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:
Anthropic’s annualized revenue run rate is approaching $7 billion, up from about $5 billion in August. The OpenAI rival is reportedly targeting $9 billion ARR by the end of this year, and $20-26 billion in 2026. While OpenAI still has an overwhelming lead in its sheer number of end users, Anthropic is enjoying growing enterprise uptake, thanks in large part to its Claude Code agentic coding platform.
Beijing-based Cambricon Technologies saw a 14-fold jump in quarterly revenues, which amounted to RMB 1.73 billion, reflecting a RMB 567 million profit. The chipmaker is riding the tide of the China-US AI race, which has seen a range of trade restrictions on both sides. The Chinese government has been pressuring AI firms to turn away from Nvidia chips and instead source their stuff domestically, which has obviously been a boon for Cambricon.
In Taiwan, TSMC beat Q3 estimates and raised its revenue outlook for 2025 to mid-30 percent. Q4 alone is expected to deliver $32.2–$33.4 billion in revenues. The news suggests that the ‘AI megatrend’ is ongoing, TSMC being the world’s biggest foundry for major clients like Nvidia, AMD, and Apple. TSMC has also announced an expansion to its capital expenditures, to further boost its upward trajectory.
CoreWeave and Poolside have announced a 2-gigawatt AI data center project in West Texas, dubbed “Project Horizon”. That’s roughly the output of the Hoover Dam. And they will be generating their own electricity via on-site gas turbines, a feature expected to ensure the steady availability of energy despite whatever may be happening on the local grid. CoreWeave is a major American cloud computing company, while Poolside is a Paris-based coding-focused AI startup.
Meta, meanwhile, has committed over $1.5 billion to an AI-focused data center project in El Paso that will scale to 1 gigawatt. Facilities will be designed for high power density, advanced liquid cooling, and high-bandwidth networking, and Meta says it will match 100 percent of energy use with renewables and pursue water-positive operations, aiming to restore twice what it consumes. The facility is expected to come online in 2028.
Google has announced $24 billion in AI infrastructure spending to be split between the US and India over the next several years. $9 billion will be used to expand South Carolina data centers through 2027, while $15 billion is being committed to building a gigawatt-scale AI hub in Visakhapatnam over the next five years. The latter project will also involve work on subsea fiber-optic cabling.
OpenAI is officially allowing estates of deceased public figures to block the use of their likeness on Sora, its video generation platform. There were some complaints from the estate of Martin Luther King Jr. after “disrespectful” videos surfaced. The recently-launched Sora 2 system has captured considerable attention thanks in part to its extremely convincing likenesses (or deepfakes, if you prefer) of users and public figures.
A 17-year-old girl from New Jersey has filed a lawsuit against the maker of an AI app that purports to digitally remove clothes from people’s photos. Its ClothOff app was allegedly used to make a nude picture from a photo the teen took when she was 14. The plaintiff is being supported by Yale Law School affiliates, and her lawsuit essentially accuses AI/Robotics Venture Strategy 3 Ltd. (BVI) of creating child sexual abuse material, with Telegram – the platform where the images were distributed – also named as a defendant.
The chatbot’s take: A bit of clarity on all the high-energy activity from Big Tech:
–
By Alex Perala
