Independent research on compute-backed credit

Independent credit analysis for the AI build-out.

For lenders and investors assessing GPU debt, data-centre finance and compute contracts: we connect cash flow, hardware value and the debt schedule to show which assumptions drive the credit.

When protections weaken together

Cash flow, collateral, parent support : Which protection will you need when the others weaken?

Compute collateral does not stand still

% of original principal

Fine lines: possible collateral paths. Bold: selected band. Dashed: scheduled debt. Click a path to follow it.

Compute-price environment 50th percentile band

WeakerStronger

What if receipts stop?

A customer dispute, counterparty distress or contractual rights could interrupt receipts. Here, test a gap beginning in year two.

What else weakens? No parent supportRemoves a source of payments. Lower collateral recoveryReduces the fallback value by 30%.

Choose an interruption to test the other protections.

Who pays the lender?

Monthly payments / % of original principal

Operating cashReserve drawParent paymentUnpaid service

What backs the remaining debt?

Fallback cover / % of original principal

Merchant collateralCash reserveParent-dependentWithout fallback coverCumulative write-off

The protection that looks strongest today may not be the one you need tomorrow.

Discuss your exposure with us Illustrative CCRFramework simulation, not a named deal or forecast. Red areas show unpaid service or debt without fallback cover; the dashed red line shows cumulative write-offs.

Our coverage

The assets, contracts and borrowers behind compute debt

GPU collateral and equipment finance

What would the lender recover if the contract stopped paying? We assess hardware resale assumptions and the security terms that determine how much of that value reaches the lender.

Data-centre structured finance

We connect asset cash flows with the structure around them: lease terms, debt-service coverage, maturity and the capacity to refinance.

Neocloud and operator credit

Neoclouds, converted miners and integrated sponsors. We assess who owes the revenue, when it can stop, and how much cash remains after operating costs and debt obligations.

Meet the analyst

Two disciplines behind one credit view

The useful answer sits between the indenture and the machine: How the structure pays, what the contracts really guarantee, and what the hardware might be worth when the base case breaks.

Martin Andrews combines CDO structured-credit experience from 2001–2011 with a PhD and work in deep learning since 2012. That background informs an analysis of how compute assets earn, what they may recover and how the financing allocates the risk. Notably, Martin's CDO experience involved pricing structures based on underlying collateral prices - enabling clients to avoid the worst of the rating agency methodology fall-out.

Publications

Planned publications

Planned titles are provisional. Publication dates have not been set.

Publications and provisional titles
Date	Title	Category
—	How independent are the protections in a GPU-backed loan?PROVISIONAL	Silicon Duration
—	Deal note: a neocloud facilityPROVISIONAL	Subscriber research
—	The economic half-life of a GPUPROVISIONAL	Silicon Duration
—	Deal note: a rated data-centre ABSPROVISIONAL	Subscriber research

Access

Getting the work

Forthcoming in Silicon Duration

How independent are the protections in a GPU-backed loan? Provisional title; the commentary is being prepared.

Sample note forthcoming

Institutional subscription

Research on individual debt instruments, released to subscribers on completion, plus the full archive and direct analyst access. Priced per seat and per desk.

Enquire about institutional access