|
|
|
Feb 24, 2026
|
|
|
|
|
Supported by
|
|
|
|
|
|
|
Happy Tuesday! Oracle shares fall after The Information's report on the struggles of the firm's Stargate joint venture with OpenAI. Anthropic CEO Dario Amodei is set to meet with Secretary of War Pete Hegseth on Tuesday. Cerebras files confidentially for a U.S. IPO.
|
|
|
|
Oracle shares were down 4.5% on Monday, after The Information reported that the cloud firm’s Stargate joint venture with OpenAI and SoftBank floundered after its flashy announcement at the White House in January 2025. Instead of the three companies pursuing a data center expansion together, OpenAI has struck individual deals with Oracle and SoftBank, though those deals fell short of OpenAI’s goal of securing commitments for 10 gigawatts of capacity by the end of 2025. OpenAI also moved last year to get compute capacity from other providers outside of the original
Stargate trio—a move that made its compute bill more expensive than it had planned. The Information reported that as part of Oracle’s 4.5 gigawatt deal with OpenAI, the firms agreed to share some of the economic risk of the expansive project. That means if there is a delay or if expenses are over budget, they would share the cost. Other tech stocks fell Monday, as investors weighed the impact of artificial intelligence on enterprise applications, which Oracle also provides.
|
|
|
|
Anthropic CEO Dario Amodei is set to meet with Secretary of War Pete Hegseth on Tuesday, following weeks of negotiations between Anthropic and the Pentagon over whether Anthropic can put limits on the Pentagon’s use of its AI. Axios first reported on the meeting. The negotiations began after Hegseth issued a memo in January calling on AI companies to allow the Pentagon to use their models for “any lawful use.” Anthropic, which won a $200 million contract with the Pentagon last summer and is the only AI company whose model is cleared for use on classified data, is asking the government
for restrictions on its model being used for mass domestic surveillance and autonomous weapons use without humans involved. As conflict between the two sides escalated, the dispute has spilled into the public. A Pentagon official told Axios last week that the department was considering declaring Anthropic a “supply chain risk,” meaning that any Pentagon contractors would have to certify that they did not use Anthropic technology in their work for the department—an unusual threat usually reserved for foreign adversaries, not an American company.
|
|
|
|
Cerebras Systems has again made a confidential filing for a U.S. public offering, according to people with knowledge of its plans. The designer of AI chips and systems has been meeting with prospective investors before a potential listing that could take place as soon as this April, they said. The company had aimed to go public last year, but announced last December it was withdrawing its IPO paperwork. Instead, in early February, Cerebras raised about $1 billion in a private funding round that valued the company at $23 billion including the new investment. It also announced that OpenAI had agreed to purchase 750 megawatts worth of computing power from the startup through 2028 for about $10 billion. The OpenAI agreement could resolve public investors’ concerns that it is overly dependent on one customer. In its
original prospectus in September 2024, it disclosed that its investor Group 42 Holding, the parent of United Arab Emirates conglomerate G42, made up a majority of its revenue. CEO Andrew Feldman late last year said it had withdrawn its earlier IPO paperwork because so it could re-file with updated financials and strategy information “including our approach to the rapidly changing AI landscape.”
|
|
|
|
IBM stock dropped 13% on Monday after Anthropic said that its Claude Code tool could be used to automate the modernization of legacy code. IBM mainframes, computers that process a large portion of the world’s financial transactions, run applications written in COBOL, a programming language developed in 1959 that is widely used by banking institutions, insurance companies and government agencies. COBOL supports 80% of in-person credit transactions, for example. Anthropic said Monday that Claude Code can be used to automate the translation of COBOL systems into modern coding
languages. The drop in IBM shares follow a selloff by other software stocks on fears new agentic AI tools will lower demand for the older software. IBM told The Information it’s been “investing in code modernization for years,” including the launch of its watsonx Code Assistant in 2024, which—similar to Claude Code—works to translate COBOL on its mainframes. Amazon also launched its own AI mainframe modernization tool called AWS Transform in 2025. Notably, IBM also has a partnership with Anthropic, announced in October, to integrate Claude into the software giant’s products.
|
|
|
|
Elon Musk said Monday that SpaceX has been cutting prices for its Starlink satellite internet service to reach more customers. He made the comments in response to an article in The Information on the topic. SpaceX has introduced cheaper plans in the U.S. and other countries in recent months, has ramped up sales and marketing efforts and has been giving hardware away for free in some instances. The changes come ahead of Amazon’s planned
launch later this year of its Leo satellite internet service, which was formerly known as Project Kuiper. In response to a post on X that linked to The Information’s story, Musk wrote: “This has nothing to do with Kuiper, we’re just trying to make Starlink more affordable to a broader audience. The lower the cost, the more Starlink can be used by people who don’t have much money, especially in the developing world.”
|
|
|
|
| |
