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.
Who writes this
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.
Reports assume you know credit and explain the compute, or the reverse, depending on which side of the desk you sit. Where a claim rests on a modelled number, the model is stated. Where it rests on a mark, the mark is sourced.
How the work is done
We read the documents. Offering circulars, credit agreements, offtake contracts, and the residual assumptions underneath the advance rate. We rebuild the collateral curve from secondary transactions rather than from vendor schedules, and we publish both so the gap is visible.
Every figure is either sourced or explicitly marked illustrative. Where we are uncertain, the note says so and says what would resolve it.
What this firm does not do
We do not trade the securities we cover. We do not accept issuer or sponsor payment for coverage, and no note is shown to a covered party before publication. We do not arrange, place, or advise on the transactions we write about.
Commissioned work is considered case by case and is disclosed as commissioned wherever it appears.
Enquire about institutional access