Meta quietly hired two more founding members from Mira Murati's Thinking Machines Lab
THE SO WHAT
Meta pulling founders and researchers out of Thinking Machines Lab is another data point that frontier AI talent is consolidating inside a few mega-platforms — not independent labs. If you’re a startup betting on proprietary AI research as your edge, your real risk isn’t model performance, it’s retention against compensation and compute you can’t match.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIHow to Ground a Korean AI Agent in Real Demographics with Synthetic Personas
Grounding Korean AI agents in synthetic personas tied to real demographics is a template for localized, behaviorally tuned agents without touching sensitive PII. If you're deploying agents in any non-US market, your edge won't be the base model — it'll be how fast you can stand up credible, compliant persona layers like this.
The drama between a software engineering veteran and Google is heating up — and playing out in public
Public debate over uneven AI use inside a major engineering org is a tell — AI adoption is now a cultural fault line, not just a tooling choice. If you're not measuring and actively managing AI usage variance across teams, you're going to get pockets of 10x productivity sitting next to teams still working like it's 2019.
Applied AIAnthropic takes $5B from Amazon and pledges $100B in cloud spending in return
AI labs are now functionally anchor tenants for hyperscalers — $5B in equity for a $100B AWS commit turns model training into pre-sold cloud revenue. If you're not a frontier lab, assume list price on GPU cloud is subsidizing these mega-deals and plan your infra mix accordingly.
Applied AIQ&A with Canva CEO Melanie Perkins on the company's growth in enterprise, competing with AI labs, token pricing, investing in its own models, and more (Nilay Patel/The Verge)
Canva treating token pricing and its own models as core strategy says the 'AI feature' phase is over — design platforms are now de facto model platforms with UX on top. If you're an application layer company, you either own your economics on tokens or you become a thin skin over someone else’s margin structure.