One of AI's most influential researchers is joining AMD in an $8 billion deal
THE SO WHAT
Bringing Fei-Fei Li in as chief scientist signals AMD wants to compete at the model-and-systems level, not just sell accelerators. Hardware buyers should assume AMD’s value prop will increasingly include reference world-model stacks for robots, factories, and physical AI—worth factoring into long-term platform bets.
READ THE SOURCE
MORE FROM THE WIRE
Startups & VentureRoboTechnik Leads Hong Kong’s $1.8 Billion Listing Debut Lineup
Four IPOs raising $1.8 billion in a single day in Hong Kong show public markets are reopening to hardware and industrial automation stories. If you’re building in robotics or deep tech, the exit stack is diversifying beyond U.S. listings and late-stage private rounds—board conversations about listing venue are back on the table.
Startups & VentureAMD makes a big bet on the next era of AI with World Labs acquisition
AMD folding World Labs and Fei-Fei Li into a “next era of AI” story is about bundling compute with native world-model expertise. For enterprises, this could mean more turnkey physical-AI reference designs tied to AMD hardware—useful leverage in negotiations with both GPU vendors and systems integrators.
Startups & VentureAMD is acquiring AI company World Labs in a deal worth more than $8 billion
An $8.2 billion all-stock deal for a 2024-founded lab valued at $1 billion months ago is AMD buying not just models but a worldview—world models tightly coupled to its silicon. If you’re building physical AI or robotics, expect AMD to push a vertically tuned stack and start asking whether your roadmap assumes NVIDIA by default or can ride a second ecosystem.
Startups & VentureSource: Inference provider Modal Labs closing in on $750M round at $15.75B valuation
Inference infra is now being priced like a strategic utility, not a dev tool — a $750M round at a $15.75B valuation, tripling in four months, is a bet that serving tokens at scale is its own category. If you’re building on third-party inference, model your vendor and concentration risk the same way you would a core cloud provider.