Welcome to the latest edition of the AI Update, where we continue to track the major developments in artificial intelligence and their implications for identity security and cybersecurity more broadly.
The Pentagon has announced deals with a number of major tech companies to bring AI to its classified military networks. Agreements have been made with Nvidia, Microsoft, Amazon Web Services, Google, OpenAI, SpaceX, Reflection, and Oracle. The deployments are meant for the Pentagon’s IL6 and IL7 environments, which refer to some of the government’s most sensitive classified network environments, and the Pentagon says the AI tools will “streamline data synthesis, elevate situational understanding, and augment warfighter decision-making in complex operational environments.” The set of deals notably comes in the wake of a very public dispute with Anthropic over how its AI tech could be used by the US military.
Nevertheless, Google is reportedly committing to invest up to $40 billion in Anthropic, with $10 billion upfront and the remainder tied to performance milestones. And Amazon is committing up to $25 billion to Anthropic. On the AI front, these companies are all competitors. But the investments are tied to cloud infrastructure, with Anthropic agreeing to purchase compute resources from Google and Amazon as it continues to progress with its own AI models – so even if it stays near the front of the pack in the AI model race, Google and Amazon will benefit.
Huawei now expects AI chip revenue of about $12 billion in 2026, which would reflect a 60 percent increase year-over-year. The surge is currently being driven by demand for its new Ascend 950PR chip, which entered mass production last month; but it also reflects the impact of US sanctions restricting the availability of Nvidia chips in China, where Huawei is based. Also helpful has been the recent launch of DeepSeek V4, the latest open-weights model from an AI lab that has been particularly strong in playing catch-up with the closed models of Western firms like OpenAI and Anthropic.
Elon Musk said he was “a fool” to support the founding of OpenAI during questioning in the first week of the jury trial for his lawsuit against the latter. Elon says he donated about $38 million to help found the firm, believing it to be a nonprofit dedicated to open source AI, and now that it’s a massively profitable firm, he says OpenAI’s co-founders managed to “steal a charity”. The defence argues that Elon was aware of potential for-profit structures for the firm in the early days, and when confronted with a 2017 document discussing such a shift, he said he “didn’t read the fine print.”
SoftBank is reportedly getting ready to launch a US-based AI and robotics firm as early as this year. Dubbed “Roze”, it will apparently involve using robots to build out AI infrastructure, such as data centers, and it could consolidate some of SoftBank’s current energy, land, and infrastructure projects. The news comes after SoftBank agreed in October to acquire ABB’s robotics business for $5.375 billion. It’s now reportedly aiming for a $100 billion valuation for Roze.
Meanwhile, Meta has acquired Assured Robot Intelligence, a robotics AI startup focused on building the software “brains” for humanoid robots. Financial terms of the deal have not been disclosed, but the startup’s co-founders, Lerrel Pinto and Xiaolong Wang, are joining Meta Superintelligence Labs and will work with Meta Robotics Studio. Meta has established a new division within its Reality Labs arm to develop AI-powered humanoid robots, led by Marc Whitten, the former CEO of Cruise, earlier this year.
The FIDO Alliance has formed an Agentic Authentication Technical Working Group to define how AI agents authenticate, act, and transact on behalf of users, with Google contributing its Agent Payments Protocol and Mastercard contributing its Verifiable Intent framework. FIDO is not alone: NIST, the IETF, and the Agentic AI Foundation are all working on overlapping pieces of agent identity, delegation, authorization, and interoperability. FIDO says some analysts estimate agentic commerce could reach $5 trillion globally by 2030, which helps explain why payment networks are moving early.
China has officially blocked Meta’s acquisition of the agentic AI startup Manus, which was pretty much considered a done deal until the government stepped in to announce a regulatory investigation, and blocked Manus’s co-founders from leaving the country. In its final decision ordering the deal’s unwinding, the National Development and Reform Commission said it has decided to “prohibit foreign investment in the Manus project in accordance with laws and regulations”.
The ID Tech team is not worried about any such issues:
–
By Alex Perala
Follow Us