Extropic uses RL post-training to accelerate thermodynamic algorithm discovery, first sparks of Thermo RSI
beffjezos · x · 2026-10-02
Extropic, with Prime Intellect, published "First Sparks of Thermodynamic Recursive Intelligence": they use RL to post-train models that accelerate thermodynamic algorithm discovery for their new hardware substrate.
- Modern deep learning co-evolved with matrix-multiplication accelerators (GPUs/TPUs); thermodynamic computing is a different "species" of hardware (claimed 100x–10,000x algorithmic efficiency) and needs its own algorithm family.
- They've open-sourced a thermodynamic programming stack: Torx (high-level programming), Thermalizers (compiling stochastic differentiable programs), and THRML (low-level sampling framework).
- Framed as the first step of hardware-algorithm co-design toward porting AI inference to thermodynamic hardware.
More from Infra
- Micron Q4 FY26: bigger bit base and more HBM support a longer memory cycle — BenBajarin · 2026-10-02
- AWS revamps onboarding: no credit card, $200 free credits, hard spend caps, one-prompt agent setup — danbri · 2026-10-02
- Lambda wins ClusterMAX Sliver as SemiAnalysis finds reliability much improved — TheZachMueller · 2026-10-02
- Micron turns green as AI compute selloff appears to cool — firstadopter · 2026-10-02
- Edge0 open-sources app layer: 35B model on-device with just 1–2.5GB memory — FinanceYF5 · 2026-10-02
- SpaceX Transporter-18 launches 130 payloads including Google's orbital AI chips — XFreeze · 2026-10-02