From left: President Donald Trump, Oracle cofounder Larry Ellison, SoftBank Group CEO Masayoshi Son, and OpenAI CEO Sam Altman in the Roosevelt Room of the White House in Washington, D.C., on Jan. 21, 2025. (Photo: Aaron Schwartz/Sipa/Bloomberg/Getty Images)

President Trump on Tuesday announced an eye-popping investment in artificial intelligence infrastructure in the U.S., funded through a joint venture called the Stargate Project. 

According to Trump, the investment will be accompanied by a spate of executive orders to ensure new data centers built in connection with the investment will have enough energy. 

Trump was joined by Oracle founder Larry Ellison, OpenAI CEO Sam Altman, and SoftBank CEO Masayoshi Son, who said the investment would start with $100 billion, plus a goal of $500 billion over the course of four years.

The project’s initial equity funders are SoftBank, OpenAI, and Abu Dhabi-based investment partner MGX. OpenAI will have operational responsibility and Son will be the chairman of Stargate. Arm, Microsoft, Nvidia, Oracle, and OpenAI are the project’s “key initial technology partners.”

Ellison said the applications produced through advancement in AI would revolutionize health care, offering the example of early cancer detection and the potential for a cancer vaccine. 

The Oracle founder added that Stargate already had 10 data centers under construction and expects to build infrastructure in areas beyond its first location in Abilene, Texas. —Amanda Gerut

Meta, which has just very publicly renounced censorship, has found itself at the center of a censorship storm.

Many Instagram users inside and outside the U.S. found this week that searches for “#democrat” were blocked, while those for “#republican” were not. Searches for terms such as “#project2025”—the right-wing blueprint for President Trump’s second term—were also blocked. 

In all of these cases, Instagram claimed it had hidden the results because they “may contain sensitive content.”

Given Meta’s current keenness to ingratiate itself with Trump, people on social media were naturally pretty outraged. But the company claims it was all a terrible mistake.

“We're aware of an error affecting hashtags across the political spectrum and we are working quickly to resolve it,” it told the BBC, which noted that searches for “#republicans” in the plural were also being blocked. And indeed, the issue did seem to be clearing up yesterday.

But even if it was a technical error, the incident gives a good indication of the reputational problems that Trump-aligned social media firms can expect over the next four years. —David Meyer

The ascendant French artificial intelligence company Mistral is “not for sale,” its founder and CEO said Tuesday.

Arthur Mensch told Bloomberg TV that the two-year-old company, which was founded by veterans of Google DeepMind and Meta, was valued last year at €5.8 billion (or about $6 billion), is taking steps toward an initial public offering. 

“Of course that’s the plan,” he said from the World Economic Forum in Davos, Switzerland. In the meantime, Mensch added, the company will open a Singapore office to serve the APAC region and focus on growing its footprint in the U.S. and its native Europe.

Mistral is best known for its generative AI model and “Le Chat” chatbot, but its recognition trails better-funded U.S. rivals like Anthropic and OpenAI. 

Not that Mensch is worried. Mistral is backed by a who’s who of Silicon Valley investors including General Catalyst, Andreessen Horowitz, and Lightspeed Venture Partners. “We have plenty [of money],” he told Bloomberg. —AN