# Compute Reservation **Entity class:** Infrastructure contracting mechanism **Domain:** Artificial intelligence / capacity markets **Maturity:** Developed ## Definition Compute reservation is the advance commitment of accelerator, memory, energy, or data-center capacity for future model training and inference. ## Mechanism and significance Long reservation horizons can secure a laboratory’s development path while preventing later entrants from buying equivalent capacity on demand. Competitive position therefore depends on infrastructure contracted before a model or product is ready. ## Relationships - [[wiki/Compute Scarcity|Compute Scarcity]] - [[wiki/Compute Allocation|Compute Allocation]] - [[wiki/Frontier-Lab Vertical Integration|Frontier-Lab Vertical Integration]] - [[wiki/Moonshots - The Coming Manhattan Project|Moonshots — The Coming Manhattan Project]] ## Sources and provenance - [[research/All Things Superintelligence Research - The Coming Manhattan Project - Moonshots|All Things Superintelligence Research — The Coming Manhattan Project — Moonshots]] — episode transcript, reconciled against the preserved ElevenLabs timing layer. - [[research/All Things Superintelligence Research - The Coming Manhattan Project - Moonshots - Reminder and Wiki Preparation|Reminder and Wiki Preparation]] — coherent-thought ledger and evidence boundaries. ## Evidence boundary Claims about how many years of capacity are reserved are time-sensitive and source-dependent. This node distinguishes the mechanism from any unverified numerical estimate. ## Simple Reminders, Quotations, and Thoughts > Frontier AI laboratories may withhold their best models and use them to develop and sell drugs themselves. Incumbent companies that postpone building or securing their own AI capacity may discover that the compute needed to recover their competitive position has already been reserved. > **— Adapted from Dave Blundin**, *Moonshots, October 7, 2026* [[reminders/Biotechnology/Frontier AI Labs May Become Drug Companies Instead of Suppliers by Dave Blundin|Frontier AI Labs May Become Drug Companies Instead of Suppliers by Dave Blundin]] > Executives are still learning how to absorb today’s models while laboratories prepare much stronger ones and the compute market tightens behind them. Organizational adoption can lag twice: first behind capability, and then behind the infrastructure required to catch up. > **— Adapted from Emad Mostaque**, *Moonshots, October 7, 2026* [[reminders/Economic Transition/Executives Are Adapting to Models the Frontier Has Already Left Behind by Emad Mostaque|Executives Are Adapting to Models the Frontier Has Already Left Behind by Emad Mostaque]]