The Great AI Capital Divide

A consequential inflection is emerging in the U.S.–China AI competition: America’s technological advantage is increasingly supported by an unprecedented mobilization of financial capital, while China’s expanding AI capabilities confront a sharp deterioration in domestic technology valuations.

Three developments reported on October 26–27 substantiate this shift. None establishes superior returns or a change in technological leadership alone — but together, they materially refine the thesis.

$97.5 billion — Meta. On October 27, Meta increased its proposed fundraising to $97.5 billion, including ~$37.5 billion in current offerings and up to $60 billion through a future share-sale program not yet launched. Nvidia separately announced a $3 billion investment in Meta.

$63 billion — Anthropic. On October 26, Bank of America, Citi, Morgan Stanley and Blackstone announced financing to purchase Broadcom AI equipment and lease it to Anthropic. About $43.5 billion would comprise senior debt, with Broadcom providing residual-value support to lenders.

40% — China’s STAR50. Reuters reported on October 26 that the index had declined ~40% from its July peak, eliminating roughly $480 billion and threatening equity fundraising for semiconductor and AI companies.

American AI companies are extending access to institutional financing while Chinese technology companies face increasingly difficult equity-market conditions.

American AI investment is increasingly supported by an interconnected system of equity issuance, structured financing, bank lending, private capital and semiconductor-supplier guarantees. This architecture can accelerate construction and distribute financing — but it introduces new vulnerabilities.

When chipmakers help finance their customers, commercial revenues become connected to those customers’ financial capacity. When AI equipment secures large loans, lenders become exposed to hardware depreciation, obsolescence and utilization risk.

The AI Supercycle is moving from a corporate investment cycle toward a broader capital-market financing cycle. The $97.5 billion and $63 billion figures are financing arrangements, not realized revenues, and should not be treated as directly additive measures of infrastructure spending.

China’s progress in competitive models and inference optimization remains significant. Yet its domestic capital markets are increasingly questioning whether that progress will produce attractive shareholder returns, with Reuters reporting substantial corrections across semiconductor and robotics companies and high valuations still a concern after the selloff.

This qualifies the diffusion thesis: lower-cost intelligence can expand adoption without generating corresponding profits for model developers, chipmakers or infrastructure suppliers. China’s challenge is no longer simply to produce competitive intelligence at lower cost — it is to finance its diffusion profitably while public-market investors demand stronger returns.

A fourth development reinforces the systemic implications. On October 27, the Bank of Korea warned that weaker American AI infrastructure spending could reduce South Korea’s 2027 semiconductor exports by $50 billion–$100 billion in a pessimistic scenario, and GDP growth by ~0.4 percentage point. These are conditional projections, not an observed slowdown.

The transmission mechanism is clear: AI financing conditions → infrastructure investment → semiconductor orders → Asian industrial revenues → national growth and capital flows. The Supercycle is increasingly influencing credit allocation, cross-border investment, semiconductor demand and macroeconomic vulnerability.

Our central thesis remains intact: America leads the AI frontier; China challenges the economics and diffusion of intelligence. But a third dimension is becoming increasingly consequential.

Five Signals to Watch

• Meta: actual proceeds raised, shareholder dilution and subsequent free-cash-flow performance.
• Anthropic: financing costs, utilization commitments and semiconductor residual-value guarantees.
• China: recovery in AI equity valuations, domestic capital formation and technology-sector profitability.
• Inference economics: whether Chinese models sustain advantages in cost per completed commercial workload.
• Global transmission: semiconductor orders, Korean memory exports and credit stress among leveraged AI infrastructure providers.

Today’s developments materially strengthen the financial dimension of our U.S.–China AI thesis without overturning the technological assessment. America is extending its lead in access to institutional capital, but taking on larger financing commitments. China continues challenging the economics of intelligence while facing a harder environment for converting progress into shareholder value. The next rotation will reward companies whose access to capital, infrastructure advantages and AI adoption translate into sustainable returns on invested capital.

America leads in frontier intelligence and capital mobilization. China challenges the cost of intelligence. The next AI Supercycle will be decided by who can finance growth without sacrificing returns.