Meta: big AI deal — and ad clean‑up

Meta spent Thursday doing two very different kinds of cleanup: it unveiled a new artificial intelligence model called Muse Spark and locked in about $21 billion of CoreWeave cloud capacity through December 2032, while also pulling Facebook and Instagram ads that were helping law firms recruit people to sue social media companies. (about.fb.com, coreweave.com, reuters.com) The first move was about speed and scale. Meta said Muse Spark already powers the Meta Artificial Intelligence app and website, and it plans to roll the model out to WhatsApp, Instagram, Facebook, Messenger, and its artificial intelligence glasses in the coming weeks. (about.fb.com) Muse Spark is a large language model, which is the part of an artificial intelligence system that predicts the next word the way your phone predicts the next word in a text, just at far larger scale. Meta called it its most powerful model yet and said it was built specifically for Meta products rather than as a research demo sitting on a shelf. (about.fb.com) The CoreWeave deal is the machinery behind that promise. CoreWeave said Meta will buy dedicated artificial intelligence cloud capacity worth about $21 billion through the end of 2032, spread across multiple locations and including some early deployments of Nvidia’s Vera Rubin platform. (coreweave.com, reuters.com) That kind of contract is less like renting a few hotel rooms and more like booking whole floors years in advance. CoreWeave said the capacity is meant to help Meta scale inference workloads, which is the expensive step where a trained model answers millions of real user requests. (coreweave.com) Investors liked the message because it said two things at once: Meta has a new flagship model to ship now, and it has lined up the computing power to keep serving it later. StockStory said the announcements helped lift Meta shares, while Reuters described the CoreWeave agreement as part of Meta’s push to support more complex artificial intelligence workloads. (stockstory.org, reuters.com) Then came the other story, which was about what shows up inside Meta’s own ad system. Reuters reported on April 9 that Meta began removing ads on Facebook and Instagram that were aimed at recruiting new plaintiffs for lawsuits claiming the platforms were designed to be addictive for young users. (reuters.com) Those ads were not random legal notices. Axios reported the pullback came about two weeks after a California jury found Meta and Google liable in a landmark social media addiction case, which gave plaintiff lawyers a reason to go hunting for more families with similar claims. (axios.com, reuters.com) The awkward part is that Meta’s advertising tools are built to find exactly the people an advertiser wants. In this case, the same targeting and distribution system that sells shoes, games, and streaming subscriptions could also help lawyers find teenagers and parents who might join litigation against Meta itself. (reuters.com, theverge.com) So the picture from April 9 and April 10 is not just “Meta is spending big on artificial intelligence.” It is “Meta is rich enough to reserve years of specialized computing while still having to police its own platforms when those platforms start feeding legal risk back into the company.” (coreweave.com, reuters.com)