...
.Nscale Locks $3.5 Billion Figure Robotics Compute Deal  ·Qatar’s Meeza Lands Major Hyperscaler Deal for 8MW ·Qualcomm Strikes Amazon AI Chip Deal, Opens Door to $4 Billion Stock ·Hitachi Energy Bets $300M on China Grid Manufacturing Corvex Builds Toward 8MW Cloud Infrastructure Footprint LITEON Bets $176 Million on DCX Liquid Cooling EdgeConneX Backs Singapore’s AI-Ready Tropical Data Center Testbed
.Nscale Locks $3.5 Billion Figure Robotics Compute Deal  ·Qatar’s Meeza Lands Major Hyperscaler Deal for 8MW ·Qualcomm Strikes Amazon AI Chip Deal, Opens Door to $4 Billion Stock ·Hitachi Energy Bets $300M on China Grid Manufacturing Corvex Builds Toward 8MW Cloud Infrastructure Footprint LITEON Bets $176 Million on DCX Liquid Cooling EdgeConneX Backs Singapore’s AI-Ready Tropical Data Center Testbed

The Procurement Question Every Operator Should Ask Twice

A cooling system can look exceptionally disciplined while its water behavior remains almost impossible to predict from the procurement documents

Share
data center cooling procurement

A cooling system can look exceptionally disciplined while its water behavior remains almost impossible to predict from the procurement documents alone. The equipment may carry a clear peak flow figure, a defined heat-rejection capacity, and a polished sequence of operating modes, yet none of those details necessarily tells an operator when the system will actually begin using water. The missing information often sits inside the control sequence rather than the mechanical specification, where a small change in an ambient threshold can alter when adiabatic assistance starts and how long it remains active. That makes the procurement conversation less about buying enough cooling capacity and more about understanding the conditions under which the equipment changes its behavior.

Service Public Policy Newsletter Leaderboard 970x118 1

The issue becomes more important as operators compare hybrid heat-rejection equipment across sites with different weather patterns and different operating assumptions. Dry operation depends on the relationship between the ambient condition, the temperature of the circulating fluid, the available heat-transfer surface, airflow, and the load presented to the system. Adiabatic assistance changes that relationship by using evaporation to lower the effective temperature of the entering air, which can improve heat rejection when dry operation approaches its usable limit. The control system therefore decides more than whether a spray pump or valve turns on because it determines where the equipment sits along the boundary between dry operation and water-assisted operation. That boundary can shift with load, return temperature, airflow, humidity, and the control deadband selected by the designer. Procurement documents that disclose only the maximum water flow leave the operator without the information needed to understand that boundary.

The Datasheet Number That Never Shows Up on Your Water Bill

Peak gallons per minute has an important place in equipment selection because the operator needs to understand the maximum water delivery that the system can demand under its defined design condition. That figure helps engineers check supply infrastructure, piping, valves, pumps, treatment arrangements, and other water-side requirements before equipment enters service. It does not, however, explain how frequently the system approaches that condition or what control sequence causes water demand to begin. A peak value can describe the upper edge of a system without describing the path that the system follows before reaching that edge. The distinction matters because hybrid cooling equipment can move through dry, partially assisted, and more intensive water-supported states as ambient conditions and thermal load change.

Peak Flow Is a Capability, Not an Operating Story

The same issue appears when a vendor presents a maximum water draw alongside a nominal heat-rejection capacity and leaves the operator to infer the annual behavior. A maximum draw tells the water system what it must be prepared to supply during a defined condition, but it does not reveal the point at which water assistance begins, the rate at which water demand rises, or the conditions that return the equipment to dry operation. Those variables depend on control logic and equipment characteristics that may sit outside the headline datasheet. A procurement team therefore needs to ask whether the quoted flow represents continuous operation at the design condition, a maximum during staged operation, or a theoretical value reached only under a narrow combination of load and weather. The answer changes the meaning of the number without changing the number printed on the specification sheet.

