{"changes":[],"country":"US","description":"Etched builds Sohu, an ASIC that hardwires the transformer architecture into silicon to run AI models far faster and cheaper than general-purpose GPUs. It targets AI labs and infrastructure providers running large transformer models at scale. By specializing the chip for one architecture, Etched claims large gains in inference per dollar.","domain":"etched.ai","employees":null,"founded":null,"fundingTotalQuarantined":false,"githubOrg":null,"githubStars":null,"industry":"semiconductors|ai|hardware","investors":[],"lat":37.787936,"lng":-122.40752,"location":"San Francisco CA","name":"Etched","rounds":[{"amountQuarantined":false,"amountUsd":120000000,"currency":"USD","date":"2024-06-01","investors":[],"leadInvestor":null,"source":{"fact":"round","publisher":"crunchbase","retrievedAt":"2026-05-05","status":"attributed","url":null},"status":"attributed","type":"seed","url":null}],"slug":"etched-ai","sources":[{"fact":"round","publisher":"crunchbase","retrievedAt":"2026-05-05","status":"attributed","url":null},{"fact":"existence","publisher":"etched.ai","retrievedAt":"2026-05-05","status":"self-reported","url":"https://etched.ai"},{"fact":"description","publisher":"etched.ai","retrievedAt":"2026-05-05","status":"self-reported","url":"https://etched.ai"}],"stage":"series-a","techStack":["Vercel","Lenis 1.3.11","Next.js","React","Webpack"],"totalRaisedUsd":120000000,"updatedAt":"2026-05-05"}