We threw a fun event on owning your ai stack today! 80 @sequoia portfolio companies attended technical workshops on how to own your AI (models, harnesses, data, evals, RL, CL, etc). Videos and takeaways coming soon. Towards a vibrant ecosystem for Democratized Intelligence 💚

Two of the people most responsible for scaling the transformer are now betting on a next act. @MillionInt ran the Reasoning 🍓 team at OpenAI. @_arohan_ was a pre-training lead on Gemini after years at Google Brain and Anthropic. They just started @coreautoai to find what comes next. Their core argument: (1) models are trained in the lab but deployed in the real world and can't keep learning once they leave; (2) AI research is done by humans today but models will be able to explore and uncover new advances more rapidly and systematically (controversial but timely w this week's petition). The conversation covers: — why Jerry expected AGI in 2025 and what changed his mind — the two kinds of learning from experience, and why RL only captures one — the computational depth problem baked into today's architectures — why the biggest labs can't afford to look for a transformer replacement — the kernel competition where humans + $100K of coding agents found a 60x speedup no frontier model comes close to — a definition of AGI you can actually test: a model that improves itself with no human in the loop 00:00 Introduction 01:46 Appreciating Transformers 02:44 Scaling Hits Limits 04:54 Why Architecture Matters 05:32 RL Reality Check 07:32 Test Time Learning 09:52 Economics Of Scaling 12:47 Why Start A Company 14:24 Rohan On Transformers 19:11 Computational Depth Problem 20:32 When Transformers Top Out 23:22 Beyond Reinforcement Learning 26:41 Optimization And Efficiency 34:24 Building An Automated Lab 39:45 Kernel Automation Roadmap

We @Sequoia are leading the Series C in @Etched Etched has built a beautiful machine in Gen 1. And Gen 2/3/++ will only be faster and more ambitious. Maximizing intelligence per flop is both insanely fun engineering and an incredibly noble mission. Honored to be on board.

The first Vera Rubin clusters are here! Yesterday, @IneffableLabs took delivery of their Vera Rubin NVL72 cluster from @googlecloud @nvidia The AI frontier jumps forward by yet another generation of hardware. Acceleration continues.

We have an awesome roster of creators already signed up for the @fal x @sequoia video hackathon, taking place July 17-19th. Reminder to register! Lots of workshops/tutorials to get your feet wet, and unlimited video inference credits to whet your creative exploration :)

I built to make it easier to browse the 1,500+ Krea 2 Style LoRAs I trained. You can try them with your own prompts, run a random loop to generate lots of examples, or blend multiple styles together. Images can be generated directly from the site using @fal API keys, or you can grab the LoRA weights from the @huggingface link on each style page. Every LoRA was trained for just 100 steps. If you like one, you can always build a larger dataset around it and retrain it with more steps.

Dylan grew up working in motels Got a screen of death on his XBOX as a kid Fixing it got him into semis Somehow this led him to starting SemiAnalysis, the premier semis research co @dylan522p is fascinating and on an all time run 👀 @sonyatweetybird

Featuring the man of the moment @dylan522p @shaunmmaguire in today's Training Data episode. Nobody is a more trusted industry insider to the biggest infrastructure build-out in history. The story of how Dylan became @SemiAnalysis_ is even more awe inspiring: a young motel kid, following his curiosity about the compute industry relentlessly down rabbit holes, from Reddit forums to audiobooks to Japanese chemicals conferences... 00:00 Introduction 01:58 Motel Kid Origins 03:11 Xbox Repair Spark 04:23 Internet Forums to Semis 06:42 From Quant to Founder 09:16 Homeless Research Roadtrip 14:04 InferenceX and Benchmarking 34:35 Sparse vs Dense Models 35:08 Interconnect Shapes Architecture 35:48 CUDA Moat Is Shifting 36:46 Ecosystems and Co-Design 38:46 Cerebras Speed and Limits 42:07 ROI Debates and Hot Takes 44:20 Ten Year Tech Bets 50:48 Compute Crunch and NeoClouds

What happens when you shift more of the context layer into the model weights themselves? Engram is building a neolab focused on memory and continual learning. Let’s go @dan_biderman @realJessyLin! Fun chat w @shaunmmaguire

📹️ Announcing the fal x @sequoia 72-Hour Video Hackathon. A 3-day global sprint for AI-native filmmakers, creative technologists, developers, designers, and storytellers building the future of video. Supported by leading AI video labs @GoogleDeepMind, @xAI, and @Kling_ai. Participants will get: - Access to frontier video models - Workshops + talks from industry leaders - Mentorship from top creators + engineers - $150k credit prize pool This is a chance to build at the cutting edge and create what wasn’t possible before. Judges, speakers and event details coming soon. Participation is by application only. Apply here:

I believe that @SpaceX is the most important company ever It transcends the scope of a traditional company It will open up the Stars Elon had the vision And then the team pushed through limitless pain to get where they are Which is just the beginning

🚀Launching: LangSmith Engine LangSmith Engine is an agent that sits on top of your traces It runs in the background and automatically identifies issues It then proactively suggests action items (code changes, evaluators to add) Try it today: