OpenAI signs record $105B Ohio data center lease with Nvidia backing

The Ohio data center deal signals a critical transition in how AI infrastructure constraints are perceived. For years, the bottleneck in artificial intelligence expansion centered on chip availability and production capacity. Now, as Huang frames it through the "LPS" acronym, the limiting factors have shifted to the physical fundamentals: securing land with sufficient power supply and constructing the buildings themselves. This shift matters because it explains why Nvidia is willing to stake $105 billion on residual asset value rather than on chip sales alone.

The broader context the Wall Street Journal uncovered—that nine tech companies hold $3 trillion in off-balance-sheet AI commitments—underscores how rapidly these infrastructure buildouts have outpaced traditional corporate financing models. Leases and purchase commitments don't appear on balance sheets until payments begin or goods are delivered. This opacity has grown sharply: leases not yet started jumped to $1.2 trillion (four times the prior year), and Alphabet's purchase commitments alone nearly tripled in three months, from $332 billion to $811 billion. For investors accustomed to parsing balance sheets, this hidden debt load makes it harder to assess the true financial health and leverage of major technology firms.