share

Find me on Linkedin

See more engineering posts

You may also find these of interest:

There could be a way that data centres could help provide an abundant supply of food - locally produced - across an entire nation for little to no cost. All whilst generating additional jobs to support the local economy, reducing food miles and making products previously unfeasible in localised climates to be grown pretty much anywhere in the world.

If there's one part of data centre construction that remains a continuous challenge it's the production of vast volumes of heat the equipment emits. This was always a notable challenge in a traditional compute data center; but with the rise of hyperscale and AI data centers, this challenge is just becoming greater all the time. In conventional setups, the heat is usually created in one of two forms:

1. Ambient heat (ie exhaustive hot air gas)

2. Thermally absorbed through water or liquid variants

Increasingly commonly, a hybrid approach to cooling is adopted; usually utilising a combination of the two. In particular, my aforementioned opportunity exists for recycling the hot air exhaust air that's dissipated.

Up to now, a significant effort is placed into the management of this hot air, and in particular, aiming to cool it or remove the heat from it. This is often through:

• Passive fans blending with cooler ambient air

• Refrigeration

• Condensation

• Heat exchange

Irrespective of which ever method is adopted, there is remains a significant capital cost (or operational cost) in implementing such solutions. And most solutions are active - ie requiring power - and will be an equally significant contributor to the colossal energy requirements a data center requires to operate.

I believe this provides an opportunity whereby instead of processing the hot air in an attempt to cool and recycle it; the data center could simply pump it out and it could be used effectively elsewhere. No need to cool it or recycle it; and fresh cool air could be taken in. A huge operational cost saved.

The environment where this hot air could be useful currently exists in common forms globally - dotted generally throughout a nation for certain products; ambiently as a climate for others. Up to now, this solution has not been widely adopted owing to the significant operational costs required to heat or manage the installations.

Here's the hot take: How about building a greenhouse, or food plant, or agricultural unit next to the data centre?

I'll flesh this idea out a little more:

Around the world, hot and humid environments create perfect climates to grow certain products: Bananas, Tobacco, Soft Fruits, leafy greens and many others. For the environments which don't have this natural climate for well over 200 years, humans have created artificial environments suitable for growing these products (traditionally greenhouses, but other indoor agricultural units exist such as hydroponics). These are a less common approach simply owing to the general expense required to operate them, in particular heating them up to the required climate to grow said products.

Herein lies the opportunity:

• We have a global demand for food

• We have the knowledge of the environments of which food types will thrive in

• We have a waste product from a growing industry that is also currently the road block for the wider implementation of artificial agricultural units.

In a world where data centers increase their interconnectivity and are networked throughout large continents, these agricultural units could assist in supporting them by utilizing their waste product, reducing their demand for energy, generating additional local jobs, - and most significantly, as far as the environment is concerned - eliminating food miles from having to grow products in tropical climates and import them thousands of miles.

Nearly any organic plant-based food type; produced locally at volume and scale, distributed locally at volume and scale. It enables an additional revenue stream for the data center operator or one of their partners. It utilizes a waste product that otherwise would have an overhead in dealing with it. And it creates a more abundant food supply to the local economy.

There are additional opportunities that exist as a result of this suggestion. The pressure on agriculture in recent years to be the savior of many of the world's problems continues to grow. There is a continuous battle between the food producers that wish to use the land for food production only; and the energy producers that wish to use the land to produce alternative crops - wheat for bioethanol, maize for anaerobic digestion, miscanthus/straw for biomass, or rapeseed oils for biofuels.

A challenge which is rarely communicated amongst the scientific literature in circles outside of the immediate agricultural industry, is a lack of knowledge surrounding soil types and their suitability to grow certain products. A soil's pH and macronutrient structure vastly affects its ability to grow a certain product. That is to suggest that products grown in one place cannot be grown in another. Some land is not sufficiently fertile to grow anything other than grass. Other land is so fertile that two crops a year can be grown on it. Staple crops, such as potatoes, can only be grown in very small pockets of countries, owing to their nutritional requirements to thrive and flourish.

