🤖 Meta to buy Chinese startup Manus to boost advanced AI ↗
Meta’s buying Manus, the “general AI agent” startup that’s been talked up as more doer than talker - like a chatbot that gets off the sofa. Terms weren’t disclosed, but sources peg the valuation in the low single-digit billions.
Manus relocated to Singapore, keeps ties in China, and has a strategic partnership with Alibaba. Meta’s plan is straightforward: fold Manus tech into Meta AI and its broader product stack, because everyone wants “agentic” features now… or so they say.
🧩 Nvidia takes $5 billion stake in Intel under September agreement ↗
Nvidia has closed its $5B Intel stake purchase, locking in a big chunk of shares via private placement. It reads like a balance-sheet move, but it still smells like a platform bet.
US antitrust clearance is in, Intel gets a welcome capital boost, and Nvidia gains another lever in the “who controls the compute stack” chess match. Friendly cooperation, sure - with elbows tucked just out of sight.
🧠 Nvidia Stock Slips as Groq Deal Raises Questions ↗
The reported Nvidia-Groq licensing deal looks huge on paper, and that’s exactly why it’s getting prodded. The arrangement is described as nonexclusive, with Nvidia licensing Groq tech and bringing in key staff, while Groq stays independent.
Analysts frame it as either a clever inference shortcut or an expensive “buy the brains” move. The vibe is - Nvidia’s widening its moat against in-house silicon efforts elsewhere, but the price tag still makes people squint.
🏗️ SoftBank Group to Acquire DigitalBridge for $4 Billion to Scale Next-Gen AI Infrastructure ↗
SoftBank is buying DigitalBridge for $4B to bulk up on the unsexy stuff AI lives on - data centers, connectivity, the whole pipes-and-power situation. It’s classic Son: grand vision, concrete assets, a dash of “we’re doing this anyway.”
DigitalBridge keeps operating, SoftBank frames it as scaling “AI infrastructure at scale” (yes, that phrase loops a bit), and the broader message is clear - models don’t run on vibes, they run on racks.
🧯 OpenAI says it's hiring a head safety executive to mitigate AI risks ↗
OpenAI’s hiring a “Head of Preparedness” to focus on emerging, severe-risk scenarios - from security misuse to broader societal harms. The job pitch is unusually blunt about pressure levels, which lands as either refreshing… or mildly alarming.
It’s another sign that “safety” is becoming an org-chart reality, not just a blog-post category. Hiring is the easy part - the hard part is what happens when safety collides with shipping.
🗣️ China is considering a raft of new controls for training AI on chat log data. Here's what it means. ↗
China is weighing new rules around using chat logs to train AI - focusing on consent, disclosure, and tighter handling of user conversations. It’s basically “you can’t just hoover up everything people say to a bot,” which lands as fair.
There are also angles around user rights like access or deletion, plus extra guardrails for sensitive groups. The vibe is privacy meets governance - two trains on the same track, not always moving in the same direction.