Ex-Google CEO Eric Schmidt says the true constraint on AI growth is capital, not energy, estimating 10 gigawatts of compute capacity could cost half a trillion dollars or more.
Summary:
- Schmidt argued the binding constraint on AI development is capital availability rather than energy supply, with costs estimated at around $50 billion per gigawatt
- At that rate, 10 gigawatts of compute capacity would cost approximately half a trillion dollars, a sum only a handful of countries or companies could realistically finance
- China is identified as capable of mobilising that level of capital, though Schmidt said he was uncertain whether it is doing so
- The US capital market’s ability to borrow at scale is cited as a structural advantage that makes large AI financing possible
- Europe is described as unable to finance AI infrastructure at the required scale, a limitation Schmidt acknowledged is a source of frustration on the continent
Former Google chief executive Dr. Eric Schmidt has argued that the artificial intelligence industry is approaching a financial barrier that may prove more restrictive than the energy constraints that have dominated discussion of AI’s growth limits.
Speaking publicly, Schmidt laid out a straightforward cost argument. At roughly $50 billion per gigawatt of compute capacity, building out 10 gigawatts would require approximately half a trillion dollars of capital. Sustaining or scaling that further pushes the figure toward a trillion dollars or beyond, a level of investment that very few nations or private actors could plausibly deliver.
That framing shifts the competitive dynamic away from questions of electricity generation and infrastructure toward questions of financial depth and access to capital markets. In Schmidt’s view, the countries and institutions best placed to win the AI race are those that can mobilise and sustain investment at that scale, and the field is narrow.
He identified China as one of the few actors with the financial capacity to mount that kind of effort, though he stopped short of confirming whether Beijing is actively pursuing it at that magnitude. The US, he argued, holds a structural advantage through the depth and sophistication of its capital markets, which allow the kind of large-scale borrowing required to finance AI infrastructure. That ability to leverage private capital at scale is, in his telling, a genuine competitive differentiator.
Europe, by contrast, does not have access to comparable financing mechanisms, a reality Schmidt acknowledged is a source of frustration among European policymakers and industry figures. The implication is that without a structural shift in how European capital markets function or how governments are willing to deploy public funds, the continent risks being priced out of the front rank of AI development regardless of its technical talent or regulatory ambitions.
The comments add a financial lens to a debate that has largely focused on compute, energy and regulation as the primary constraints shaping the AI landscape.
—
Schmidt’s framing of capital availability as the binding constraint on AI development has direct implications for how markets should think about the AI infrastructure buildout. If the effective ceiling is financial rather than physical, then the competitive advantage shifts decisively toward deep-pocketed sovereigns and economies with sophisticated capital markets, reinforcing the US and China duopoly narrative. European AI and tech equities face a structural headwind if the continent genuinely cannot mobilise capital at the required scale