The creation of artificial agriculture units alongside data centers would enable the new surplus land area to be allocated towards alternative energy production if needed. Poorer quality land can be used for livestock production.

The data center could be effectively housed subterranean in the basement of a building where the temperature is more stable; and then greenhouse installations are at ground level above, enabling the heat to naturally rise. Or, they could both be above-ground structures stored adjacent to each other.

Additionally, it enables the production of food in more hostile climates which are sometimes, bizarrely, the most suited to data centers. These are typically colder climates where raw intake air can be used to cool the servers and the exhaust hot air can be used to grow food for the local economy where it has never been able to be grown before owing to the harsh native climate. As there’s reduced requirement to cool any air, these data centers could operate at a fraction of the energy demand of an installation in a warmer climate.

As with many solutions, this is not a universal one - but rarely does a solution exist which proposes a universal win for the operators, local economies and global environment.

And even more remarkable - the bigger the data center, the more people it can help.

Definitely a topic that wants exploring a little further.

TH

Share

Find me on Linkedin

See more engineering posts

You may also find these of interest:

This could either be a billion-dollar idea or a complete failure. However, as data centers begin to reach hyperscale, the construction challenges - and subsequently the costs associated in overcoming them - are soon becoming prohibitive.

No matter how fast we innovate, the bottleneck is implementation. We simply cannot produce enough of everything fast enough; and even if we can, we're still relying on archaic infrastructure which requires significant modernization first. Without that investment, we don’t stand a chance at keeping up with the future demand that will be required. Therefore, my take on this is that perhaps we've got to look at a different part of the installation in order to make this a more viable solution for the time being.

So hear me out: how about instead of redesigning the whole data center to suit all the servers - the construction, cooling, power distribution, networking, everything…

Why don’t we just redesign the servers so that the roadblocks being faced don’t even exist in the first place?

I completely accept that any amendment to the design cannot impinge on it’s ability to perform as they currently do; so we’ll start with a brief, and then my solution may make a little more sense:

• Data centers have two main roadblocks in the construction - firstly, they get sodding hot; and secondly, they require a ludicrous amount of power. For a single AI hyperscale facility - it can use 20+ times the amount of power that New York City can use (the whole of NYC uses about 5GW; an AI hyperscale data center can use over 100GW).

  • So our server design or chipset architecture needs to be able to function ideally on less power.

  • Or, at the very least - the design needs to be able to run cooler. If the building has a lesser requirement for an extensive HVAC system, that’ll reduce a good lump of the power to run the building.

• We need to be able to stack them high and wide. Historically, two main types of servers are used - blade servers and rack servers. Both are favoured for a largely modular approach as an efficient way of cramming as much computing power in as compact space as possible, and also being easy to scale. If more capacity is needed, another server can just be installed into a rack.

  • Our solution also needs to be a modular design that’s an efficient use of space. (spoiler alert: if you were expecting anything circular, you’re about to be disappointed).

Quite simple really from initial impressions. The most common types of servers currently deployed are rack servers. I completely understand the modularity of them, but some parts of the rack system just don’t make sense to me from a practical perspective. I’m totally willing to be proven wrong and corrected on the reasoning for this, but I’ll cover my main points of confusion below:

1. Second-grade physics teaches every single child that heat rises. We’ve identified heat generation is a huge problem. So why on earth do we stack servers in a rack on top of each other? The further up the stack we go, the hotter those servers will become as they’re subjected to the heat rising from the servers below it. I'm fully aware that solutions have been provided to provide cooling in between the servers in the rack, but this is still a relatively inefficient solution in my view.

   A lot of effort has been put into overcoming the physics, whereas I would suggest it makes much more sense to redesign the server to align with physics. This is to suggest particular culprits for causing heat (the processor and potentially power supply) are put towards the top of each enclosure, and the cooler components are put towards the bottom.

2. More basic physics teaches us that a larger hot object dissipates more heat into its surrounding environment than a smaller one. Therefore, I'm intrigued to know why all components of the server are stored in the same space - the hot power supply and hot processor are stored together, a few centimeters apart; and then they are stored next to other hot power supplies and hot processors.

