OpenAI CEO Sam Altman has publicly pointed to the emergence of a structural imbalance between supply and demand in the artificial intelligence hardware market. According to him, a significant portion of chip manufacturers are building their business models around selling products to a narrow circle of buyers—OpenAI, Anthropic, and other leading AI labs—yet the aggregate order volume of these companies may prove insufficient to utilize all planned manufacturing capacity. Altman suggested that while demand for compute and data centers will continue to grow in the long term, supply risks outpacing actual buyer needs in the short horizon.

“We Have a Limited Amount of Demand”

“A lot of people are making chips right now. And a lot of those people think they are going to sell a significant portion of them to us, Anthropic, and others. But we have a limited amount of demand and data centers. And I think a lot of the chipmakers are now negotiating with component suppliers assuming we are going to buy all of their chips,” Altman stated. He emphasized that multiple manufacturers are simultaneously relying on the exact same future orders, even though the actual needs of end-buyers are finite. This concentration of expectations on a narrow circle of clients, in his view, creates the risk that some manufactured products will not find buyers in the foreseeable future.

Who Sells and Who Buys: Market Figures

The scale of planned investments in AI infrastructure remains staggering. In August, Nvidia CEO Jensen Huang announced that OpenAI had committed to deploying 12 GW of compute capacity by 2030. Nvidia’s financial results confirm the scale of the boom: company revenue for the second quarter reached $96.22 billion, rising to $107.6 billion in the third quarter; AI companies funded by Nvidia resources are estimated to account for about a quarter of its total revenue next year. AMD, in turn, has deployed Helios solutions to power workloads for Anthropic, Meta, Microsoft, OpenAI, and Oracle. Company CEO Lisa Su confirmed that demand for AMD products exceeds supply and announced plans to increase revenue by over 80% by 2027. Micron has contracted over 75% of its supply for fiscal 2027 and reported a 379.3% increase in revenue for the fourth quarter. Optical component manufacturer Coherent has virtually all of its 2027 manufacturing capacity already reserved, with customers now booking orders for 2028.

Financial Risk Falls on Infrastructure Operators

The most vulnerable link in the chain turns out to be data center operators who shoulder the financing of infrastructure construction and operation. CoreWeave’s planned capital expenditures for 2026 range from $37 billion to $39 billion; in the third quarter, the company’s debt interest expenses reached $846 million, with a net loss of $709 million. CoreWeave CEO Michael Intrator stated that the company does not intend to grow its business at any cost. Since the beginning of the year, CoreWeave shares have fallen by 33.12%. A similar dynamic is observed at IREN, a neocloud provider whose net debt rose from $216 million to $4.43 billion, while its share price dropped by 42.39% in a month. According to sources, IREN investors have shifted their focus from contract sizes to the company's ability to fulfill obligations, secure funding, and generate profit from infrastructure operations.

What This Means for the AI Boom

Altman’s warning does not mean that demand for compute power is exhausted; on the contrary, all key suppliers are reporting supply deficits in the current cycle. However, his statement points to a systemic risk: if multiple chipmakers simultaneously bake sales to the same narrow circle of buyers into their production plans, any slowdown in the pace of AI infrastructure deployment will leave a portion of capacity underutilized. For end-consumers—AI labs—this could mean temporary hardware price drops as well as the need to renegotiate long-term contracts. For suppliers and data center operators burdened with tens of billions of dollars in debt obligations, such a scenario carries direct financial risks that have already been reflected in the stock quotes of CoreWeave and IREN.