Service Advisory Services Leaderboard 970x118 1

A useful vendor response should connect peak flow to the physical and control conditions that produce it. The operator should ask for the entering-air condition, leaving-air or fluid condition, heat load, fan state, pump state, and water-control state associated with the quoted maximum. The response should also identify whether the system can operate at intermediate stages and whether water demand follows thermal demand or simply follows an ambient trigger. This prevents a procurement team from treating peak flow as though it were a proxy for annual consumption. The operator should also ask whether the quoted figure includes only evaporated water or whether the calculation incorporates other water losses associated with the system architecture. That level of disclosure turns a static capacity number into the starting point for understanding actual operating behavior.

The Missing Curve Between Dry and Wet

The more revealing procurement document is not a single point but a description of the operating curve between dry operation and water-assisted operation. An operator should ask the vendor to show how the equipment behaves as ambient temperature rises while the heat load remains controlled, and then repeat the exercise with the heat load changing while ambient conditions remain similar. That comparison can reveal whether the control system responds primarily to outdoor temperature, heat-rejection approach, leaving-fluid temperature, or a combination of signals. It can also show whether water assistance begins gradually through staged modulation or arrives as a binary change from dry to wet operation. Such information matters because two machines with similar peak water capability can create very different water profiles when their control sequences differ.

The operator should also request the assumptions behind every water-use representation because modeling can otherwise create a false sense of precision. If a vendor calculates water demand from a fixed design condition, the resulting figure may say little about how the equipment behaves when load and weather move together throughout a normal operating period. A more useful submission explains the weather source, load profile, fluid temperatures, control sequence, water chemistry assumptions, and treatment strategy used to produce the modeled result. The model should make clear whether water demand represents evaporation alone or includes blowdown, drift, leakage allowances, and other losses where relevant. This approach does not eliminate uncertainty, but it identifies where uncertainty enters the procurement decision and gives the operator something concrete to challenge before contract award.

At What Degree Does Your System Start Drinking Water?

The most important question in a hybrid cooling procurement can be remarkably simple: at what ambient condition does water assistance begin? The answer should never appear only as a broad phrase such as “high ambient,” “peak conditions,” or “extreme weather” because those descriptions leave too much room for interpretation. The operator needs the actual control variable, the associated threshold, the applicable load condition, and the temperature or thermal margin that the controller is protecting. A system may initiate adiabatic assistance because a measured ambient condition crosses a programmed threshold, because the circulating fluid approaches a limit, or because the controller predicts that dry operation will no longer maintain the required condition. Each approach creates a different water-use profile even when the underlying cooling equipment remains unchanged.

Find the Activation Setpoint Before You Sign

The questionnaire should ask the vendor to identify the primary activation signal and every secondary condition that can override it. The request should cover ambient dry-bulb temperature, ambient humidity or wet-bulb condition where applicable, fluid entering and leaving temperatures, fan speed, pump state, thermal load, and any equipment protection limits. It should also ask whether the setpoint remains fixed or changes with load, fluid temperature, time delay, or another control input. A fixed threshold can be easy to understand, but a variable threshold may better reflect the actual heat-transfer capability of the equipment. What matters for procurement is that the operator can see the complete rule governing activation rather than receiving one temperature value detached from the control sequence around it.

The activation point should also be treated as a contractual operating parameter rather than a casual commissioning adjustment. If a vendor can change the threshold after installation without documenting the effect on water behavior, the operator may inherit a system whose actual water profile differs from the one evaluated during procurement. That does not mean the threshold can never change because operating requirements, equipment aging, water quality, or control optimization may justify later adjustments. It means the baseline setting should remain visible, traceable, and linked to the performance assumptions used in the original selection. A procurement document can therefore ask for the factory default, the commissioned value, the allowable adjustment range, and the conditions under which an adjustment should occur. That record creates a direct line between the equipment specification and the behavior the operator expects to manage over its operating life.

One Degree Can Change the Operating Window

A small movement in an activation threshold can matter because outdoor conditions do not arrive as isolated design points. They move continuously through mornings, afternoons, evenings, seasonal transitions, and changing humidity conditions, while the thermal load inside the system also changes. When the activation boundary moves, the system can spend more or less time using evaporative assistance even though its peak cooling requirement remains unchanged. The resulting water behavior therefore depends on where the control sequence places the boundary between dry capability and water-supported capability. This is why a procurement document that contains only peak water flow cannot explain the annual operating pattern of the machine.

