
Tutor is one of the companies I'm most excited about, and I'm particularly glad to support it because it is run by some of my favorite people! (And also robots are cool)
✓pilot of @flappyairplanes. i make ml go brr.

Tutor is one of the companies I'm most excited about, and I'm particularly glad to support it because it is run by some of my favorite people! (And also robots are cool)

Good morning friends. Flappy's quest for extreme learning efficiency has led us to build the 'Kittens Virtual Machine' (KVM) framework, which launches a megakernel that acts as a persistent runtime and exposes its own instruction-based programming model. The KVM (a) provides unprecedented control and visibility over the hardware (something we've been extensively utilizing to accelerate our internal research), exposing a completely new kind of systems–training co-design space, (b) reaches vastly SoTA speeds without hand-tuned kernels, and (c) tremendously increases GPU utilization by co-packing multiple independent workloads into the same kernel.

Hello again, it's been a while. We're writing to let you know that we have raised quite a lot more money. We're letting you know because we are on the lookout to hire smart people and this will help us do that. See you again when we have a product or have raised more money!

i propose, tonight only, we get to light off a nuke over the pacific. for america's 250th the world can let us have one special firework, just for the occasion

All reliable clusters are alike; each unreliable cluster is unreliable in its own way.

megakernels remain underrated. if you haven’t dug into them before go look them up! flappy seems to be hinting at some really powerful training megakernel stuff which is sick ex: fully contained training megakernel could be great for automated research

Fun to share a little taste of some of the things we've been working on with @sequoia

We’ve been thinking a lot about scaling laws, wondering if there is a more effective way to scale FLOPs without increasing parameters. Turns out the answer is YES – by looping blocks of layers during training. We find that predictable scaling laws exist for layer looping, allowing us to use looping to achieve the quality of a Transformer twice the size. Our scaling laws suggest that for a fixed parameter budget, data and looping should be increased in tandem! 🧵👇

Very grateful and excited to be working together every week!


The new guard in AI has emerged: next generation of top tier AI founders just started a new kind of frontier lab. Good luck!

Very proud of the team we've assembled! Back to work!

Introducing Ricursive Intelligence, a frontier AI lab enabling a recursive self-improvement loop between AI and the chips that fuel it. Learn more at ricursive.com

Thrilled to share that @annadgoldie and I are launching @RicursiveAI, a frontier lab enabling recursive self-improvement through AIs that design their own chips. Our vision for transforming chip design began with AlphaChip, an AI for layout optimization used to design four generations of TPUs, data center CPUs, and smartphones. AlphaChip offered a glimpse into a future where AI designs the silicon that fuels it. Ricursive extends this vision to the entire chip stack, building AI that architects, verifies, and implements silicon, enabling models and chips to co-evolve in a tight loop. We sat down with WSJ’s @berber_jin1 to discuss Ricursive: wsj.com/tech/this-ai-startup…

Part 2 of our MLPs blog post is out! 👀 This time, we’re here to tell you the story 📖 of our quest for a construction that: ✅ Handles general embeddings 🌐 ✅ Asymptotically matches the information-theoretic limit 📊📈 ✅ Is usable within transformers 🤖✨

The U.S.–China AI race won’t be decided by who builds the most datacenters, but by who deploys the most intelligence. We call this Gross Domestic Intelligence (GDI): intelligence per watt × usable power. If the U.S. activates its dense installed base of local AI accelerators in a hybrid local–cloud system, it could add ~30–40% inference capacity and ≈2-4× GDI for single-turn chat and reasoning queries without building any new datacenters or grid infrastructure. Winning the GDI race means treating local compute as critical infrastructure and making hybrid inference the default. (1/N)