OpenAI compute tab seen at $614 billion, not $750 billion

The $614 billion estimate rests on an assumption Bernstein itself calls unlikely: that every identified contract proceeds on schedule. Construction delays, cancellations and flexible hardware purchases could reduce or defer payments, so the true figure may land lower or later than the headline number suggests. That distinction is central for investors worried that AI developers have promised more computing capacity than their revenues can support.

OpenAI's revenue growth is the other side of the ledger. It reportedly reached nearly $50 billion in annualized revenue in September, growing 77% year over year, which could help fund the expanding infrastructure requirements — though projections remain uncertain. The provider mix also matters: OpenAI accounts for less than half of Oracle's remaining performance obligations, and Microsoft's diversified cloud operations could cushion delays better than smaller infrastructure specialists.

The bigger risk may arrive later, when computing shortages ease. That could expose speculative construction projects and providers dependent on expensive financing, while existing contracts with established cloud companies appear more resilient than future demand for specialized computing rentals.