What if the biggest bottleneck in AI wasn't the models themselves, but the fundamental math underneath them? That's the question a new company called Subquadratic is answering — and their answer is pretty remarkable. This comes from Justin Dangel, co-founder and CEO of Subquadratic, published on their site at subq.ai.

Here's the core problem. Every transformer-based AI model — and that includes basically everything you've used, from ChatGPT to Claude to Gemini — has a scaling issue baked i...

---

Source: Subquadratic

Original article: https://subq.ai/introducing-subq

Generated by cjav.dev TLDR