Chamath Palihapitiya
· Chamath Palihapitiya
· 14 августа 2026, 15:46
· ⏱ 4 мин чтения
| Читать в Substack ↗
Резюме
AI compute has become too large and price-volatile to keep trading bilaterally: 2026 AI capex of $765B overtook oil & gas capex, and CME/Silicon Data are launching the first compute futures contracts to hedge that exposure. Success depends on solving GPU concentration and interchangeability problems; if futures scale, compute becomes a tradeable asset class and materially de-risks AI infrastructure financing. For markets, the clearest near-term beneficiaries are the exchange/index infrastructure and the dominant compute hardware supplier.
•AI capex hit $765B in 2026, surpassing oil & gas capex at $681B for the first time, and is projected to nearly double by 2031.
•CME Group and Silicon Data announced plans to launch compute futures contracts on October 5, 2026, pending regulatory review.
•Silicon Data ran the same workload across 3,500 GPUs at 11 cloud providers and found H100 performance varied by as much as 34.5% in one test, with the widest spread reaching 38%.
•Neocloud revenue passed $25B in 2025 across more than 60 providers, showing broad and growing compute supply.
•Prior attempts to build futures markets for onions, uranium, DRAM memory chips, and bandwidth failed due to concentration and interchangeability problems.
•Larry Fink argues a new asset class will buy compute futures because there is not enough compute power, and Morgan Stanley projects AI diffusion creates a $40T opportunity.
Article reports CME Group and Silicon Data plan to launch compute futures on October 5, 2026 pending regulatory review; if the contract scales, CME captures a new derivatives revenue stream and become
Article reports CME Group and Silicon Data plan to launch compute futures on October 5, 2026 pending regulatory review; if the contract scales, CME captures a new derivatives revenue stream and becomes the venue for hedging AI's largest input cost.
Risk: Regulatory review, low initial volume, and GPU interchangeability problems could prevent the contract from achieving meaningful liquidity.
Article states NVIDIA supplies most AI chips and that compute demand has risen sharply; compute futures would hedge GPU rental volatility, making it easier for buyers to commit to capacity and preserv
Article states NVIDIA supplies most AI chips and that compute demand has risen sharply; compute futures would hedge GPU rental volatility, making it easier for buyers to commit to capacity and preserving Nvidia's role as the core supplier of AI infrastructure.
Risk: Article also notes that each new Nvidia generation destroys prior-GPU rental and collateral value, a dynamic that could pressure financing economics for older chips and create product-cycle risk.
This newsletter, published August 14, 2026,
features Chamath Palihapitiya
discussing CME, NVDA.
2 trade ideas extracted by AI with direction and confidence scoring.