{"city":"Sunnyvale","name":"Cerebras","slug":"cerebras","facts":[{"date":"2026-10-01","type":"revenue_band","value":">500m","source":"reported"},{"date":"2026-10-01","type":"partnership","value":"Cerebras and AMD are partnering on disaggregated inference in production","source":"reported"},{"date":"2026-09-18","type":"partnership","value":"OpenCall switched from Cerebras custom chips to Inception diffusion-based LLMs for voice-agent speed and cost efficiency","source":"No Priors: AI, Machine Learning, Tech, & Startups — 2026-09-18"},{"date":"2026-09-12","type":"partnership","value":"Cerebras, SambaNova, and d-Matrix are developing competing disaggregated AI-compute systems","source":"The Register"},{"date":"2026-09-09","type":"product_launch","value":"Cerebras CS-4 launched with double the tokens-per-user performance of CS-3","source":"Semiconductor Engineering"},{"date":"2026-08-27","type":"deployment","value":"Cerebras CS-4 serving GPT-5.6 Sol at up to 750 tokens per second","source":"Cerebras"}],"peers":[],"stage":null,"state":"California","aliases":[],"country":"United States","domains":[{"why":"Cerebras designs and sells wafer-scale AI accelerators for training and inference to AI labs, cloud providers, and enterprise customers.","list":"AI Infra 100","rank":92,"tier":"member","layer":"Chips & accelerators","total":85,"domain":"AI Core","scores":{"gate_ne":["technology","firepower"],"leverage":10,"position":6,"progress":10,"firepower":8,"commercial":8,"deployment":9,"production":10,"technology":8},"edition":"November 2026","segment":"Inference-specialist chips"}],"founded":null,"signals":[{"date":"2026-10-29","header":"Culper Research alleged that Cerebras chips have shorter lifespans than the company’s statements suggest.","outlet":"Data Center Dynamics"},{"date":"2026-10-13","header":"CASPA’s 2026 CEO Summit will feature more than 35 Silicon Valley CEOs.","outlet":"EE Times"},{"date":"2026-10-07","header":"Cerebras enables rapid AI-assisted video-game and software decompilation.","outlet":"Peter H. Diamandis"},{"date":"2026-10-07","header":"Cerebras’s Astra model delivers 1,200 tokens per second with restricted access.","outlet":"Peter H. Diamandis"},{"date":"2026-10-07","header":"OpenAI did not make Ultra Ultra Fast publicly available because its Cerebras capacity was reportedly sold out.","outlet":"ARK Invest"},{"date":"2026-10-07","header":"OpenAI's Ultra Fast service is powered by Nvidia hardware rather than Cerebras chips.","outlet":"ARK Invest"}],"website":"cerebras.ai","category":"AI Infra · Chips & accelerators","categories":[{"name":"Accelerator design","domain":"AI Core","parent":"AI Infra · Chips & accelerators"}],"last_round":null,"updated_at":"2026-10-30T10:01:07.526+00:00","facts_total":30,"positioning":"The wafer-scale inference specialist that IPO'd in Q2 2026; sells fast-inference capacity to OpenAI and through AWS rather than chips alone, and is pairing with AMD and Trainium for disaggregated inference.","sub_category":"Accelerator design","signals_total":86,"url":"https://www.frontiertechintelligence.com/companies/cerebras","title":"Cerebras · ai core, Sunnyvale, California, United States · scores, facts and signals · Frontier Tech Intelligence","description":"Cerebras designs and sells wafer-scale AI accelerators for training and inference to AI labs, cloud providers, and enterprise customers. AI Infra 100 rank 92 (score 85) 86 signals and 30 facts with sources.","publisher":"Frontier Tech Intelligence (Future Frontiers, a trade name of Essential Cognition Inc.)","note":"Assessment by Frontier Tech Intelligence; facts and signals carry their sources."}