Skip to content
synpute

For insurers & investors

Comparable data on compute risk.

Synpute Price Reference and the Synpute Loss Index turn registry data into comparable marks and loss history across operators, so you can hold compute exposure without underwriting each cluster.

Synpute Loss Index

Preview
Hardware: B200Region: USTenant: startup
Default rate
— —

% of clusters

Recovery rate
— —

% of exposure

Time to re-lease
— —

days

Replacement rate
— —

$/GPU-hr

Loss history · by quarterawaiting data

Layout only

v1

No loss data is published until the methodology and governance are.

The gap

A large market with no loss history.

Capital is arriving faster than the data. There is no public default or recovery history for compute, and price marks disagree. Each cluster is underwritten from scratch.

~$135B

Data center debt held by private credit funds, globally

Source: Datacentres.com, September 2026

None

Public default and recovery data on compute today

Methodology previews

Two data products, one registry.

Methodology and governance are published before any data. Here is what each product will measure.

Synpute Price Reference

Prices drawn from real contracts.

  • Transaction-verified marks by hardware generation, term and region
  • Drawn from contracts recorded in the registry, not from quotes
  • Methodology, governance and an oversight committee published before any mark, aligned with the IOSCO Principles for Financial Benchmarks
  • Built for collateral values and re-leasing floors

Synpute Loss Index

Public data on compute credit.

  • Default, recovery and time-to-re-lease across registered clusters
  • Cut by hardware, site and tenant type
  • Recorded as it happens on each Synpute ID, from Monitor and Transfer
  • Built for lenders, insurers, investors and rating agencies

Data governance

What's aggregated, and what never leaves.

Operators and buyers own their data. The data products publish aggregated, anonymized figures, and nothing that identifies a party.

Aggregated and anonymized

  • Contract rates, by hardware generation, term and region
  • Defaults, recoveries and time to re-lease
  • Replacement rates, by hardware, site and tenant type

Never published

  • Contract-level data, unless the owner authorizes it
  • Names of operators, buyers or lenders
  • Any data on individuals

Methodologies are published, versioned and reviewed by an advisory board of lenders, operators and buyers. A written conflict policy bars Synpute from trading capacity or holding positions in anything it marks.

Subscription tiers

Access for every kind of holder.

Indicative pricing, so you can plan. Tiers are set when the products launch.

Data subscription

$75K to $120K

per institution a year

Price Reference, Loss Index and registry access

For lenders, insurers, investors

Platform license

$150K

per platform a year

Price Reference with the Verify API and Synpute ID

For marketplaces, exchanges, brokers

Data license

By agreement

scoped to use

Loss Index and Synpute Standard data feeds

For rating agencies and analysts

Indicative prices, not quotes. No marks or loss data are published on this site; the products open to the waitlist first.

Data waitlist

Join the data waitlist.

Tell us which products you'd use. You'll hear first when methodology notes and data are released.

What happens next

  1. 1Methodology notes for Price Reference and the Loss Index
  2. 2Early access when the first data is released
  3. 3An invitation to review the methodology
  4. 4Business contact details only
Products of interest *

Business contact details only. See our privacy policy.

FAQ

Questions we hear

Straight answers to what comes up in the first call. If yours isn't here, ask us directly.

Talk to the team

06 questions · reply within one business day

Price compute risk on data, not guesswork.

Transaction-verified marks and public loss history, built from the registry and published with their methodology.