Tech
tech
Jon Keegan
7/22/25

OpenAI and SoftBank’s $500 billion AI data center “Stargate” stumbles

On the second day of his second term, President Trump stood alongside SoftBank CEO Masayoshi Son, Oracle founder Larry Ellison, and OpenAI cofounder and CEO Sam Altman to announce the $500 billion “Project Stargate,” which Trump said included “the construction of colossal data centers, very, very massive structures.”

While it sounded like SoftBank’s Son was ready to “immediately start deploying $100 billion with a goal of making $500 billion within the next four years,” the company started looking for a $16.5 billion loan to cover the investment. And Trump’s chaotic tariff announcements have reportedly delayed SoftBank’s initial $100 billion investment, amid chatter of a possible oversupply of AI data centers.

Now The Wall Street Journal reports that there are deeper problems with Stargate, and the plans may be significantly scaled back. Apparently the organization is not fully formed yet, according to an Oracle executive on an investor call last month, and the Journal reports that Stargate has “yet to complete a single deal for a data center.”

OpenAI is moving forward on its own, including a massive data center under construction in Abilene, Texas, that the company is applying the “Stargate” name to, though it technically is not part of the SoftBank-OpenAI partnership.

Bloomberg is reporting that OpenAI’s relationship with Oracle is bearing more fruit. Oracle is reportedly supplying OpenAI with 2 million of Nvidia’s top-of-the-line GB200 GPUs as they develop 4.5 gigawatts of additional AI data center capacity in Texas, Michigan, Wisconsin, and Wyoming.

While it sounded like SoftBank’s Son was ready to “immediately start deploying $100 billion with a goal of making $500 billion within the next four years,” the company started looking for a $16.5 billion loan to cover the investment. And Trump’s chaotic tariff announcements have reportedly delayed SoftBank’s initial $100 billion investment, amid chatter of a possible oversupply of AI data centers.

Now The Wall Street Journal reports that there are deeper problems with Stargate, and the plans may be significantly scaled back. Apparently the organization is not fully formed yet, according to an Oracle executive on an investor call last month, and the Journal reports that Stargate has “yet to complete a single deal for a data center.”

OpenAI is moving forward on its own, including a massive data center under construction in Abilene, Texas, that the company is applying the “Stargate” name to, though it technically is not part of the SoftBank-OpenAI partnership.

Bloomberg is reporting that OpenAI’s relationship with Oracle is bearing more fruit. Oracle is reportedly supplying OpenAI with 2 million of Nvidia’s top-of-the-line GB200 GPUs as they develop 4.5 gigawatts of additional AI data center capacity in Texas, Michigan, Wisconsin, and Wyoming.

More Tech

See all Tech
tech

Nebius soars after signing a 5-year deal with Microsoft to supply nearly $20 billion worth of AI computing power

Artificial intelligence infrastructure group Nebius jumped more than 50% in early trading on Tuesday after the company announced after the close on Monday a major deal to supply computing power for Microsoft’s AI operations.

Under the agreement, Nebius — which rose from the ashes of Russian tech giant Yandex — will provide Microsoft “access to dedicated GPU infrastructure capacity in tranches at its new data center in Vineland, New Jersey over a five-year term.” The New Jersey data center has a capacity of 300 megawatts. The total contract value through 2031 is $17.4 billion, though, if further capacity is required, the contract value could rise to $19.4 billion.

The deal represents a sizable portion of Microsofts proposed annual capital expenditure on AI, which is expected to reach $120 billion by the end of fiscal 2026.

Nebius and competitor CoreWeave are both on the short list of startups that Nvidia has invested in. Nvidia’s small stake in the former is now worth about $120 million.

Under the agreement, Nebius — which rose from the ashes of Russian tech giant Yandex — will provide Microsoft “access to dedicated GPU infrastructure capacity in tranches at its new data center in Vineland, New Jersey over a five-year term.” The New Jersey data center has a capacity of 300 megawatts. The total contract value through 2031 is $17.4 billion, though, if further capacity is required, the contract value could rise to $19.4 billion.

The deal represents a sizable portion of Microsofts proposed annual capital expenditure on AI, which is expected to reach $120 billion by the end of fiscal 2026.

Nebius and competitor CoreWeave are both on the short list of startups that Nvidia has invested in. Nvidia’s small stake in the former is now worth about $120 million.

President Trump hosts tech executives and their guests to a dinner at the White House in the Oval Office.

Here are the Trump ties among the tech leaders who had dinner at the White House

Many of the attendees have donated to, vocally supported, or even worked for the president.

tech

Tesla’s EV market share declined to 38% in August

In August, Tesla’s share of the US EV market fell to 38%, according to new data from Cox Automotive reported by Reuters. Tesla’s market share fell below 50% for the first time last year, as competitors’ EVs began hitting the market. Now, as Tesla’s own sales slip more drastically than they had last year, it’s giving up even more ground. Tesla’s market share fell from 48.7% in June to 42% in July to 38% in August, according to Reuters. That slide has come even as buyers rushing to take advantage of the federal tax credit that ends this month provide a near-term boon for sales at Tesla and other EV makers.

$115B

OpenAI now expects to burn around $115 billion through 2029 — a full $80 billion higher than the company had previously estimated, The Information reports.

Just how much is that? It’s roughly equivalent to:

Fortunately for OpenAI, which is raising money at a $500 billion valuation, its revenue is also growing faster than expected. The ChatGPT maker now expects to make $13 billion in revenue this year and $200 billion in 2030.

Latest Stories

Sherwood Media, LLC produces fresh and unique perspectives on topical financial news and is a fully owned subsidiary of Robinhood Markets, Inc., and any views expressed here do not necessarily reflect the views of any other Robinhood affiliate, including Robinhood Markets, Inc., Robinhood Financial LLC, Robinhood Securities, LLC, Robinhood Crypto, LLC, or Robinhood Money, LLC.