← Napkin Math, Cost & Capacity / 16

mediumCrusoeCoreWeaveMicrosoft

What is the power budget for a cluster of 1,024 H100s?

Build a power budget from complete servers to the facility meter, then check rack limits and distinguish a design maximum from an annual electricity bill.

Updated Oct 2026 · Learn how AI infrastructure works through explanations, worked examples and diagrams, then practise applying it to interview questions.

Build a power budget from complete servers to the facility meter, then check rack limits and distinguish a design maximum from an annual electricity bill.

This answer is included with a free account. Sign in to read it and save your progress.

free account · no card needed

Work through the numbers

Change the assumptions in a free calculator, then explain what the result establishes and what still needs measurement.

The concepts behind this question

Use these concept explanations to review the ideas behind this question.

🧮 Napkin Math & Capacity🔒 Premium

Power and Datacenter ConstraintsGPU capacity also needs electrical and cooling capacity. Build the facility estimate from complete system loads, rack density and the site’s operating limits.

🖧 Hardware & Cluster Build-Out

Colocation, Power Contracts and Site SelectionA suitable site must deliver the required rack power, cooling and service capacity. Contract terms and facility lead times shape both cost and deployment dates.

🖧 Hardware & Cluster Build-Out

Direct-to-Chip Liquid Cooling and CDUsDirect-to-chip cooling carries heat from cold plates to a coolant distribution unit. Flow, temperature, pressure and coolant quality all matter to the design.

🖧 Hardware & Cluster Build-Out

Rack Power Delivery and BuswaysRack power determines current, distribution equipment and redundancy requirements. Training workloads can also create load swings that average power hides.

Explore the full curriculum →

UP NEXT ON YOUR JOURNEY

Next in this trackShould we buy or rent 1,000 H100s for the next two years?Next in this trackHow much does moving from bf16 to fp8 save in serving cost?Next in this trackHow long does an all-reduce of a 70B model's gradients take on eight GPUs?

FEDITOR'S NOTE

Check system boundaries, explicit overhead assumptions, rack power and space, and the difference between average and maximum demand.

DISCUSSION · 0

No comments yet — be the first to share your approach.