Erik Torenberg

Andy Jassy × Ben Horowitz × Erik Torenberg

Amazon expects about $220 billion of 2026 CapEx, yet AWS capacity is sold out for 2026, largely for 2027, and partly for 2028; AWS AI has exceeded a $25 billion revenue run rate.Data centers require investment roughly two years before monetization, pressuring free cash flow; broader enterprise inference adoption also hinges on costs Jassy still calls an order of magnitude too high.

Erik Torenberg × Nathan Labenz × Anna Patterson × Lukas Petersson × Zvi Mowshowitz × Naveen Verma

Ceramic AI offers $0.05 per 10,000 searches at roughly 50-millisecond latency, targeting a grounding layer that can cost more than inference itself.GPT-5.5 matched Opus 4.6 on single-agent Vending-Bench and beat Opus 4.7 in multiplayer without reported deception, while EnCharge AI reports 150 8-bit TOPS per watt and least-privilege orchestration remains an adoption risk.

Erik Torenberg × Nathan Labenz × Alex Boris × Dean Ball × Peter Wildeford

H100 clusters, billion-dollar raises, and enormous pre-training costs are turning frontier-model competition into a capital-and-infrastructure game, while fine-tuning and inference remain accessible to a much broader application economy.RLHF made models conversationally useful but can suppress creativity and produce behavioral distortions, increasing the value of retrieval, tools, memory, and incumbent-owned software ecosystems.Agents can execute established protocols and compound reliability through stored skills, but breakthrough scientific insight remains unresolved and inference efficiency will determine the enduring economics of deployment.

Dylan Patel × Erin Price-Wright × Guido Appenzeller × Erik Torenberg

GPT-5’s router is an economic release, directing simple queries to mini models while reserving “ungodly amounts of compute” for transactions OpenAI could monetize through agentic commerce.Flat-rate subscriptions face heavy-user losses, while custom silicon threatens Nvidia mainly if demand stays concentrated among hyperscalers; powered sites, grid equipment, and Intel’s capital needs remain near-term constraints on broader deployment.

Vincent Weisser × Johannes Hagemann × Erik Torenberg × Nathan Labenz

Prime Intellect is building an asset-light marketplace across fragmented clouds and data centers, while Intellect-1 demonstrated 10-billion-parameter distributed training using DiLoCo and 8-bit updates to cut communication roughly 400x.R1-style reinforcement learning could improve decentralized scaling by shifting compute toward inference-heavy rollouts, but efficient participation beyond about 16 workers, fault tolerance, safety, governance, and the planned tokenized foundation remain unresolved.