TL;DR: Beijing just ordered Meta to unwind its $2 billion acquisition of AI agent startup Manus—a Chinese-founded firm that’s now headquartered in Singapore. The unexpected move comes as AI tensions between the US and China ramp up—and it sends a warning to both Chinese AI startups and US tech firms hoping to strike deals with them.

What happened: Originally founded in China, Manus relocated its headquarters to Singapore last summer before Meta announced it would acquire the firm in December 2025. On Monday, China’s National Development and Reform Commission said it would block the “foreign acquisition” of Manus and require the parties to unwind it—though whether Beijing can actually block the purchase of a company that’s no longer based in China is unclear. A Meta spokesperson told CNBC that the deal “complied fully with applicable law.”

Beijing’s Ministry of Commerce started its probe into whether the deal violated China’s export control and cross-border technology transfer rules in January. By March, Manus CEO Xiao Hong and Chief Scientist Ji Yichao had been summoned to Beijing for a regulatory review of the Meta deal and barred from leaving the country.

“China-shedding” hits a wall: It’s not uncommon for Chinese-founded startups to cut domestic ties to insulate themselves from US-China geopolitical tensions. Singapore tends to be an attractive destination for its low taxes, regulatory environment (including a US tariff of just 10%), and a friendly diplomatic posture toward both Washington and Beijing (at least, apparently, until now). Exactly how many Chinese-founded firms have moved there is unknown, but Reuters reports that the trend is accelerating.

Tit for tat: The Manus block comes as the AI competition between the US and China heats up. Last week, the White House accused Chinese firms of “industrial-scale” AI model theft through distillation attacks and said it would work with US AI companies to combat it. The House also just advanced a slate of bipartisan bills to further tighten chip export controls. Beijing, meanwhile, is now also looking to require government approval before its tech firms can accept US investment.

Bottom line: For Chinese tech startups, expatriating was a workaround to partner with US tech firms. Beijing is now indicating it’s no longer letting that slide. Even if the order proves toothless, it could have an effect on every Chinese AI startup eyeing a Western buyer—and every US tech firm hoping to scoop one up. —WK

Also at Meta…

• It just signed a deal with a startup that could let it harness solar power at night for its AI data centers.

Google's latest Pixel update is draining batteries faster than users can complain about it

Imagine unplugging your phone and watching the battery percentage drop in real time—even when you're barely touching it. Google's recent security update, which rolled out April 8, was supposed to address battery issues introduced by the March update, but made things worse instead of better. Thousands of Pixel users have been taking to Reddit threads and a Google Issue Tracker to vent ever since, with new comments pouring in every day.

Users in the threads suspect a background CPU bug is the culprit—though Google hasn't confirmed what's causing it or when a fix is coming. So far, the tech giant's response amounts to: We know; we're looking into it. Pixel owners are, predictably, thrilled:

"Battery down 40% after 2 hours in a plane w/ the phone turned off! Happened again on return flight!!!!!!!!!"

"I used it for 2 hours, and the battery dropped from 100% to 35%. It seems like my Pixel 8 Pro drains faster than my Pixel 3."

"I've troubleshot this to the end of the earth and back and have had no luck. […] Just utterly abysmal battery no matter what I do."

A fix is probably coming. Maybe. Eventually. Dare we suggest investing in a really good portable charger? —SM

If you have a valid, funny, strange, or petty rant about a new software update or a technology that just released, fill out this form and you may see it featured in Tech Brew.

Inside you are two wolves: one who resists the urge to make an AI-generated image of a real wolf on the loose—and the other who gleefully gives in, only to have the joke come back to bite you. A 40-year-old man in South Korea was arrested after he posted a deepfake of a real runaway wolf that escaped one of the nation’s zoos in early April. The AI image led to emergency text blasts warning people to avoid the area of the reported sighting and misdirected the police search.

The actual animal, a 2-year-old wolf named Neukgu who’s part of an effort to restore the wolf population in South Korea, was found a little over a week ago after nine days on the lam—and has become a beloved viral sensation. (The city where he resides is reportedly weighing whether to make him an official city mascot.) The AI deepfaker’s reason for sowing chaos and wasting police resources? For some “fun,” according to the BBC. Neukgu is heading home a national hero; the man who cried AI wolf is facing up to five years in prison. —WK

Chaos Brewing Meter: /5