The procurement template should finally require a written explanation of why the selected activation threshold exists. The vendor should identify the thermal constraint that the threshold protects, the performance relationship that changes beyond it, and the reason the controller does not wait for a later condition before introducing water. That explanation allows an operator to separate an equipment limitation from a conservative control choice, which can have very different implications for long-term water behavior. The same request should cover the conditions that allow water assistance to stop because the return path matters as much as the activation path. When procurement records both sides of that transition, the operator can evaluate the equipment as a controlled thermal system rather than as a machine with a single water-flow rating.

Dry Hours Are Not Free Hours, They Are a Strategy

Dry operation should not be treated as the absence of a decision because the controller actively decides whether the equipment can remain within its thermal limits without evaporative assistance. That decision can depend on ambient conditions, fluid temperatures, fan capacity, thermal load, and the margin available before another cooling stage becomes necessary. A system designed to maximize dry operation will therefore monitor those conditions and use the available heat-transfer capability before introducing water. A system with a more conservative sequence may introduce water earlier to create additional thermal margin even when dry operation remains technically possible. Procurement needs to expose that control philosophy because the physical equipment alone cannot reveal how aggressively the system uses its dry capability.

Ask What Keeps the System Dry

The vendor questionnaire should ask what conditions must remain true for dry mode to continue and which condition has priority when several signals disagree. It should ask whether the controller considers the actual thermal load before enabling water assistance, whether fan speed can increase before water begins, and whether fluid temperatures can float within an approved operating range. The operator should also ask whether the system uses predictive logic, simple threshold control, staged response, or another sequence when ambient conditions approach the dry operating boundary. These questions matter because a controller can protect the same thermal target through very different combinations of fan power, fluid temperature, airflow, and water use. A procurement process that does not request those combinations cannot properly compare two systems that carry similar capacity ratings.

The questionnaire should also ask what happens when the system moves back toward favorable ambient conditions. Returning immediately to dry mode may create unnecessary cycling, while remaining in water-assisted operation longer than necessary can extend water use beyond the period that justified it. A sound control description should therefore explain the return threshold, the required stability period, and the conditions that prevent repeated switching. The operator should ask whether those values can be reviewed through trend data after commissioning and whether the controller records each mode transition. This creates an operational feedback loop in which the intended dry-first strategy can be checked against actual behavior instead of being accepted as a design statement.

The Control Sequence Is Part of the Equipment

A procurement specification becomes stronger when it treats the control sequence as part of the cooling system rather than as software that arrives after the mechanical design has already been selected. The sequence determines how the machine uses its available heat-transfer surfaces, fans, pumps, valves, and water-assist components as conditions change. That makes the control logic directly relevant to the resource behavior the operator will experience. Simulation research has shown that alternative control strategies can materially change plant behavior even when the underlying equipment configuration remains comparable. The procurement process should therefore require enough sequence information to reproduce the intended operating logic in a model and verify it during commissioning.

The vendor should be asked to provide the control hierarchy in plain technical language before the equipment receives final approval. The hierarchy should identify which condition starts the first response, which condition starts water assistance, which condition increases water demand, and which condition forces a return to a more protective operating mode. It should also identify alarm states, sensor failure responses, communication failures, and fallback modes because a water-saving sequence that depends on unavailable sensor data may not behave as intended. Operators should ask whether the equipment defaults toward dry operation, water-assisted operation, or another safe state when a critical input becomes unreliable. That answer belongs in procurement because fail-safe behavior can materially change water demand during abnormal operation.

Staging Logic Beats System Size Every Time

Cooling capacity tells an operator how much heat a system can reject, but it does not explain how that capacity arrives as operating conditions change. A large system can use water in a coarse sequence, while a smaller system can use water through several controlled stages that respond more closely to the actual thermal requirement. That difference matters because water demand does not necessarily need to rise in proportion to the nameplate size of the equipment. The controller can instead combine fan speed, fluid temperature, airflow, and progressively enabled water-assist stages to match the heat-rejection requirement. Procurement should therefore ask how many independent stages exist, what activates each stage, and whether the stages can operate simultaneously at different levels.

