WillBuild
Mission-critical AI infrastructure · Northeast USA

Most data centers are built for average workloads. I engineer for the extreme.

I design, build, and assure the thermal and power environments that keep high-density AI and low-latency workloads running without throttling or failure — surgical, telemetry-backed retrofits for clients who treat latency and downtime as existential threats.

Independent & vendor-neutral · Northeast USA

20+
Years in mission-critical mechanical
GW-scale
Capacity enabled in North America
130 kW
Per-rack densities engineered
Zero
Thermal-throttling events on assured pods
Workload-to-Watts

Don't take my word for it. Take your own numbers.

Generalists ask what power you need. I tell you — from your hardware. Enter your deployment and see exactly where your current cooling throttles it.

Run the math

Your numbers, a deterministic answer. No email required.

47 kW
Per-rack load
0.47 MW
Total IT load
Throttles at ~32% of this workload

Your cooling tops out at 15 kW/rack against a 47 kW load — a 32 kW/rack deficit (0.32 MW across 10 racks).

Recommended approach
Rear-door heat exchangers / hybrid

Air alone is marginal — hybrid liquid assist recommended.

Get the full CFD model for your room

Directional estimate for orientation only — not a stamped engineering calculation. A real assessment runs CFD against a digital twin of your actual room.

What I do

De-risk it. Build it. Stand behind it.

Three phases, one accountable engineer — from the upfront thermal reality, to a commissioned pod, to a telemetry-backed guarantee on the environment it runs in.

Thermal & Power Audit

Before any capital is committed, I show you the exact thermal and power reality of your space — with the data to prove it.

Learn more

Surgical Design-Build Retrofit

Turnkey high-density retrofits for power-constrained, occupied footprints — delivered in weeks, not quarters.

Learn more

Performance Assurance

I don't hand over the keys and leave. I guarantee the thermal and power envelope your workload runs in — backed by live telemetry.

Learn more
Why WillBuild

I know the detail. That's the whole job.

At 100 kW+ per rack, the margin for error disappears. The difference between a facility that scales and one that melts down lives in the details most firms never look at — the manifold, the busbar, the displacement when a door opens. That's where I work.

  • Independent and vendor-neutral — I specify what's right, not what I resell
  • The data first: CFD and telemetry before opinions or capital
  • Surgical retrofits in occupied, power-constrained footprints
  • A bounded, telemetry-backed guarantee on the environment
How the work runs
  1. 01

    Audit

    I walk your facility and pull real telemetry — the honest starting point, not a rule of thumb.

  2. 02

    Model

    CFD and a digital twin prove the thermal and power outcome before any capital is committed.

  3. 03

    Build

    Prefabricated, pre-tested systems are dropped into your live room with minimal disruption.

  4. 04

    Assure

    Live telemetry underwrites the SLA and tunes the envelope for the life of the deployment.

Selected work

Where the details won

View all work
High-Frequency Trading

Beating a 15 kW/rack ceiling

Secaucus, NJ

Bypassed the building loop with modular CDUs to land 90 kW/rack and eliminate market-open thermal throttling. No interruption to live trading.

Generative AI

6 MW pod inside a capped envelope

Manhattan-adjacent

Densified within the existing fixed power envelope — more usable compute landed against the same MW, no new interconnection required.

Low-Latency Platform

Sub-microsecond thermal stability

Fairfield County, CT

Hybrid air-liquid balancing and predictive telemetry held inlet temperatures flat through peak — spikes eliminated.

What clients say

Trusted where failure isn't an option

He told us exactly where we'd throttle — to the rack — before we spent a dollar. The CFD model was dead on. Nobody else we spoke to could show us the math.
Head of Infrastructure
Quantitative trading firm
We couldn't get more power from the utility for years. Will got us the compute we needed inside the envelope we already had. That alone saved the roadmap.
CTO
Generative AI company
The telemetry caught a pump going bad two weeks before it would have taken down a pod. The assurance model is the real product.
VP, Platform Engineering
Low-latency data platform

Illustrative client outcomes — references available under NDA.

Running out of thermal headroom?

Send me your facility and your workload. I'll run the model and show you the honest ceiling — before you commit a dollar.

Request a thermal audit