   This is to suggest an intrigue as to why the majority of the transformers for power supplies cannot be stored away from each server rack in a separate plant room, simply served by a cable, akin to a laptop charger. This will effectively split the heat production between two rooms and make the thermal management requirements much less in each respective space. I accept this isn't the biggest challenge to overcome; and it could result in equivalent requirements - but is perhaps a consideration that may yield some success.

3. The active cooling obsession. It appears that every solution to cooling that is required seems to be some form of active management. A fan that sounds like a Boeing 737 trying to take off or an expensive liquid cooling solution.

   The most recent iterations of direct-to-chip cooling utilizing glycol and other compounds are perhaps the most efficient iteration presently, but I would suggest they are far better than methods of passively dissipating heat through a refined choice of material and geometry to better suit this purpose.

4. Speaking of materials, why is everything made out of thin sheet steel? If these installations are known to get hot, steel is a relatively mediocre method of dissipating the heat. I accept its cheap, but fundamentally, if the same amount of cost has to be spent in keeping the installation cool, it would be a better investment to design the server correctly in the first place to mitigate ever needing such cooling solution. Therefore, a novel approach will suggest some alternative materials that wouldn't necessarily add large amounts of cost per unit but would produce a unit that’s more efficient in passively cooling itself.

So my solution, which I’m dubbing the ‘wafer server’. I’ll address each point roughly in the same order I mention them above:

• We need that modular goodness - so it’ll remain a flat square that can be packed tightly into a space, in close proximity to other servers.

• But, to help with the heat problem:

  • Spread the heat over a much larger surface area. So instead of a rack server measuring about 800mm deep laid on top of each other; we’ll make ours 2 metres. As the same amount of componentry can be spread over that whole surface area, It can obviously be a lot thinner now (hence the ‘wafer’ name). There may be space for more componentry, to house factorial capacity per server.

  • We won’t store them on top of each other, but upright, akin to blade servers. Very few server rooms have a problem with eaves height, and so there’s no issues there. We’ll stack them side-by-side, stood in lines like soldiers. If we still need to put cooling in between, we have that opportunity if needed.

  • We’ll do our level best to passively cool them with heatsinks, which can be finned for efficiency - similar to air-cooled motorcycle engines. Whilst we’re at it, those could be made of anything that’s a better heat dissipator than thin sheet steel. As these are barely going to be moved, they can be made even from fragile or brittle materials. My initial thoughts is intrigue to see how ceramics could help in this situation.

  • Finally, we reconfigure the servers so that as much as practicable, the processors and other heat-producing chipsets are placed near the top of the wafer. Power supplies are stored in an adjacent plant room with it’s own independent cooling solution - the server racks are simply fed by cables ran from floor level (could be ceiling mounted, but likely issues with condensation running down cables into the servers)

  • How that will look roughly:

    

    

As I stated above - I’m fully prepared for the pushback on this, and I’ve not come at this from any computer science standpoint; but I do find it fascinating that the buildings to house servers are being constructed with bleeding edge technology; and the components in the servers are being upgraded to bleeding edge technology - but we’re still relying on enclosure configurations designed by Compaq in 1993.

Even if it’s not the above, I’m certain there’s a better solution which will help plug the gap whilst infrastructure investment catches up.

TH

See more posts

Find me on Linkedin

Send me an email

You may find these of interest:

But there could be another opportunity that exists; instead of a large centralised manufacturing system whereby all of the above is contained under one roof, perhaps the future is better served by a more decentralised approach. In simple terms - close the big national factory down, and distribute the work through many smaller manufacturing shops operating nationwide on a local level.

I accept there will be a little conceptual inertia pushing back against this argument. For the past 200 or so years, pursuits have been made to minimize the amount of human contact required in manufacturing overall, and for good reasons:

1. For efficiency. Automated solutions don’t need to sleep; they don’t fatigue, and they can even be designed in such a manner to rarely require consumables or maintenance. They work in harmony, and do not have the contend with the dynamism of human emotions.

2. For safety and liability. Ultimately, if no humans are involved, no humans can get hurt; if no humans are hurt, there's no risk of lawsuits that threaten the bottom line of the manufacturing business.