Staged Water Delivery Changes the Procurement Question

A vendor questionnaire should require the sequence for every water-assisted stage rather than accepting a single statement that the equipment has variable control. The response should identify whether individual nozzles, spray banks, pumps, valves, or equivalent components can operate independently and what signal causes each additional stage to engage. The operator should also ask whether the control system can reduce water delivery when the thermal requirement falls without first returning completely to dry operation. That answer reveals whether the system has meaningful turndown or merely alternates between broad operating states. A procurement comparison becomes much clearer when vendors must describe the same thermal progression using the same sequence of questions rather than presenting different interpretations of the word “modulating.”

The next question should address how the controller handles changing heat load at the same ambient condition. A system that responds only to outdoor temperature may introduce the same water-assist stage even when the thermal requirement differs materially, while a load-aware sequence can adjust its response to the actual heat that must leave the system. The operator should ask for a control narrative showing the response to rising load, falling load, rising ambient temperature, and falling ambient temperature separately. The narrative should also explain whether the equipment prioritizes additional airflow, greater water application, or another available cooling resource at each step. This makes staging a measurable procurement characteristic rather than a general claim about flexibility.

Turndown Reveals What the Peak Rating Hides

Turndown becomes especially important when the equipment does not operate continuously at its design heat load. Cooling systems often encounter changing thermal conditions, and the ability to reduce active equipment in response can affect both electrical demand and water behavior. A system with coarse staging can remain in a higher water-use state after the thermal requirement has fallen if its next lower operating state would provide insufficient margin. A system with finer staging can reduce the active water-assist capacity while maintaining the required thermal condition. The procurement document should therefore ask for the lowest controlled water-assist state, the conditions that support it, and the sequence used to move between stages.

The final staging question should connect water delivery to thermal load rather than treating temperature as the only trigger. The operator should request a sequence that shows what happens when heat load rises while ambient conditions remain stable and what happens when ambient temperature rises while load remains stable. Those two scenarios can reveal whether water demand follows actual heat rejection requirements or whether the control system reacts primarily to a weather threshold. The procurement record should preserve those responses because they provide a basis for commissioning tests and later operating reviews. Staging then becomes part of the equipment’s expected behavior rather than a feature that disappears behind a general capacity specification.

Why Identical Machines Keep Different Thirst in Different Cities

A cooling machine does not experience a generic climate because every site exposes it to a different sequence of temperature, humidity, load, and operating conditions. The same equipment can therefore follow different control states at different sites even when the installed hardware and design heat load remain unchanged. A procurement comparison that uses only one design weather condition removes the sequence of conditions that actually drives mode changes throughout the year. The operator should ask for a location-specific simulation based on hourly weather rather than accepting a universal annual water assumption. Such modeling allows the procurement team to see how the activation threshold interacts with the actual weather pattern at the proposed site.

Climate Must Enter the Model Before Procurement

Hourly weather modeling also creates a better way to test the activation setpoint established earlier in the procurement process. The operator can see how frequently the modeled conditions approach the threshold, how long the equipment remains in water-assisted operation, and whether the controller returns to dry operation as conditions improve. The model should preserve the load profile alongside the weather sequence because the same ambient condition can produce different cooling responses at different loads. It should also identify the weather source and the treatment of missing, abnormal, or representative conditions so that the result remains auditable. This approach replaces a generic climate assumption with a site-specific operating story that the operator can challenge before purchasing the equipment.

The request should extend beyond outdoor temperature because humidity and psychrometric conditions can affect the usefulness of evaporative assistance. A dry atmosphere can support a different evaporative cooling response from a humid atmosphere even when the measured dry-bulb temperature appears similar. The operator should therefore ask which weather variables drive the model and which variables directly influence activation, staging, and water demand. The submission should also identify the thermal fluid conditions and load assumptions used alongside the weather data because those variables determine how much heat the system must reject under each modeled condition. A climate model becomes useful for procurement only when it represents the control inputs that actually cause the equipment to change state.

The Site Determines the Operating Story

The procurement questionnaire should require the vendor to rerun the cooling model when the proposed site changes rather than transferring the original water estimate to the new location. This matters because climate can alter not only the amount of water assistance required but also the timing of when the equipment needs it. A system selected using conditions from one location can therefore produce a materially different operating sequence when installed somewhere with another weather profile. The operator should ask for the same control logic and equipment assumptions to be applied consistently across locations so that the comparison isolates the effect of climate. This produces a much more useful procurement record than comparing unrelated vendor assumptions for different sites.

