{"changes":[],"country":"GB","description":"Fractile designs in-memory inference chips for large language models, targeting at least one order of magnitude better cost and energy per token. It serves AI labs, model providers, and inference platforms running frontier-scale LLMs. The thesis is that compute-in-memory beats the von Neumann bottleneck that limits incumbent GPUs.","domain":"fractile.ai","employees":null,"founded":2022,"fundingTotalQuarantined":false,"githubOrg":"fractile","githubStars":null,"industry":"ai","investors":[],"lat":51.507446,"lng":-0.127765,"location":"London, UK","name":"Fractile","rounds":[{"amountQuarantined":false,"amountUsd":15000000,"currency":"USD","date":"2024-09-01","investors":[],"leadInvestor":null,"source":{"fact":"round","publisher":"sifted","retrievedAt":"2026-05-06","status":"attributed","url":null},"status":"attributed","type":"seed","url":null}],"slug":"fractile-ai","sources":[{"fact":"round","publisher":"sifted","retrievedAt":"2026-05-06","status":"attributed","url":null},{"fact":"existence","publisher":"fractile.ai","retrievedAt":"2026-05-06","status":"self-reported","url":"https://fractile.ai"},{"fact":"description","publisher":"fractile.ai","retrievedAt":"2026-05-06","status":"self-reported","url":"https://fractile.ai"}],"stage":"seed","techStack":["Cloudflare","Vercel","Prismic","Vue.js 3.5.14","Nuxt.js","Java","Nuxt"],"totalRaisedUsd":22000000,"updatedAt":"2026-05-06"}