AI compute research and GPU market publications

Sovereign compute

On Sovereign Compute

Advanced AI chips have crossed from commercial input to sovereign commodity. Export licensing, rare-earth retaliation, national subsidy races, and a wave of AI-sovereignty pledges now treat GPU capacity the way states once treated oil, rare earths, and LNG: once a resource is contested, governments secure access and permanently change how the market clears. Fabrication, high-bandwidth memory, and advanced packaging sit in a handful of jurisdictions and firms, so a single policy or geographic shock can reprice fleets in days. The essay’s thesis is that compute is already on that historical arc, yet it still lacks the benchmarks and hedging tools those earlier commodities eventually required. The practical market question is how operators, buyers, and lenders should budget and finance GPU capacity when jurisdiction—not just hardware generation—can move the price overnight.

Procurement

The Basis Gap in Compute Acquisition Structures

GPU procurement is not a search for the cheapest hour. Buying, leasing, and reserving capacity each package three exposures differently: replacement-price risk, unused-capacity risk, and residual value. The essay’s thesis is that the hidden variable is basis—the gap between what a contract promises and what a workload later needs. A cheap commitment without reserved machines can be right on rate and wrong on access; a reservation without a matching spec can strand capacity while the buyer still pays for a second, better-fit block. At thousands of GPUs those mismatches become project-finance problems, because collateral values and contracted cash flows move on different clocks. The market question is which risks a fleet owner should keep, and which to push to a lessor or lender, when hardware, utilization, or demand diverge from the term sheet.

Compute markets

Compute in Space. Why, How, and When?

Ground datacenters cannot expand as fast as AI demand because power interconnects, water for cooling, and suitable land all take years. The essay argues that orbit is becoming a real alternative, not a slogan: launch costs have fallen sharply, solar collection is far stronger above the atmosphere, and radiative cooling can reject heat without consuming rivers. Early orbital H100 hardware and reusable-rocket economics are the proof points. Radiation, servicing, and extra latency still decide whether space is a hedge or a science project, especially for delay-sensitive inference. The practical question for buyers and financiers is when orbital capacity should enter a terrestrial fleet’s cost and obsolescence model—and how to hedge the chance that ground hardware is stranded if compute migrates off-planet.

GPU economics

On GPU Depreciation

Investor alarm that five-year GPU lives overstate earnings treats an accounting schedule as if it measured physical wear. The essay’s thesis is that this conflates two different things. GAAP depreciation allocates capex, moves reported earnings, and creates a tax shield. Economic loss of GPU value is driven mainly by new architectures and training regimes, not by parts failure. Older accelerators can remain cheap for batched, throughput-heavy work long after they look obsolete on a latency benchmark, and hyperscaler retirement histories imply useful lives closer to seven to nine years. The market question for operators and lenders is how to report and finance useful life without using a depreciation table as a residual-value forecast—and whether economic obsolescence should be hedged rather than modeled as linear wear.

Memory

Memory: How It Works and Why It Sets the Cost of Compute

Frontier models wait on memory more than they wait on arithmetic. Prefill and decode keep weights and a growing key-value cache in high-bandwidth memory, so HBM capacity and bandwidth set how much context, occupancy, and therefore usable compute a GPU can deliver. The essay’s thesis is that memory, not transistors, is now the price-setting input. HBM cannot expand smoothly: stacking yield, CoWoS packaging, and a three-vendor allocation market reprice at contract resets rather than on a continuous spot curve. Because memory already dominates the bill of materials on leading accelerators, a discrete HBM step change reprices whole clusters at once. The practical question is whether buyers and lenders can treat HBM as a scarce, hedgeable commodity instead of a background component cost.

Compute markets

The AI Chip Wars: From Supplier Dominance to Contested Equilibrium

Headlines about OpenAI training on Amazon Trainium, or NVIDIA licensing Groq inference designs, invite a winner-take-all reading of the chip race. The essay’s thesis is narrower: large buyers are building credible outside options so they can contest pricing power even if NVIDIA remains the performance leader. Hyperscaler silicon does not need to win every workload; it needs to be viable enough to change contract terms, optionality, and where margins sit in the stack. Competition will appear first in cloud pricing and flexibility, not in merchant-GPU share. The market question is how to budget and hedge GPU-hours when the price of compute fragments across architectures, regions, and bargaining positions rather than remaining a single supplier quote.

Infrastructure finance

Circular Financing in AI

Cross-investments, take-or-pay contracts, prepayments, and utilization guarantees among chip vendors, labs, clouds, and builders look like a closed money loop. The essay argues they are a familiar infrastructure workaround: they pull future demand into the present so physical capacity can be financed before transparent markets exist. That stopgap concentrates utilization, construction, funding, and obsolescence risk on a small set of counterparties and hides true prices. The structure works only while demand assumptions hold. Markets mature when contracts standardize, benchmarks appear, and risks can be unbundled. The practical question for lenders and operators is which balance sheet absorbs the shock if spending, power timelines, or GPU generations break those loops.

Derivatives

Compute Futures

Capital is pouring into AI infrastructure while the value of a GPU-hour remains hard to trade. The essay’s thesis is that a cash-settled future, marked to a daily compute index and paying an Asian-style average over the tenor, matches how compute is actually used: as a flow, not a single expiry print. Incremental daily settlement plus margin lets consumers, producers, and financiers hedge without taking delivery or operational SLA risk. Oil-style terminal settlement would misalign the hedge with the period of use. The market question is whether operators and investors can lock in a period-average GPU-hour cost—and transfer counterparty risk through collateral—instead of managing exposure with opaque bilateral contracts.

Derivatives

Compute as a commodity isn't oil. It's electricity.

The first hedge design for compute copied oil: lock a future price for a storable barrel. The essay’s thesis is that this analogy fails at the unit of account. Compute is quoted per GPU-hour, and unused capacity cannot be recovered, so it is a flow good. Electricity is the better analog: an unused megawatt-hour vanishes, and power markets already price location, real-time utilization, and forward delivery around that temporal structure. Compute markets are beginning to show the same features. The practical question for anyone hedging GPU cost or revenue is whether the instrument should settle like a stock commodity at expiry or like power over the period of use.