Gnosly Feed - fintechhttps://gnosly.com/feed/fintech/Business Intelligence Console and Prediction Algorithms3855https://www.linkedin.com/in/rubendominguezibar/Rubén Domínguez IbarRevolut is now officially worth $115B. The fintech confirmed the valuation today. Revolut serves over 80M retail customers and posted $2.3B in pre-tax profit last year.<p>The climb in just over two years:</p><p>1️⃣ August 2024: $45B</p><p>2️⃣ November 2025: $75B</p><p>3️⃣ November 2026: $ 115B. </p>https://www.linkedin.com/in/rubendominguezibar/0.00202026-11-04T11:25:17.100413+00:001linkedin.comrevolutfintechstockseconomy3770https://media.licdn.com/dms/image/v2/D4E22AQFSXwGm-sLdXQ/feedshare-shrink_800/B4EaGl6sYIKUAc-/0/1793784989448?e=1795651200&v=beta&t=QZ-SSkvT3j1z-vJ8Mhq5uz_VOjqD8m0_5ngWQjkuMf0View image2026-11-04T11:25:18.115210+00:003764@maxkarpisMax KarpisRevolut’s quiet bid for AI Independence: Cutting coding costs by 90% - Frontier models remain available but are now the exception.<p>In a recent presentation, David Abitbol, head of AI operations at Revolut, explained how the company has reduced its dependence on expensive frontier models. By self-hosting open-weight models (primarily GLM) on dedicated GPUs, Revolut cut the cost of AI coding assistance by roughly 90%. The blended cost is now about 30 cents per million tokens — roughly one-tenth of previous Anthropic/OpenAI spend.</p><p>After the switch around April 2026, results across ~2,000 engineers showed coding time improving by ~60% and lead time by ~70%. Earlier tools (Gemini, early Claude Code under budget caps) delivered smaller gains. </p><p>The bigger lift came once token limits were removed. Engineers stopped rationing AI for the hardest tasks and used it freely for codebase navigation, config and surrounding work. Coding itself was never the main bottleneck, but lead time was.</p><p>Revolut focused on making existing hardware efficient rather than simply buying more. Over several months, with the same GPU count, performance improved substantially even as demand rose, delivering interactive speeds engineers accept at a fraction of the old cost. </p>https://x.com/maxkarpis/status/21171474672757311630.00102026-11-02T11:31:14.060246+00:002x.comairevolutfintech3662https://pbs.twimg.com/media/HWGeV9gW4AE2PEB?format=jpg&name=smallImage2026-11-02T11:31:14.421051+00:00