3. For profit. Often, margins are thin enough from a highly leveraged capital investment, and they don't need to be reduced any further than absolutely necessary. Automated approaches can retain waste to barely conceivable levels, even operating at vast scale.

I’m continuously in awe of many of the automated solutions, but the elephant in the room with them is the cost - they’re typically ludicrously expensive; and many machines are bespoke to the products also.

A trend in the market over the past few years has resulted in a compounding issue occuring - whereby the assets required to house such manufacturing are sharply rising in capital value and rental value; along with the costs of utilities, insurance and local taxes to support their continued operation.

It must be considered that constraints can often come with operating out of a singular site; many companies find you can only extend so many times before an entire relocation is required in order to scale.

So, what's the answer?

Well, broadly speaking - it's an about turn, full circle, back to where they started - close the large manufacturing facility, and open multiple smaller ones.

The difference being, is that scale comes from the volume of facilities rather than the output of a singular production facility. I foresee a world where decentralized manufacturing will be the future, whereby localized facilities can exist to serve the existing customer base through small-scale batch manufacturing far more suited to the modern world and its requirements than the existing. But it begs a question - why this approach? Why can't we double down and push for even more scale?

Scale comes with its intrinsic difficulties. Often, the problems become exponential. A factorial increase in manufacturing size often means a significant amount of investment, which can not always be proportional or realistic within the scope of what's being produced, or the increase in production output.

For instance, doubling the manufacturing area on the factory floor is fine; but that could also require significant updates to the utilities, facilities, and infrastructure. It could result in uprated connections to the mains electric, and you may find the local grid infrastructure does not always have the capacity to serve your ambitions.

If there is a requirement for heating or cooling that becomes exponentially more difficult with a much larger space. Should that space be then segregated? How is it segregated? Is there a need for disparity between those spaces - in temperature, humidity, pressure, cleanliness, hygiene, or otherwise? If so, it needs to be sufficiently insulated between the two spaces to accommodate each. Is that going to require a larger HVAC system? Does that produce a lot of noise? What about dust? Will filtration be required? How will protect your employees from that increased risk factor? Is there pollution bleeding into the local community?

You get the point - scale seemed great until the mitigation works exceeded the costs of the extra manufacturing equipment. There's also an increased amount of scrutiny amongst private equity firms, and you must be willing to operate at a scale that makes sense for their investment targets as much as it does for your business ambitions. If the two do not align, you could be incredibly stuck with what's available to you. You may be producing something great - you want to make more of it, and your customers may want to buy it from you - but there’s too many second and third order effects.

So if scale isn’t the answer, what’s the argument for the decentralized approach?

Well, firstly - There’s employment:

In a new world where AI will inevitably replace many of the white-collar professions, there will be a large cohort of people perfectly capable of providing some form of economic output; but must accept that will involve retraining in new industries which do not necessarily leverage their existing skill set to a full degree.

Although the costs of labor may be decreasing, the cost of automation and robotics (especially the more advanced applications), will remain relatively expensive for the time being - until such scale operates to bring prices down.

So if the robotics are too expensive, and there’s likely going to be some cheaper labour, a decentralized approach makes perfect use of this scenario. Multiple local manufacturing facilities will facilitate creation of a significant amount of jobs within local communities. Manufacturing skills can develop to eventually create equivalence in terms of output and quality - and also has the benefit of some of those intrinsic human traits that can sometimes just put a little extra finesse into the manufacturing process.

Secondly - There’s the agility and local prosperity:

There will never be a time when a customer wants to wait longer for their products. If anything, the modern pressure is to reduce the friction between ordering and delivering.

Smaller facilities will be able to serve customers faster. Supplies can be sourced from local businesses. Transport, logistics, storage and distribution costs nearly disappear.

Performance metrics will be assessed just like any other previous approach. Except with decentralization, there is a possibility for employees to be incentivized through targets, profit share, or other such ways. This system intrinsically develops profitability and prosperity for the wider community - delivering a far more effective service for all consumers; in addition to maintaining (and even improving) the scalability of the output from the manufacturers overall.

