Anthropic is days away from a deal that could rearrange who actually runs the AI sector. The company is wrapping up a raise of at least $30 billion at a valuation above $900 billion. Sequoia, Dragoneer, Greenoaks and Altimeter sit on the lead. If the paperwork clears by end of May, Anthropic jumps past OpenAI’s $852 billion price tag from March. Bear in mind, the same company was worth $380 billion in February. NewsTrackerToday spotted this curve back in Q1, right when the PwC and Goldman Sachs deals went into production.
Roughly 137 percent in three months. That kind of repricing doesn’t usually happen to a private tech company. And it says more about how AI money is moving than about Anthropic specifically. Capital here isn’t chasing growth in any old-fashioned sense. It is chasing three things: compute, model capability, and a path into the enterprise. Dario Amodei has been open about where the cash goes. Straight into infrastructure. Mostly the AWS and Google Cloud commitments that come online through 2027.
Liam Anderson, who watches financial markets for a living, didn’t hedge: “At 900 billion you’re betting Anthropic eats a real slice of every enterprise software dollar AI ever touches. Full stop. The growth multiple is wild. But so is the gap between today’s ARR and whatever terminal value you’d need to justify the number. This isn’t an upside trade. It’s a winner-takes-most trade.”
On paper the math looks ridiculous. Dig into the revenue and it gets less ridiculous fast. Anthropic just disclosed that Q1 2026 revenue grew 80x year over year. ARR is now north of $44 billion. Customers paying a million or more annually doubled, from 500 to over 1,000, inside two months. PwC, Goldman, Blackstone, Hellman & Friedman, the Gates Foundation. None of those are pilots anymore. They are production. NewsTrackerToday mapped this shift toward production-grade enterprise contracts a couple of months back as the most underrated driver of the next rerating.
There is something else hiding in the round structure. Sequoia, Dragoneer, Greenoaks, Altimeter. Those aren’t late-stage funds chasing the next markup. They are concentrated holders that win when their bets turn into operating-system-grade companies. The joint lead reads like a long bet to slot Anthropic next to Microsoft, Alphabet and Meta. Not a flip.
Sophie Leclerc, who covers the technology sector, walked it out further: “We are watching the next class of platform companies get assembled in real time. The frontier labs stopped being software vendors a while ago, though most of the press hasn’t caught up to that yet. They are infrastructure now. Their models will sit underneath legal work, finance, medical research, the whole code stack. You can only price them this high if you actually believe that. The cap table here suggests these investors do.”
Stack this up against OpenAI and the picture shifts. OpenAI had the bigger price tag through 2025. Anthropic closed the enterprise gap faster and then used the revenue to fund infrastructure plays the other labs simply can’t match. The SpaceX agreement for the entire Colossus 1 supercomputer, 220,000-plus NVIDIA GPUs, 300 megawatts of compute, closed earlier this month. That covers the bridge before AWS and Google Cloud come fully online. NewsTrackerToday broke that deal down the day it landed, and the conclusion held: this is semiconductor-era vertical integration, not 2010s SaaS playbook.
The risks aren’t on the balance sheet. They sit upstream, in capability. Anthropic has to keep shipping frontier-class models that hold up against whatever Google rolls out at I/O on Monday and whatever OpenAI does next. Stall for a quarter and the price-to-product gap opens fast. Regulation piles on top. The US Commerce Department now requires pre-deployment evaluation on every major model release, which stretches the runway between announcement and money in the door.
So where does that leave the rest of 2026? Anthropic walks into the back half of the year with the deepest enterprise pipeline among AI-native vendors, a compute footprint that matches the hyperscalers, and enough fresh capital to commission its own infrastructure instead of renting it. The real question is no longer whether $900 billion holds in May. It is whether the entire pricing model for AI businesses is about to break free of software-style multiples and start moving with compute economics. News Tracker Today has been pushing that argument since the Colossus paperwork leaked. If this round closes on the reported terms, the argument stops being a thesis and becomes the new default.
