Welcome to the newest edition of ID Tech’s AI Update. Here’s the latest big news on the shifting landscape of AI and identity technology:
DeepSeek is reportedly getting ready to launch the latest version of its coding AI model, dubbed V4, in mid-February. Word is that it may actually outperform OpenAI and Anthropic models on complex coding tasks, which would be a very big deal, given the amount of value priced into agentic coding platforms like Claude Code. China-based DeepSeek makes open source models, and its launch of a highly capable chatbot last January was extremely disruptive in market valuations of proprietary AI systems.
Bearing that in mind: Anthropic is in talks to raise about $10 billion at a $350 billion pre-money valuation, with GIC and Coatue expected to lead. The funding round, expected to close within weeks, would almost double Anthropic’s valuation from four months ago. Anthropic had recently hired Wilson Sonsini to prepare for an IPO that could take place as early as 2026, and has reported particularly strong enterprise demand for Claude Code.
Meanwhile, Anthropic and OpenAI rival xAI has raised $20 billion in an upsized Series E funding round, helping to fuel its own AI scaling efforts – which are apparently pretty expensive, given that Elon Musk’s AI company reportedly spent $7.8 billion in cash in the first nine months of 2025. And there are other headaches: On the X social platform in which xAI’s Grok model is integrated, the system recently acceded to user requests to generate unclothed images of individuals, including minors. xAI has now restricted Grok’s image generation/editing on X to paying subscribers as a result.
AI-focused cloud provider Lambda is in talks to raise $350 million in a pre-IPO round led by Mubadala Capital, with a market debut reportedly planned for the second half of this year. The news comes after Lambda disclosed a “multi-billion-dollar” agreement giving Microsoft access to tens of thousands of GPUs, including Nvidia GB300 NVL72 systems, toward the end of last year.
Meta has signed a set of nuclear energy deals with Vistra, Oklo, and TerraPower that could supply up to 6.6 gigawatts of “new and existing” nuclear power by 2035 to support its growing AI and data center needs. In announcing the agreements, Meta called its efforts “one of the most significant corporate purchasers of nuclear energy in American history.” The structure mixes nearer-term supply (Power Purchase Agreements from operating plants with uprates starting in late 2026) with longer bets on advanced reactors, as the Oklo and TerraPower timelines are centered in 2030–2035).
Chinese authorities are preparing to approve the import of Nvidia’s H200 chips as soon as this quarter, after the White House reversed trade restrictions that had been meant to keep the advanced chips out of the country. Still, the Chinese government is wary. A limited number of imports will be approved for commercial use, while use of the chips in military applications, sensitive government bodies, and critical infrastructure will be prohibited. Beijing has also asked some Chinese tech companies to halt orders for H200 chips and is expected to mandate domestic AI chip purchases.
The chatbot’s take: We asked GPT-5.2 for some strategic market analysis:
–
By Alex Perala
