MacroTerrain

MacroTerrain research brief

How Should Investors Compare Contracted and Available AI Compute?

Contracts, energization, hardware deployment and customer acceptance are separate gates.

Executive answer

Contracted compute represents a customer obligation or reservation; available compute represents installed and operable infrastructure. Revenue may require energization, commissioning, hardware installation and customer acceptance after the contract is signed. Comparisons should keep MW, GPU counts, service revenue and lease commitments in their native units.

Key findings

  • A take-or-pay contract can improve financing visibility while delivery risk remains.
  • Available capacity may be uncontracted; contracted capacity may be undelivered.
  • Customer concentration and contract maturity can affect refinancing and residual-value risk.

Definitions

Contracted
Covered by an executed agreement, subject to conditions.
Available
Technically ready for service.
Accepted
Confirmed under contractual testing or delivery terms.

Evidence and methodology

IREN’s annual report explains that revenue generally starts after delivery, commissioning, service placement and acceptance. CoreWeave’s presentation describes financing supported by take-or-pay contracts and related infrastructure.

Claims are classified as documented facts, calculations, inferences or MacroTerrain interpretations. The brief does not promote an inference to fact when the documentary path ends.

Source-backed example

A developer can report a signed 100 MW lease while only the first 25 MW phase is energized. Both are useful facts, but the 100 MW should not be displayed as current operating capacity.

Relevant MacroTerrain visualization
Contract-to-revenue bridgesigned capacityacceptance

Contract-to-revenue bridge from signed capacity through acceptance. The illustration explains the analytical structure; it is not a quantitative model output.

Limitations and unresolved facts

Public disclosures may not identify contract ramp, utilization or acceptance status. GPU and MW measures cannot be converted reliably without hardware and facility assumptions.

What investors should monitor

  • delivery and acceptance milestones
  • hardware procurement and deployment
  • customer ramp and renewal terms
  • energy availability and curtailment

Sources