The Evaluation Checklist

Everything in this course reduces to a single practical output: a checklist you can run against any neocloud to separate the durable from the doomed. The checklist is organized around the few variables that actually decide survival. Work through it in order, because the early items gate the later ones.

Start with the anchor and the demand book. Who is the largest customer, and what share of revenue do they represent? Is that share falling over time as independent customers are added, or is it holding lopsided as the company grows? What form do the contracts take, take-or-pay or best-efforts, long or short, prepaid or unsecured? Weight the backlog by counterparty creditworthiness, not headline size. A book dominated by long, take-or-pay contracts from strong, independent buyers is the single best sign a neocloud will last. A book that is one entangled name in a trench coat is the single best sign it will not.

Next, examine the debt against the chips. Over what period does the financing amortize, and how does that compare with the realistic competitive life of the accelerators? If the debt outlives the chips earning power, the company is on a clock. What depreciation schedule does it use, and is it stretched to flatter margins? What is the cost of the debt, when was it struck, and when does it mature? A near-term maturity in an uncertain credit market is a hidden fragility. Match the tenor of the debt to the tenor of the contracts and the life of the chips; where those three fall out of alignment, you have found the risk.

Then interrogate power. Is electricity secured under long-term contracts at a known price, or bought opportunistically? How strong is the interconnection position, and how quickly can new capacity actually be energized? Power is the largest recurring cost and often the binding constraint on growth. An operator that solved power early has a durable advantage; one that left it to chance has left its margins and its expansion exposed to the most volatile input in the business.

Now assess chip access and supplier dependence. Does the operator have preferred allocation, and how much of its growth depends on that favor continuing? Is it building alternative supply relationships and qualifying other accelerators, or is it wholly captive to one vendor? Preferred allocation is an advantage today and a dependency tomorrow. The durable operator treats it as the former while preparing for the latter.

Trace the circularity. Is the largest customer also an investor, or the chip vendor, or otherwise financially entangled with the company survival? How much of the revenue would exist if every counterparty acted purely at arms length? Follow each major dollar of demand to its ultimate source and ask whether someone outside the loop is paying for intelligence they value. Revenue that survives that trace is real; revenue that depends on the capital continuing to circulate is borrowed time.

Check utilization and operational quality. Is the installed fleet running hot, with high sustained utilization across reserved and on-demand capacity? Idle chips depreciate and owe interest while earning nothing, so utilization is the truest measure of whether the fleet pays its own way. Consistently high utilization signals both genuine demand and competent scheduling; chronically low utilization signals a fleet bought ahead of the demand to fill it.

Finally, test the growth discipline. Does fleet expansion follow signed contracts, or lead them on faith? Does the operator secure power and demand before it commits capital, or build speculatively into hoped-for demand? Speculative building turns a demand wobble into a solvency crisis; disciplined building carries a company through the first air pocket. The pace of growth matters less than whether each increment is underwritten by something contracted and certain.

Run the whole checklist and a pattern resolves. The durable neocloud scores well on the same axis at every question: its commitments are matched to its certainties. Contracted, diversified, arms-length demand. Debt sized to the true life of the chips. Power locked in. Supply diversified. Growth that follows proof rather than hope. No single item guarantees survival, but a company that clears all of them has no structural reason to fail, and a company that fails several has no structural reason to survive, the only question is timing.

Use the checklist as a filter, not an oracle. It will not tell you which neocloud will have the best quarter, and it is not designed to. It is designed to tell you which ones are built to still be standing after the first serious stress, the demand pause, the rate rise, the early chip generation, the anchor that walks. In a category defined by companies that appear fast and fail fast, the ability to sort the two apart before the market does is the entire edge. That sorting is what this checklist gives you, and it is the one durable skill to carry out of this course.