Key Takeaways:
OpenAI's plan to spend more than $30 billion on a new data center shows the staggering scale of capital deployment reshaping the global technology industry.
OpenAI plans to invest more than $30 billion in a new data center, the latest escalation in an AI infrastructure arms race that has pushed industry capital spending up 600% since 2020.
"The AI industry's financial architecture depends on a single company continuing to exist," Ed Zitron, a technology critic and author of the newsletter Where's Your Ed At, said. He warned that OpenAI's failure would function as a market-shaking collapse comparable to the fall of Lehman Brothers.
The company intends to spend more than $50 billion on compute this year and has made roughly $748 billion in performance obligations to Microsoft, Amazon, and Oracle, according to Zitron's analysis. OpenAI is also carrying a $122 billion funding round that has not fully closed, with SoftBank Group contributing $30 billion in tranches — the third of which is due Oct. 1, 2026. The company posted a net loss of $38.5 billion in 2025 on $13.07 billion in revenue.
The spending spree has created a web of financial dependencies that extends far beyond OpenAI. Oracle has committed more than $340 billion to build data center capacity for the AI lab as part of a $300 billion compute contract, and has seen its credit rating cut to the lowest investment-grade level by S&P Global, with OpenAI named as a key credit risk.
The data center buildout is part of a broader transformation in how the technology industry allocates capital. Between 2020 and 2025, data center capital expenditures grew by 600%, surpassing upstream oil and gas investment and matching total renewable energy spending, according to Rystad Energy. Almost 60% of senior energy executives name AI and data centers as one of the most significant power challenges over the next five to 10 years, a Reuters report found, with nearly 80% expecting data centers to consume more than 11% of total U.S. electricity by 2030.
Who Wins, Who Loses
The beneficiaries of this spending cycle are concentrated among a handful of chipmakers, cloud providers, and infrastructure operators. Broadcom has signed multiple long-term agreements with Google, Meta Platforms, Anthropic, and OpenAI, including a contract for a 1.3 gigawatt deployment in 2027 as part of a 10 GW agreement through 2029 with OpenAI. The company's AI semiconductor revenue surged 143% year-over-year to $10.8 billion in the second quarter of fiscal 2026, with more than $30 billion in AI semiconductor bookings against $10.8 billion in shipments.
Marvell Technology, which designs chips for data center networking and custom silicon, has seen its data center segment revenue climb 27% year-over-year to $1.8 billion. The company now expects full-year revenue to grow 40% to nearly $11.5 billion, driven by demand for 800G connectivity products and 1.6-terabit solutions that move data across thousands of processors inside AI clusters.
Nvidia, the dominant supplier of AI training chips, has deployed cross-investments across the ecosystem, including a $2 billion stake in Nebius, an Amsterdam-based AI cloud company, and participation in OpenAI's $110 billion funding round announced in March. Some analysts have flagged what Goldman Sachs called "the increasing circularity of the AI ecosystem," in which a concentrated group of chip companies, cloud providers, and AI labs finance one another's buildouts.
The Risk of Overbuild
The scale of OpenAI's financial exposure has drawn scrutiny. The company submitted a confidential IPO filing with the Securities and Exchange Commission last month at an $852 billion valuation, with Goldman Sachs and Morgan Stanley leading the process. Advisers have warned that a $1 trillion valuation — which Chief Executive Officer Sam Altman has called a minimum — may not be achievable in current market conditions, and the company is leaning toward delaying its public offering until 2027.
Zitron argues that a payment stoppage to infrastructure partners such as Oracle and CoreWeave would leave those companies without the cash flow needed to meet their own debt commitments. "Should it fail, the reverberations would mark a turning point — the AI era's Lehman Brothers moment, closing one chapter of economic history and violently opening the next," he wrote.
For investors, the question is whether the current spending cycle generates measurable returns. Broadcom trades at a forward P/E that reflects its AI exposure, while Marvell's projected EPS growth of 42% in fiscal 2027 and 67% in fiscal 2028 suggests the market is pricing in continued demand acceleration. But if OpenAI's financial position deteriorates, the ripple effects would hit every layer of the AI infrastructure stack — from chip designers to cloud operators to power utilities — making the $30 billion data center bet a bellwether for the entire industry.
This article is for informational purposes only and does not constitute investment advice.