The batch manufacturing methods will additionally reduce waste and other intrinsic benefits which come as a result of operating a smaller facility overall.

Thirdly - There’s a drastic reduction in nearly all overheads:

There’s typically few large sites in local areas, but plenty of commercial real estate for smaller ones. Hybrid and remote working has freed a lot of commercial space which could easily be repurposed for industrial use. A plentiful supply ensures a lower rental or capital cost to acquire a facility.

And of course, these facilities are smaller - with less requirement for excessive power or other infrastructure. A gentler print on the local economy from production side effects or pollutants, and can easily be purchased as turnkey solutions if needed. Equipment is not as specialist; and smaller machines are cheaper to buy, easier to repair, and there’s a greater safety of continuity. If one manufacturing facility needs to be closed for maintenance - the workload can temporarily be spread amongst the other local facilities whilst repairs are carried out.

That’s a good start, but it could be better:

Finally - The holy grail of production. Not only eliminating operational overheads altogether, but actually monetizing them.

There could certainly exist a time whereby licensing agreements could ensure that manufacturing could operate out of existing facilities and other businesses.

For instance, car parts could be manufactured in the back of service centers of dealerships. Medical devices could be made in-house in hospitals.

• Many existing businesses have the spaces themselves already, and they simply have to allocate a portion of it towards manufacturing.

• Meanwhile, the original manufacturers simply license their staff, intellectual property, manufacturing expertise, and equipment to the business to exist and operate out of their own premises.

• Parts could even be produced on-demand, ensuring that no inventory needs to be held. Additive manufacturing and CNC machining have made this a very realistic and achievable premise.

The franchisee model could also exist in perfect harmony with a decentralized manufacturing setup, which is a business model that has served titans of some industries - if you need convincing, look at the fast food industry. McDonald's operates an extremely successful model whereby their products are manufactured, prepared and sold globally by teams from all types of backgrounds - and with no variation in size, appearance, taste, or otherwise.

Nearly sounds too good to be true, doesn’t it?

It should be stated that this solution is not a universal one; and it’s applicability will ultimately depend on the business, the product, the skillset of the staff available, and the availability and distribution network it requires.

Centralised manufacturing may absolutely be the correct procedure in some cases. For some businesses, a decentralized approach would invite more risk of security or increase cost from having a multiplicative effect for complex systems. Such examples would be data centers or defense weapons manufacturers - ensuring sufficient security protocols for many, many sites would largely be prohibitive as a practice; and it would be far safer to ensure all assets are contained under one roof space.

However, this does highlight that in doing a decentralized approach for those that can facilitate the shift, will ensure that larger industrial units can be allocated in the appropriate areas for such businesses with a centralized requirement; and as a result less pressure is put on certain districts of the existing infrastructure as it serves less larger businesses in any one area.

The biggest challenge in the decentralized approach is the increased staff management and support structure that would inevitably need to exist. However, one could regard this as a relatively minor inconvenience in return for the significant cost saving that comes from passing much of operational overhead onto a third party and even monetising it.

This problem is simple enough to solve though - extra bodies, and implementation software SOPs, monitoring and quality control are all problems which have been tackled numerous times by businesses with alot of success.

This is a problem which is an easier problem to solve than magically attempting to find a few extra terawatts of energy in a grid infrastructure that is already creaking at the seams; or attempting to magic out additional area from sites which are already overdeveloped as they are.

This may not be perfect, but it's a suggestion of how manufacturing can return as an industry that is desirable, economic, profitable, and worthwhile for white-collar workers. It leverages their development, intelligence, and nurtures and values the qualities unique to human beings.

This solution may not be the silver bullet, but it could just be the start of a golden era for manufacturing.

TH

Find me on Linkedin

See more Engineering posts

So we absolutely have a demand for more data centers, but the main problem to meeting this demand isn’t what you might suspect.

It’s not the supply of land. There’s plenty of land to build them; or existing industrial sites that can be converted.

It’s not the supply of capital. There’s plenty of money to fund them; generally from real estate funds and venture capital firms.