Location-specific modeling should finally include the quality and characteristics of the available makeup water where the system depends on recirculation and controlled blowdown. Water chemistry can influence how the system manages concentration, treatment, and discharge, which means the same cooling design can require different water-management strategies in different locations. The procurement question should ask what water-quality assumptions support the vendor’s modeled behavior and what happens when actual makeup-water characteristics differ from those assumptions. The response should identify any treatment constraints that can force additional blowdown or restrict the intended operating range. That information connects climate modeling with water-side reality and prevents the annual model from treating the water source as an abstract input.

The Second Number That Makes Procurement Honest

The procurement template should ask for two water numbers, with each number answering a different question. The first should describe the maximum instantaneous water requirement under the defined operating condition, while the second should describe modeled annual water consumption under a stated site-specific load and weather profile. Neither number replaces the other because peak demand governs water-system readiness while annual demand describes the cumulative operating consequence. The annual figure should come from the same control sequence used to support the peak figure so that the two numbers describe one coherent system rather than two separate calculations. This pairing gives the operator a much clearer view of what the equipment can demand and how that demand is expected to unfold over time.

Peak Draw Needs an Annual Companion

The annual model should not stop at evaporation because water can leave a recirculating cooling system through several pathways. Blowdown removes concentrated water to manage dissolved solids, while drift and other losses can add to makeup-water requirements depending on the equipment arrangement. The operator should therefore ask the vendor to separate evaporation, blowdown, drift, leakage assumptions, and any other modeled losses instead of presenting one combined water figure without explanation. That breakdown makes it possible to identify whether the annual result depends primarily on heat rejection or whether water-treatment and control practices materially influence the result. It also prevents a procurement team from confusing water that performs the cooling function with water that enters the system because of water-quality management.

Recirculation deserves the same level of scrutiny because the water moving through the equipment is not necessarily the same as the water entering the system from the source. The operator should ask how the model represents recirculated water, makeup water, blowdown, and any water recovered for reuse within the system. A vendor should also state whether the modeled annual result assumes a particular treatment regime or concentration-control strategy because those assumptions can change makeup-water requirements. This makes the second number more than an annual estimate because it becomes a structured water balance that the operator can compare against measured flows after commissioning. Procurement becomes much more transparent when the model explains where the water goes rather than simply stating how much water the system uses.

Make the Water Model Auditable

An honest procurement model should allow the operator to reproduce the major logic without requiring access to proprietary software or undocumented vendor assumptions. The vendor should provide the weather source, load profile, activation setpoints, staging rules, water-quality assumptions, and calculation methodology that produced the annual result. The operator should then be able to change one assumption and understand why the modeled water behavior changes. This does not require disclosure of proprietary algorithms or intellectual property because the procurement requirement concerns operating inputs, outputs, and control behavior. The objective is to make the water forecast reviewable rather than turning the forecast into an unchallengeable vendor number.

The questionnaire should also ask the vendor to provide separate results for dry operation, water-assisted operation, and abnormal or protective operating states where those states affect water demand. The model should explain how each operating state begins, how it ends, and which conditions cause movement between them. It should also identify whether control overhead, such as additional pumping or auxiliary operation associated with water assistance, appears separately from the water balance. Those details help the operator understand whether the system reaches its annual water result through frequent short periods of assistance or longer periods of continuous operation. The same structure can later support commissioning because the operator can compare actual mode transitions with the states represented in the procurement model.

Ask Twice Today to Avoid Explaining for Ten Years

The strongest cooling procurement questions do not stop at whether the equipment can meet the design condition. They ask when the equipment changes state, what causes that change, how much water each state requires, and how the sequence behaves as thermal and weather conditions move together. They also ask for the control assumptions that sit between the mechanical specification and the annual resource requirement. This approach turns the activation setpoint from a commissioning detail into a procurement parameter that can be reviewed before the equipment is ordered. The operator then buys a defined operating behavior rather than relying on a headline capacity number to describe the system’s future water profile.

