Jurors eye Musk and Meta's $2.5B unravel

Jury side-eyes Musk. The Musk v. Altman trial is officially underway, and jury selection quickly revealed the obvious: Elon Musk is a polarizing guy. In federal court in Oakland, several potential jurors admitted they weren’t exactly fans of Musk, though Judge Yvonne Gonzalez Rogers made clear that disliking a celebrity CEO doesn’t automatically make someone unfit to serve. Only one juror was dismissed over especially strong anti-Musk views. The final nine-member jury includes a pretty eclectic mix, from a painter to a psychiatrist, and some also voiced broader skepticism about AI itself. At the heart of the case is Musk’s claim that OpenAI, under Sam Altman, drifted from its nonprofit mission. Outside the courtroom, the drama kept humming: Altman and Greg Brockman showed up, Musk didn’t, protesters rallied against the AI race, and both sides sparred on X. Opening statements begin next. More here.

Data center split. America’s data center boom is turning into a tale of two grids. On one side: power-rich states and regions racing to welcome AI infrastructure, jobs, and investment. On the other: areas hitting hard limits on electricity, transmission, and local tolerance, where giant server farms are starting to look less like economic candy and more like a strain on the system. The divide matters because AI’s expansion now depends less on chip supply alone and more on whether utilities can actually keep the lights on. That’s creating winners and losers across the US, with developers chasing cheap land and reliable power while communities weigh tax revenue against water use, noise, and grid stress. In short, the next big AI bottleneck may not be silicon—it may be geography, politics, and plain old electricity. More here.

Meta's deal mess. Meta may have a pricey AI headache on its hands. After buying Manus—a Singapore-based, China-linked startup focused on AI agents—for $2.5 billion in December, the company now appears to be preparing for the possibility that it has to unwind the whole thing. The reason: China has reportedly banned the transaction on national-security grounds. That’s a big problem because Meta didn’t just buy Manus and leave it alone—it quickly started weaving the startup’s technology into its own systems. So if the deal gets reversed, this won’t be a simple paperwork exercise; it could mean a messy and expensive effort to separate tools, teams, and tech that are already intertwined. It’s the latest reminder that in AI, cross-border deals can move fast, but geopolitics can move faster. More here.

Chips over people. Big Tech is making the AI trade-off loud and clear: fewer employees, more infrastructure. Microsoft and Meta are the latest to slim down in the name of AI, with Meta planning to cut about 8,000 jobs and Microsoft offering voluntary retirement to roughly 7% of its U.S. workforce—a move that could still end in layoffs if not enough people bite. They’re hardly alone. Oracle, Snap, and Block have all made fresh cuts, with Block alone slashing 40% of its staff, or more than 4,000 roles. The bigger picture is getting hard to ignore: 45,800 tech layoffs were announced in March, the highest monthly total in at least two years, per Layoffs.fyi. The bet is that chips, servers, and AI capacity will drive the next wave of growth—but some companies may find that swapping talent for compute comes with a real cost. More here.

Samsung's glasses play. Samsung is gearing up to show off its Galaxy Glasses later this year, and the pitch is pretty clear: take the Meta Ray-Ban formula, add Android XR, and let Gemini do the heavy lifting. The first model, codenamed Jinju, reportedly skips a display entirely and leans into AI features like live translation, photos, directions, and quick answers. Specs floating around include a Snapdragon AR1 chip, 12MP Sony camera, 155mAh battery, Bluetooth 5.3, and a lightweight ~50g frame with photochromic lenses. A pricier display-equipped version, Haean, is said to land in 2027. The catch? Samsung is entering a market Meta already dominates, with millions of smart glasses sold and serious brand momentum. So while Samsung may have the better AI story, it’s arriving fashionably late to a very crowded party. More here.

Cloud ties reset. OpenAI and Microsoft are giving their blockbuster partnership a cleaner, more flexible sequel. Microsoft stays OpenAI’s main cloud partner, and OpenAI products will still launch first on Azure—unless Microsoft can’t or won’t support what’s needed. The big shift: OpenAI can now sell its products across any cloud, a notable loosening of the old setup. Microsoft also keeps access to OpenAI’s models and products through 2032, but that license is now non-exclusive, signaling OpenAI wants more room to maneuver. Financially, Microsoft stops paying OpenAI a revenue share, while OpenAI will keep paying Microsoft through 2030 at the same rate, though with a cap. Microsoft also remains a major shareholder, so it still has skin in the game. Bottom line: less knotty structure, more strategic flexibility, and plenty of joint ambition ahead in datacenters, custom silicon, and cybersecurity. More here.

Terminal gets chatty. Bloomberg is giving its famously dense Terminal a chatbot-style overhaul with ASKB, an AI layer designed to help traders and analysts dig useful signals out of an ever-growing mountain of data. Instead of hunting through endless screens, users can ask big-picture questions—like how geopolitics or oil prices might hit a portfolio—and get synthesized answers fast. Bloomberg says the tool won’t magically turn weak analysts into stars, but it could help strong ones test more ideas in less time. ASKB also brings workflow automation, letting users build templates for recurring research tasks like earnings prep. The catch, of course, is hallucinations: Bloomberg says it’s using multiple validation checks, semantic reviews, and citation controls to keep answers grounded. Traditional Terminal screens aren’t disappearing, but Bloomberg’s CTO says this AI interface is set to become the main front door. More here.