Meta and CoreWeave Just Committed $21 Billion to AI Infrastructure
Meta expanded its deal with CoreWeave, the GPU cloud company, to a $21 billion agreement running through 2032. That brings the total value of their announced deals to around $35 billion.
That is not a typo. Thirty-five billion dollars in compute commitments between two companies.
Here is why it matters: Meta is not just building features for Instagram. They are building the backbone for a fully AI-native future and they know it takes a mountain of compute to get there. CoreWeave, which went public last year, now has the kind of anchor customer that makes their business basically bulletproof for the next six years.
My take: this kind of deal is a signal, not a product announcement. When a company locks in $21 billion of cloud compute through 2032, they are telling you exactly what they believe about the next six years. Meta believes AI is going to eat their entire product stack and they are making sure they have the horsepower to run it. Hard to argue with them on that.
Florida Is Going After OpenAI Over a Mass Shooting
The Florida Attorney General launched a full investigation into OpenAI and ChatGPT over their alleged role in last year's FSU mass shooting that killed two people and injured five more.
Details are still coming out, but the allegation is that interactions with ChatGPT played a role in what happened.
This matters for a few reasons. First, this is a state AG going after OpenAI directly, not just firing off a warning letter. Second, it sets up a real legal test around AI liability that the courts have not had to sort out yet. Third, if Florida wins anything here, every other AG in the country is going to take notes fast.
My take: I genuinely do not know if OpenAI is legally responsible for what a person did after talking to a chatbot. That is a hard question and I want to see how courts work through it. But I do think the industry has been way too comfortable hiding behind "we are just a tool" as a legal shield. These cases are going to force some real answers out of people. That is uncomfortable, but it is necessary. The accountability has to land somewhere.
AI Is Moving Into Your Wallet and It Is Not Asking Permission First
Two finance stories hit today that are worth looking at together. Revolut launched AIR, an in-chat AI assistant that can actually take actions in your account. We are talking blocking payments, buying eSIMs, breaking down your spending history. Rolling out to UK users first.
At the same time, Perplexity integrated with Plaid so you can now link your bank accounts, credit cards, and brokerage accounts and query your entire financial life in plain English.
Both of these are the same story told two different ways: AI is moving from telling you things to doing things with your money.
My take: the Plaid integration is the more interesting one to me. Perplexity already has a reputation for being fast and direct. Adding your actual financial data to that turns it into something closer to a personal CFO than a search engine. Whether people trust it enough to hand over bank credentials is a whole other question. Revolut's AIR is the smarter trust-building play since it is baked into an app people already use for their money.
Either way, the era of AI assistants that can only talk is ending fast. The ones that can act are coming and they want your login.