Faris Sbahi 🏴☠️
@FarisSbahi
cofounder & ceo @normalcomputing | steering thermodynamics to reason | formerly, @theteamatx @googleai @palantirtech @dukeu | فارس سباهي
Come out and join us, so we can talk about physical world AI and thermodynamic chips. Looking forward to hosting with @nathanbenaich.
next up in nyc: @airstreet ai happy hour on 5/7 hosted with friends @normalcomputing at 5pm no talks, no agenda, just 2 hours with interesting people across research, eng, product, design, startups, academia and bigger cos drinks and food incl. airstreet(dot)com/events
Another session with @headinthebox Erik Meijer about LLMs and why we should listen but verify, proof carrying code from LLMs, #include “Asimov.h”, and not squandering the joys of freedom. No vibe coding today because I don’t currently have a laptop.
Builders compete but they never destroy. It’s the unspoken rule.
surely this new AI-for-chip-design startup founded by children who have never taped out a chip will be successful!
We had IMO gold at home the whole time!
🚨 Olympiad math + AI: We ran Google’s Gemini 2.5 Pro on the fresh IMO 2025 problems. With careful prompting and pipeline design, it solved 5 out of 6 — remarkable for tasks demanding deep insight and creativity. The model could win gold! 🥇 #AI #Math #LLMs #IMO2025
Very interesting details shared by Garry: - Windsurf had ~250 employees - ~55 eng, ~185 sales - Hired ~150 ppl (mostly sales) the last 12 months (likely all enterprise sales) - ~40 core engineers went to Google Reads like Google only wanted (most of the) eng team, and no sales
Varun Mohan and the team at Windsurf built something great. What I heard is every founding engineer (over 40 of them) cleared 7 figures. The remaining org of 200 was only 15 engineers who had worked there less than a year, the rest were sales and ops, and even then they left…
Varun Mohan and the team at Windsurf built something great. What I heard is every founding engineer (over 40 of them) cleared 7 figures. The remaining org of 200 was only 15 engineers who had worked there less than a year, the rest were sales and ops, and even then they left…
Check out this bold new work from @NormalComputing — forecasting a golden age of computing driven by physics-based hardware and architectures!
The era of Physics-based ASICs has arrived. On the surface, computing infrastructure seems unready to meet the demands of AI. But underneath, a Cambrian explosion of new computer architectures is rising to meet this challenge, with the goal of using physical processes as a…
The era of Physics-based ASICs has arrived. On the surface, computing infrastructure seems unready to meet the demands of AI. But underneath, a Cambrian explosion of new computer architectures is rising to meet this challenge, with the goal of using physical processes as a…
Good PMs gather feedback. Great PMs design UIs to collect it. In an AI-driven world, relying on feedback forms or even thumbs up/down isn't enough. Feedback often comes dressed in great UI design. You have to bake it in from the outset.
Here @elonmusk proves once again that his methods work. Assemble a small team of world-class engineers who work at a hardcore pace. An excerpt from the Elon Book (coming soon!)
We just unveiled Grok 4, the world’s smartest artificial intelligence. 🧵 Grok 4 outperforms all other models on the ARC-AGI benchmark, scoring 15.9% - nearly double that of the next best model - and establishing itself as the most intelligent AI to date.
In my experience, the best founders aren't loud. The best ones aren't shitposting, rage-baiting, nor glorifying their ego on social media. Instead they're mission-driven (not fame/ego-driven), and are quietly and maniacally focused on product, customers, and product-market fit.…
Underrated life advice: Reliability is an opportunity magnet. Opportunities accrue to the person who shows up. The one who keeps their word. The one who follows through. The one who makes things easier. Those who show up for the small will eventually earn trust for the big.
I suspect some sort of Berkson’s paradox happens here. Being good at software things A and B are typically correlated. But because of the market, at the lower tier of companies, those will be _negatively_ correlated. Which leads people to think you can’t do both A and B.
We spoke with @crmiller1 (Author, Chip War) about the competitive dynamics in data center build-outs between the U.S. and China. “China's smuggled in a large number of chips... but it appears a lot of them are sitting in data centers with low utilization.” “All the anecdotes we…
While I briefly have no employer, let me tell you what's really happening with AI companies training on public data: [1/n]
I'm not affiliated with nor endorse Symbolica anymore ( x.com/bgavran3/statu…) But training in languages like Lean is something I believe is deeply important! Less so because of the formal math aspects, but more about the mechanistic abillity of the compilers of these...
I have an unfortunate personal announcement: I was fired from Symbolica. This was a complete surprise. I am in shock and, honestly, still not sure how to process this. If you've been following me, you understand just how excited and optimistic I am about the programme of…
(1/n) Sampling from the Boltzmann density better than Molecular Dynamics (MD)? It is possible with PITA 🫓 Progressive Inference Time Annealing! A spotlight @genbio_workshop of @icmlconf 2025! PITA learns from "hot," easy-to-explore molecular states 🔥 and then cleverly "cools"…
VerilogEval is an odd coding dataset - it gives the agent access to the golden testbench, which you will use to evaluate the final design from the model. I'm honestly surprised how papers like VerilogCoder or ChipAgents only get ~95% accuracy in this setting... That's what…