Last Update: March 2026
The Index for Compute Supply
The largest compute build-out in history is underway — but the supply side remains opaque. There is still no clear, consistent view of what is being built, what is actually progressing, what the grid can support, and where new compute capacity can realistically come online.
Operators publish incompatible metrics. Utilities are flooded with speculative requests. Investors, operators, and analysts are left stitching together filings, press releases, and fragmented datasets without a trustworthy baseline.
Clarke is building the index for compute supply: a structured, forward-looking view of compute infrastructure across the U.S. We track and forecast when, where, and how much capacity is coming online — turning fragmented physical infrastructure signals into supply-side intelligence for compute markets.
We map, classify, and forecast the infrastructure underlying frontier-scale AI — from facility-level capacity to regional power constraints and operator build-out trajectories — using transparent methodology and auditable public sources.
01 — Intelligence Layer
What We're Building
The Clarke Intelligence Layer
Compute infrastructure is scaling faster than the frameworks used to understand it. Clarke is building the intelligence layer that makes the physical compute landscape legible, comparable, and decision-ready.
Capacity Forecasts (Core)
Where compute capacity can realistically grow through 2030 — shaped by power availability, siting constraints, regional delivery dynamics, and operator build-out trajectories. Confidence-scored. Primary-sourced. Built on transparent methodology.
The Clarke Index
A structured, evolving index of U.S. compute infrastructure built from primary documentation: utility filings, regulatory dockets, operator disclosures, and other auditable public records. Not directories. Not recycled marketing claims.
Facility Intelligence
Consistent classification and verified signals around capacity, cluster presence, cooling systems, density, development timelines, and delivery status.
Regional Power Dynamics
A coherent view of how utility constraints, transmission realities, interconnection bottlenecks, and planned infrastructure shape where frontier-scale compute can actually grow.
Clarke consolidates the signals that matter — capacity, timelines, power constraints, and feasibility — into a single interpretive layer. Inputs are tied back to public, auditable sources, including operator disclosures, utility and regulatory filings, and other verifiable documentation.
02 — Market Context
Why This Matters
The market is scaling faster than the language describing it.
Today, the physical supply side of AI is described through inconsistent terms and incomplete signals. Capacity metrics are not standardized. Power metrics vary by operator. Utility demand signals are noisy and distorted by speculative requests. Regional comparisons are often built on incompatible assumptions.
Without shared definitions, transparent methodology, and evidence-backed forecasting, the market lacks a reliable view of compute supply.
Clarke provides the interpretive layer this market needs — built on consistent classification, transparent standards, and supply-side forecasting grounded in public evidence.
03 — Standards
Built on Clarke Standards
The Foundation for Compute Infrastructure Intelligence
Clarke Standards provide an open, versioned vocabulary for compute infrastructure.
They define what constitutes an AI-oriented facility, distinguish AI-Native from AI-Capable, and anchor consistent interpretations of capacity, power, cooling classes, density, cluster presence, and related infrastructure signals.
Built from systematic analysis of operator disclosures, utility filings, and regulatory documentation across the U.S., Clarke Standards provide the common baseline that makes the Clarke intelligence layer possible.
Clarke Standards are open, versioned, and neutral. Definitions may be reused with attribution.
04 — Access & Pricing
For Enterprise
Launching H1 2026
Decision-grade intelligence for the compute economy.
Capacity forecasts, facility profiles, and regional power dynamics — verification-tiered, confidence-scored, and built for long-horizon capital, infrastructure, and market decisions.
05 — About Clarke
About Clarke
Clarke is the supply-side intelligence layer for compute infrastructure — a coherent, forward-looking view of the physical capacity powering frontier-scale AI.
Founded by Chris D'Andrea, Clarke is built on a simple premise: the AI economy needs a trustworthy view of when, where, and how much compute capacity is actually coming online. The company approaches compute infrastructure as a systems problem — building standards, forecasts, and intelligence from first principles rather than inherited assumptions.
We're selectively engaging early partners and investors aligned with the long-term vision for compute-market intelligence.
Based in New York. Enterprise platform launching H1 2026.