Image: Zozulinskyi/Envato
OpenAI’s Foxconn deal ties US data center hardware into its $500B Stargate buildout and $1.4T spend, raising fresh questions about risk and an AI bubble.
OpenAI has tapped the iPhone manufacturing giant Foxconn to design and build artificial intelligence data center components in the United States.
The Taiwan-based company has five US factories in Wisconsin, Ohio, Texas, Virginia, and Indiana, which are primarily used for advanced specialist manufacturing and research and development. Foxconn is already rewiring much of its US manufacturing output towards data centers, aiming to become the lead supplier of components to the new facilities opening over the next decade.
For OpenAI, the deal includes access to core data center components such as power supply, cabling, networking, and cooling systems, which OpenAI and others can tailor to meet specific requirements.
The exact terms were not disclosed, which is surprising given the scale of OpenAI’s past agreements with Oracle, Nvidia, Microsoft, and AMD. OpenAI has signed roughly $1.4 trillion in spending commitments so far, prompting some investors to warn that we may be in an AI bubble.
“This partnership is a step toward ensuring the core technologies of the AI era are built here,” OpenAI CEO Sam Altman said in the announcement. “This work will strengthen US leadership and help ensure the benefits of AI are widely shared.”
Foxconn, primarily known as the iPhone manufacturer, already has a working relationship with Jony Ive, Apple’s former head of design, who is now collaborating with OpenAI on a hardware device. If that project ever reaches production, Foxconn is expected to be the main manufacturer.
This is another major deal connected to OpenAI’s Stargate venture, which seeks to build 10 gigawatts (GW) of compute capacity over the next decade, with 7 GW already planned and more than $400 billion in investment earmarked for the next three years. Stargate has already broken ground at several facilities in Texas.
Under the Stargate structure, data centers are financed by SoftBank, investment firm MGX, and a host of other investors, then managed by Oracle to run OpenAI’s services and research. OpenAI recently committed to spending $300 billion on compute from Oracle over the next decade.
While Foxconn is not a formal partner, it sits within the manufacturing pillar alongside companies like Nvidia, with the capability to produce the components needed for rapid AI data center expansion at scale.
Even with construction underway in Texas and both the Trump administration and state governors eager to cut red tape for the company, many observers are concerned that OpenAI’s financial commitments exceed what it will ever generate in profits.
Altman sought to reassure skeptics by revealing the company has surpassed a $20 billion annualised revenue run rate, but that remains a drop in the ocean compared with its obligations, and it is projected to burn through $8 billion in 2025.
At the same time, OpenAI’s exuberance is distorting the broader AI market. Rival Anthropic has reportedly been valued at around $350 billion in a recent deal with Microsoft and Nvidia, despite having a revenue run rate of about $5 billion and being far from profitable. Elon Musk’s xAI, estimated to generate roughly $3.2 billion in revenue in 2025 (only about $500 million of which will come from AI services), is reportedly close to securing a $200 billion valuation.
If OpenAI fails to deliver Stargate for any number of reasons, there is a real possibility that the rest of this house of cards could collapse, as investors reassess whether demand for AI is truly as high as expected, or whether compute costs make the entire model unsustainable.
Recent coverage of OpenAI’s $7 billion Michigan data center campus explains how Stargate investments are reshaping regional infrastructure and jobs.
David Curry is a tech journalist and analyst with over a decade of experience writing for established outlets. He holds a master’s degree in International Journalism from the University of Leeds and has covered the technology sector since the early 2010s. His work focuses on B2B technology, data journalism, mobile apps and app markets, artificial intelligence, digital platforms, and emerging technologies. He earned a BA from the University of Lincoln and an MA from the University of Leeds.
