New details emerge about Project Stargate as Meta discloses $60B+ AI investment – News

New details have emerged about Project Stargate, the $500 billion initiative through which OpenAI hopes to build a cluster of artificial intelligence data centers.

The Financial Times reported on Thursday that the project currently lacks a fully developed plan. The initiative’s funding has not been finalized either, sources told the publication.

OpenAI, SoftBank Group Corp. and Oracle Corp. unveiled Project Stargate during a Tuesday press briefing at the White House. The companies intend to build an expansive network of data centers that will be optimized to run AI workloads. OpenAI will be the sole user of the data centers, the Financial Times’ sources said.

A few hours after it was announced, Project Stargate drew criticism from Elon Musk, who is currently suing OpenAI for alleged breach of contract. Musk also runs a competing AI startup called xAI Corp. that develops large language models. He stated in a post on X that SoftBank, one of Project Stargate’s main backers, currently lacks the funds necessary to cover the project’s hefty budget. 

“They don’t actually have the money,” Musk wrote. “SoftBank has well under $10B secured.” The post drew a quick response from OpenAI Chief Executive Officer Sam Altman, who replied “Wrong, as you surely know.”

According to the Financial Times’ Thursday report, OpenAI and SoftBank will initially each commit about $15 billion to Project Stargate. That’s a fraction of the up to $500 billion they plan to spend on the new data centers. The companies reportedly plan to cover the rest of the project’s cost by raising equity funding from “existing investors” as well as debt financing.

The project won’t receive any government funding, the paper’s sources added.

Separately, Bloomberg on Thursday revealed new details about the data centers that OpenAI and its partners are building. Citing regulatory filings, the publication reported that 10 Project Stargate facilities are currently under construction in Abilene, Texas. OpenAI told the paper that its 875-acre data center campus in the city is “the first of many we will be building across the country.”

OpenAI is not the only AI provider that announced a large data center investment this week. Meta Platforms Inc., which develops the Llama series of open-source LLMs, disclosed today its capital expenditures will reach between $60 billion and $65 billion this year. Most of the capital will go towards building data centers and hiring AI experts, Chief Executive Officer Mark Zuckerberg wrote in a Facebook post.

The initiative will see the company add one gigawatt’s worth of computing capacity over the next 12 months. One gigawatt is enough to power about 700,000 households. By the end of the year, Meta expects to have more than 1.3 million graphics processing units running in its data centers.

Zuckerberg also outlined the company’s AI software roadmap. Later this year, Meta plans to launch a new addition to its open-source LLM series called Llama 4. Internally, the company will roll out an AI-powered programming assistant that is expected to contribute “increasing amounts of code” to the company’s software projects.

Image: Oracle