Carlo CocuzziAI Expert, Pod Cast Host, Real Estate Investor and Funding Specialist

AI Is Now Designing Better AI | MOONSHOTS

19 Jul 2026 · 1 min

Watch on YouTube

1:16

In this brief dialogue, two speakers discuss a novel approach where two AI systems collaborate to design superior AI chips. The first AI is a convolutional neural network trained to act as a fast physics simulator, reducing traditional solver times from minutes or hours to milliseconds. However, this method depends on large, proprietary training datasets—referred to as the "Magnum Opsta"—currently held by companies like NVIDIA. The speakers explore whether unlocking such data or achieving sufficiently accurate simulators could enable a recursive self‑improvement loop, potentially triggering an intelligence explosion in chip design. They highlight the strategic importance of chip efficiency for AI advancement and note the ongoing race between data‑locked incumbents and open‑source simulation efforts.

Chapters

AI is now designing better AI chips, and training data is the catch-22. Here's the clever part: it's not one AI, but two working together. First, they trained a convolutional neural net, the same kind of model built for image recognition, to predict the physics. What used to take traditional solvers minutes to hours now takes milliseconds. But here's the catch, the tease: the AIs require training data, and all that training data is locked up in— yes, you got it— the Magnum Opsta. Companies out there. If this training data can be unlocked, can we see an intelligence explosion in the design of AI chips, which is the inner, innermost loop?
Just to clarify one part of that, the convolutional neural net is effectively acting like a simulator. And any place you can build a simulator, the AI can have a field day because it can check its own work and it can work for weeks or months improving itself if the simulator is accurate. The chip area, you know, is going to be massively impactful for the recursive self-improvement of AI. And it's an open question right now whether that data is truly locked inside NVIDIA and a couple of other companies, or whether the simulators are good enough to allow you to just generate a circuit, see if it would have worked, generate the next, see if it would have worked. So those are in a footrace right now.
But, you know, it's a moment in history that's pretty unprecedented.