It’s not in the supply of equipment and hardware. And there continues to be sufficient technology to enable the construction of them, and the challenges the latest hyperscale installations bring with it - the cooling, equipment density, hardware, and facilities management.

The problem is exclusively the supply of power.

The first (minor) problem - Generating enough of it.

Between conventional methods and renewables, we have no shortage of methods in terms of generating power. There is definitely enough potential to generate sufficient power to meet the demand; but it will require a lot of investment, and very quickly.

In terms of power requirements, a conventional data center is an incredibly hungry beast; but compared to a hyperscale AI data center, it is nearly incomparable. To give some illustration of this:

• The whole of New York City consumes around 5 Gigawatts of power per day.

• A conventional data center operating cloud or traditional compute functions - consumes around 60 Gigawatts per day.

• A hyperscale AI data center consumes closer to 300 Gigawatts per day.

Up to now, power providers and utilities companies have typically achieved adding around 5-6 Gigawatts of additional power capacity per year in most areas.

In order to keep up with the scaling, this rate needs to quadruple effectively. Current research from data center investors suggest that the traditional methods of building power is now completely insufficient for modern requirements - and it is going to require every solution to be rolled out to meet the demand - renewables and non-renewables - solar, wind, hydroelectric, geothermal, gas, coal and nuclear.

There’s compounding issues here also though - the turbines used to convert the heat into electricity are only manufactured by around 3 companies in the USA (Siemens, GE Vernova and Mitsubishi Electric). There’s currently a 4 year waiting list for one of these.

The second (major) problem - Getting it distributed.

Much of the distribution grid infrastructure in certain parts of the world dates back to the second world war - and it was never designed with sufficient capacity for running these levels of demand or generation.

This makes site selection extremely challenging, and many are unsuitable on this premise alone. Grid upgrades would be slow, expensive, and typically involve government cooperation - the rate of development required could simply not be practicably met.

It is thought that eventually grid-independent infrastructure will be built adjacent or on data center sites, likely encompassing their own power generation sources in order to meet their own demand and to take reliance off the grid structure. Data centers have a stable and continuous demand requirement - requiring power load for 24 hours of the day, so the eventual solution would likely not be entirely renewable by virtue of the building’s requirements.

The most promising solution would appear to be Small Modular Reactors (SMRs) - often referred to as Nuclear Microreactors; which have been tested and proven in many applications such as submarines for a number of years. Scaled versions suitable for the datacenter application are still anticipated to be around 6-8 years away from becoming commonplace however.

Solutions?

In short, not many. Currently efforts are being directed into enabling as many existing sites as practicable. It’s unlikely that industry will be willing to wait longer than necessary to develop sites - approximate figures suggest that the average data center would cost around $7900 per minute of downtime. Therefore, delaying it’s implementation could easily be a costly exercise.

Whilst grid and generation upgrades are going to be necessary in the long-term (and therefore will be continued to be proactively worked on), efforts will turn into quicker solutions in the meantime - efficiency in the chipsets and compute architecture itself, or in the cooling, insulation and building design.

My two takes are to consider alternative designs for data centers; and also to consider decentralising the manufacturing of other industries, to enable this space and capacity to be utilised by data center operators. Both articles on those solutions below.

TH

Find me on Linkedin

Other Engineering Posts

I couldn’t possibly even begin to comment on the efficacy of that, or suggest the likely truths - ultimately, I don’t think any of us will either know. But if we humour the premise that it was, then it could suggest as a more stark indicator of a new global threat - bioterrorism.

It’s a concept featured in plenty of films over the years. Mission Impossible 2 (released in 2000) had a plot which centered around the attempted distribution of ‘Chimera’ - a biological viral agent design to infect populations en masse. Before that, Outbreak (1995) was loosely based on a similar premise.

But it’s never seems to have escaped from the screen into reality. And if the COVID virus was fabricated, then it’s probably the first example of a widespread use of a biological weapon on the world at large.

Whilst there were many remarkable aspects about the pandemic, one worth highlighting was the incredible application of virologists, chemists, biologists and medical engineers to develop, test and release a vaccine in a record time; and swift implementation saved many from any further harm being caused.

