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:
The White House and AI titans have jointly announced a $500 billion AI infrastructure project. OpenAI, Oracle, SoftBank, Microsoft, and the United Arab Emirates’ VC arm MGX will all pitch in, with $100 billion to be deployed immediately, and the rest to follow over the next four years. The goal is to establish massive AI data centers in the U.S., and the project’s backers say it will create 100,000 jobs.
Separately, President Trump signed an executive order requiring the development of a new plan for American “AI dominance”. Certain agency heads, the director of the Office of Management and Budget, and tech advisers including David Sacks are called upon to advise on the plan, which seeks to promote the development of AI systems that are “free from ideological bias or engineered social agendas.” They have 180 days to put it together.
Meta plans to spend somewhere between $60 billion and $65 billion on capital expenditure this year, with the majority destined for AI data centers. The aim is to bring about a gigawatt of computing power online. According to CEO Mark Zuckerberg, a planned data center will be so big that it “would encompass a considerable portion of Manhattan”.
Google is investing another $1 billion in Anthropic, thought to be one of OpenAI’s biggest rivals thanks to its Claude AI platform. Google has already put $2 billion into the firm so far, and has a 10 percent stake. Anthropic’s annualized revenues reportedly reached about $1 billion in December of last year.
OpenAI has launched Operator, its first AI agent. The tool can perform relatively complex actions through web browsers, such as booking a hotel reservation, after a simple prompt from the user. It’s one of the pioneering products of the anticipated “agentic” era, in which AI assistants will be able to perform complex tasks autonomously. At the moment, Operator is only available to users with a Pro subscription, which costs $200 a month.
MiniMax Hailuo AI has launched a “Subject Reference” feature for its video generation model, enabling users to make videos featuring a character depicted in an uploaded image. The system shows strong character consistency across different videos, meaning the same individual can be depicted in different scenarios and attire.
China-based DeepSeek has launched an open source AI model that is competitive with some of the most advanced commercially available models, including OpenAI’s o1 “reasoning” model, across various benchmarks. DeepSeek claims it trained its new R1 model in just two months, and at a relatively low cost of $5.5 million. Its advantage to users lies in the much lower costs of its API (compared to rivals like o1), and the capability of being deployed locally on a device.
The chatbot’s take: We decided to try out the flashy new Chinese AI model.
–
January 24, 2025 – by Alex Perala
Follow Us