The phrase “ask twice” captures the structure of the procurement decision rather than a literal repetition of the same question. The first question asks what the equipment can demand at its most demanding defined condition, while the second asks how the control system is expected to behave across the conditions that occur during normal operation. The first protects the water infrastructure from an underestimated peak requirement, while the second protects the operating model from an unexplained annual demand. Together they force the vendor to connect equipment capability, control logic, climate conditions, staging, and water accounting. That connection gives the operator a technical record that can remain useful after procurement teams, project teams, and commissioning teams have moved on to other responsibilities.

Buy Predictability, Not Just Capacity

The long-term value of this approach comes from preserving the relationship between procurement assumptions and operating behavior. A cooling system will encounter changing loads, changing weather, changing water conditions, maintenance events, and adjustments to its control sequence, but the original procurement record can remain the reference point for understanding those changes. When the activation threshold, staging rules, and annual water model remain visible, later decisions can be tested against the assumptions that shaped the original purchase. When those details disappear into undocumented control settings, the operator has to reconstruct the equipment’s intended behavior from operating evidence after the fact. Procurement documentation therefore becomes part of the technical memory of the cooling system rather than a document that loses value once the purchase order closes.

The operator’s final question should be whether the vendor has explained the system well enough that its future water behavior can be understood without the vendor standing beside the machine. If the answer depends on undocumented control settings, generic climate assumptions, or an annual water figure with no supporting balance, the procurement record remains incomplete. If the answer includes the activation threshold, the control hierarchy, the staging sequence, the site-specific weather model, and the relationship between peak draw and annual consumption, the operator has a far stronger technical baseline. That baseline does not promise that future water use will never change because real operating conditions inevitably change over time. It does provide a defensible way to understand why water use changes and whether the cooling system continues to behave according to the operating philosophy that procurement originally approved.

Service Podcast Leaderboard 970x118 1
[simple-author-box]

More from AI Infrastructure

AI Load Recovery: What ERCOT, PJM and National Grid Are Signaling

The period after a grid disturbance can become an important part of the electrical

AI Capacity Planning Needs a Thermal Upgrade Calendar

The servers can arrive on time while the capacity they represent remains unavailable. Power

Designing For Chip Generations, Not One Generation: The Modular Refresh Cycle

The first rack in a new high-density hall marks a particular point in computing

COMPUTE WEEKLY

The briefing that 40,000+ tech leaders read every Monday. Sharp, fast, essential.

Building an AI Startup Without Owning GPUs

Not owning GPUs has become the default, deliberate strategy for building an AI company — not a compromise founders accept reluctantly. H100 rental rates fell 64-75% in fifteen months, a dense ecosystem of neoclouds and inference-as-a-service providers now lets startups skip infrastructure entirely, and credit programs can fund a company’s first year before a founder writes a check
Most Read

Ireland’s experience with data-center expansion became less about stopping construction than about changing the

A data hall can meet its opening-day layout and still contain a future expansion

A data center budget can remain numerically intact while its economic position deteriorates around

A data center schedule can begin moving well before major site construction starts, because

A compute node sitting behind a garage door can perform the same basic computational

Disruptor Spotlight

Cerebras Systems

The chip that makes Nvidia nervous. Cerebras’ Wafer Scale Engine is rewriting the rules of AI inference at scale.
Faster
0 x
YoY Revenue
0 x
Transistors
0 T
Market Pulse
MSFT
+1.02%
NVDA
+0.66%
AMZN
-0.078%
AMD
-6.95%
TSMC
-2.98%
Indicative only · Not financial advice
Upcoming Events