If we flip to another related domain in 2020, Deepmind released its findings on the development of utilizing AI to model protein folding in Amino Acids; which, broadly speaking, will enable the development of medications at a far more rapid degree. Such findings continue to be built upon with newer models, enabling pharmaceutical developments to occur at a rate that could have only been dreamed of only a few years ago.

There’s a dark dichotomy to such advancements however - where the implementation of one side of this technology can serve to mitigate humanity’s biggest threats; the other side could be used to create it.

Criminality increasingly has lesser requirement for physical presence - braun was once the staple tactic of intimidation to manipulate; now the brain, and psychological mechanisms appear to persevere in the modern age. Cybercrime is a perfect example of this - with ransomware attacks (where data is encrypted by hackers; and it’s decryption and release is dependent on the payment of a ransom) suggested to have increased by 126% in the past year alone. Cybercrime overall typically continues to rise by approximately 47% per year on current estimates, expecting to cost the global economy around $10.5 trillion in 2025.

If criminality is able to remain faceless; and technological advancements now enable threats detrimental to the health of the entire global population, it is entirely plausible that sooner or later, the next ransomware won’t be for a decryption key, but for a chemical formula of what a populus has been infected with.

Herein lies the opportunity for pharmaceutical companies - one of triumph, redemption and profit.

Refinement of spectroscopy and other analytical methods should be prioritised in order to identify such threats instantaneously; and building and safeguarding AI models that are capable of synthesizing vaccines, anti-venoms and cures as rapidly as the criminals can produce them.

It enables the opportunity for pharmaceutical companies to be able to monetise the development of such drugs and compounds; monetise the production and manufacture of them, and finally the distribution of them. This unique skillset safeguards their role and position in society - as no longer a desirable entity, but a necessary one.

It would be true that this does place a societal dependence on privatised corporations, but a capitalist economy does help to mitigate some of such risk - as competing corporations will assist in keeping the economic factoring to a fair degree, ensuring market value is preserved for the benefit being delivered.

It may even be argued the power this is thought to potenitally enable already exists within the pharmaceutical companies; and is cited as a critical reason for much of their unpopularity throughout some demographics. By virtue of the sector they operate in, they are largely seen as a necessary evil by many who disagree with their pricing structures and fiscal models.

The above opportunity facilitates some of their revenue to be generated as a subsidized commodity by global governments - enabling the entire global population and economies to effectively assist in contributing to ensuring the health of the population as a whole is retained and managed. The subsidization would permit lower market prices to be offered for desirable medications for lifestyle conditions and comfort.

Although I would suspect we are some time off bioterrorism become a hugely tangible threat, the potential does exist. Fortunately, there remains a large imbalance in the skillset of the two sides - whereby the top medical scientists and pharmaceutical experts are typically representing the vaccine and solution-based side, and those attempting to synthesize new viral weapons are more likely to have a significantly more rudimentary understanding of the mechanisms.

I believe that a proactive approach in this sector would be the wisest course of action - completed and refined in private, and ready to deploy instantaneously should a threat arise. There should be no distribution or publication of the extent of resources that have been developed or gathered in preparation for vaccine and anti-venom production - adhering to similar safety and clearance protocols typically reserved for military and intelligence projects. The reality is that any publicised mechanisms used to develop vaccines could easily be reverse-engineered to produce more deadly biological agents.

There will always remain some skepticism amongst the wider population when assessing the morality of pharmaceutical companies, as they will naturally be for anyone appearing to profit from others ill-health. However, these profits enables the wealth of drug synthetization, manufacture and distribution to grow and support the entire populus. So perhaps this could be the time where these companies are no longer seen as the necessary evil; but just necessary.

Of all the challenges big pharma faces, I’d argue that Corporate Social Responsibility remains one that little progress has been made on. There’s an opportunity for redemption in the eyes of the public perception; reinforcing that although profits may be generated, they are just the byproduct of being an instrumental component of all global communities.

TH

Find me on Linkedin

See more Medical Engineering Posts