Anand Sampat

Very proud to announce Buoyant Ventures latest investment in Hammerhead to help unlock more power for AI and data center growth. My partner Laura Katzman (Dyer) led this deal and our research into efficient computing which helped us win this competitive and over-subscribed deal. Thanks to Maeve Allsup for your excellent explanation of the opportunity. https://lnkd.in/gSM4PadB

64

2 Comments

AMI and Yann LeCun are betting the real future of AI is specialized. Their take: world models built on JEPA that reason causally, plan under constraints, and work in messy real-world environments. Their Feb 2026 paper coins it: Superhuman Adaptable Intelligence (SAI). The idea in short is “Outperform humans on specific tasks, tackle domains we simply can’t.” Nabla (clinical AI) is already on board as AMI’s first partner. More architectural innovation instead of more compute. Curious to see where this goes. https://lnkd.in/gmB4k4EA

19

2 Comments

How much of AI's next step change is a compute problem to solve and how much is a connectivity problem we've been misattributing to compute because compute is what we've been measuring? We scaled aggressively on the axes of more GPUs, more parameters, more training data but the fundamental network topology stayed largely unchanged with local connectivity fast and remote slow leading to scale-up and scale-out being separate worlds. We treated connectivity as infrastructure to be optimised within rather than a constraint worth removing and leveraging for performance gains. This post is an attempt to take network and connectivity seriously. Some companies working in this and adjacent spaces are Oriole, Callosum and Nexthop AI. Oriole Networks is building at the fabric layer. Callosum is demonstrating what heterogeneous compute orchestration can do for performance and speed across AI workloads. Nexthop AI is focused on the Ethernet switching layer. Each rethinks this constraint from a different angle. https://lnkd.in/eHsbsCwh

29

With continued discussions and claims of Generative AI's capacity to overtake human labor, research, and problem solving, it becomes consistently more pertinent to differentiate sensationalism from reality. "Actually deploying systems that are reliable is where things tend to falter," says Yann LeCun, Chief AI scientist at Meta, in a podcast with Alex Kantrowitz. "It's basically why we had super impressive autonomous driving demos ten years ago but we still don't have level five self driving cars. It's the last mile that's really difficult." Using an AI agent to meaningfully impact a company's workflow requires an intense level of specification, especially in the face of the omnipresent hallucinations of LLM's. I highly recommend giving this podcast a listen - especially interesting to contrast with the messaging of other leaders in the space. Link here: https://lnkd.in/efmm7rJi

14

1 Comment

The next frontier of AI isn't just about choosing the right model. It's about building moats through implementation. Across the Emergence Capital portfolio, companies like Bland and Federato are scaling faster by embedding Forward-Deployed Engineers. These engineers bridge the gap between AI and reality, ensuring real results in production. Read more from my colleagues Gordon and Kyle on why FDEs are the key to unlocking AI's full value. https://lnkd.in/gcsRRGX2

29

3 Comments

Gemini 3 Deep Think is officially here! 🚀 This launch brings a big step in advanced reasoning designed specifically for science and engineering. It builds on our long-term work in inference-time scaling, proving its power not just in IMO and ICPC competitions, but in solving new research problems. The new state-of-the-art metrics speak for themselves: an impressive 84.6% on ARC-AGI-2, alongside repeating Gold-medal performance at IMO '25 and ICPC '25 (with a full score!). Beyond the incredible competition benchmarks, it's been a privilege to benefit from this model early on in our own work. As we shared in our blog post yesterday, we used Gemini to solve 17+ open research problems across Physics, CS, and Economics. It's been so much working with this amazing team. The model is improving so fast that we already know it can push several of the results we just published even further. We are actively exploring these new capabilities and will have more to share about it soon. The pace of AI-assisted scientific discovery is unprecendented. Check out the official launch details and try it out yourself: 📄 Our Research from Yesterday: https://lnkd.in/eK4Skchv 📄 Our arXiv Paper (17+ Agentic Results): https://lnkd.in/eB87ntYp 🔗 Gemini 3 Deep Think Launch: https://lnkd.in/ehY-__kU

309

4 Comments

Barbarians at the Gate (https://lnkd.in/gkEBX79s) is a very interesting new paper from some great folks at Berkeley, with some exciting results showing the potential for AI in systems research. But I think the authors aren't quite asking the hardest problem about where this takes systems as a field. I wrote a new blog post about it: https://lnkd.in/gxS3s4v4 The core point of the paper is that AI can be effective at automatically driving the 'number go up' part of systems research. Given an evaluator, AI can automate progress on implementation. I believe this is 100% correct. The paper also talks about how more focus will have to go into developing evaluators. Again, I 100% agree, but think they're understating their case. Building evaluators (basically tests) is, I believe, going to become the focus of systems and software over the next decade. Which leaves researchers with the hard problem of choosing which problems to go after. That's never been easy in academic systems, and is only going to become more of a challenge. But the best systems research is driven by choosing problems well! This vision for AI could lead to lots more of the least interesting systems research. The "number go up" kind. Which could make it harder than ever to identify the truly innovative and visionary work. We're already seeing that elsewhere in science. So, mostly, I think the paper is right, but not yet asking the biggest question: what should systems research be?

276

12 Comments