NEW YORK — The AI funding market, already running hot, accelerated into a different gear this week. Anthropic closed a round that values the Claude maker at $965 billion — eclipsing OpenAI on paper and positioning the company as the most valuable private AI lab on earth. The company simultaneously confirmed that its next flagship model, internally called Mythos, is in the pipeline. At that valuation, Anthropic is priced at roughly 2.5 times OpenAI's last disclosed private valuation, a spread that would have seemed implausible eighteen months ago.
The week's other deals reinforce how broadly capital is now flowing. Nvidia led a $300 million round into Israeli AI startup Decart, which now carries a $4 billion valuation. The chip giant's direct equity participation — rather than a supply agreement — signals a strategic hedge: Nvidia is increasingly investing in the application layer it supplies with silicon. Meanwhile, LMArena, an AI model evaluation startup, raised $150 million at a $1.7 billion valuation — a reminder that the picks-and-shovels layer of the AI stack is now attracting institutional scale capital, not just seed checks.
The geographic dimension is equally notable. Kuaishou's video generation unit Kling closed a $2 billion deal, prompting veteran investor commentary that China's full-stack AI buildout remains systematically underpriced by Western markets. That gap may be closing: the Kling transaction is being read as a repricing event for Chinese AI assets broadly.
Against this backdrop, talent attrition at Google DeepMind draws fresh scrutiny. Several senior researchers have departed in recent months, raising questions about whether one of the field's foundational labs can maintain competitive velocity when the equity upside at startups has never been larger. DeepMind's structural constraint — operating inside a public company rather than as an independent entity — may be repricing its talent pool in real time.
For enterprise software buyers and technology operators, the aggregate signal is unambiguous: AI infrastructure is moving from experimental budget lines to core capital allocation. The valuation multiples suggest investors have already priced in that transition. The question is whether the underlying models — Mythos included — can grow into the numbers.