TBC

The AI Infrastructure Race

WEBINAR · ONLINE
The AI Infrastructure Race: Won on Power, Land and Trust — Not Capital
MAY
0
AI Infrastructure Summit
DUBAI · IN PERSON
MEA’s premier AI infrastructure event.
JUN
0 0
Compute Forecast Summit
SINGAPORE · IN PERSON
Our flagship APAC event. Early bird open.
Latest Moves
Live
ecolab
Ecolab Deepens Cooling Strategy With $4.75B CoolIT Acquisition
Ecolab is making one of its biggest moves yet into AI infrastructure after completing its $4.75 billion acquisition of liquid cooling specialist CoolIT Systems
Pure DC AVK Europe data center microgrid Dublin 110MW AI infrastructure Ireland 2026
Pure DC and AVK Deploy Europe’s First 110 MW Data Center Microgrid in Dublin
The Pure DC Dublin microgrid has made history as Europe’s first large-scale on-site data center microgrid, launched in partnership with power solutions provider AVK at Pure DC’s campus in Ireland.
Pace Digitek
Pace Digitek Partners With MEGMEET to Expand AI Data Center Power Business
India’s AI infrastructure ecosystem continues to mature as domestic technology manufacturers move beyond traditional telecommunications and industrial markets toward high-growth digital infrastructure opportunities
Follow Compute Forecast
11K followers
1200 followers
Companies to Watch
CW
CoreWeave
Neo Cloud · $19B · IPO Watch
CB
Cerebras Systems
AI Hardware · $4.25B · Pre-IPO
G42
G42
Sovereign AI · Abu Dhabi
H
Humain
Saudi AI · $40B Fund
Latest Podcast
AI Capex, Cloud Margins & the Nuclear Bet
48 MIN · 25 APR 2026

The Procurement Question Every Operator Should Ask Twice

A cooling system can look exceptionally disciplined while its water behavior remains almost impossible to predict from the procurement documents

Share
data center cooling procurement
0
847 SHARES

0
SHARES

[simple-author-box]

More from AI Infrastructure

Ireland’s experience with data-center expansion became less about stopping construction than about changing the

A data hall can meet its opening-day layout and still contain a future expansion

A data center budget can remain numerically intact while its economic position deteriorates around

A data center schedule can begin moving well before major site construction starts, because

COMPUTE WEEKLY

The briefing that 40,000+ tech leaders read every Monday. Sharp, fast, essential.

Great! We’ve received your information.

Global AI Infrastructure Outlook 2026

The briefing that 40,000+ tech leaders read every Monday. Sharp, fast, essential.
Download Free
Most Read

Ireland’s experience with data-center expansion became less about stopping construction than about changing the

A data hall can meet its opening-day layout and still contain a future expansion

A data center budget can remain numerically intact while its economic position deteriorates around

A data center schedule can begin moving well before major site construction starts, because

A compute node sitting behind a garage door can perform the same basic computational

Disruptor Spotlight

Cerebras Systems

The chip that makes Nvidia nervous. Cerebras’ Wafer Scale Engine is rewriting the rules of AI inference at scale.
Faster
0 x
YoY Revenue
0 x
Transistors
0 T
Market Pulse
NVDA
$924.60
+2.4%
MSFT
$421.30
+1.1%
AMZN
$192.80
-0.6%
NVDA
$924.60
+2.4%
NVDA
$924.60
+2.4%
Indicative only · Not financial advice
Upcoming Events
MAY
0 0
DCD Global — London
LONDON · IN PERSON
World’s largest DC event. CF is media partner.
MAY
0
AI Infrastructure Summit
DUBAI · IN PERSON
MEA’s premier AI infrastructure event.
JUN
0 0

Compute Forecast Summit

SINGAPORE · IN PERSON
Our flagship APAC event. Early bird open.
Latest Moves
  • Live
Sam Altman
OpenAI appoints new Chief Infrastructure Officer to lead $100B DC programme
27 APR · OPENAI
Sam Altman
OpenAI appoints new Chief Infrastructure Officer to lead $100B DC programme
27 APR · OPENAI
Sam Altman
OpenAI appoints new Chief Infrastructure Officer to lead $100B DC programme
27 APR · OPENAI
Follow Compute Forecast
18.4K followers
12.1K followers
9.3K subscribers
41 episodes
Companies to Watch
CW
CoreWeave
Neo Cloud · $19B · IPO Watch
CB
Cerebras Systems
AI Hardware · $4.25B · Pre-IPO
G42
G42
Sovereign AI · Abu Dhabi
CW
Humain
Saudi AI · $40B Fund
Latest Podcast
AI Capex, Cloud Margins & the Nuclear Bet
48 MIN · 25 APR 2026
Scroll to Top
Seraphinite AcceleratorOptimized by Seraphinite Accelerator
Turns on site high speed to be attractive for people and search engines.