Alphabet raises $80 billion in equity for AI capex; GPU supply still tight

Alphabet announced plans to raise $80 billion through equity issuance to fund AI infrastructure spending. The offering includes a $10 billion private placement to Berkshire Hathaway. This is the largest equity raise by a technology company on record. The financing structure is notable: dilutive equity, not debt, not retained earnings. The hyperscalers are not waiting for return signals before committing capital.

Nvidia's CEO told Asian investors the company remains supply-constrained. Not demand-constrained. Supply. GPU allocation is still the binding constraint on model training and high-concurrency inference workloads. The capex is racing ahead of rack availability, which helps explain why Alphabet is pre-financing at scale rather than staging deployment against utilization.

HPE reported 37 percent share-price gains on booming data-center equipment demand. Micron crossed a $1 trillion market cap as memory pricing firmed on AI infrastructure orders. AirTrunk committed $21 billion to India data-center construction. These are not software-margin businesses. These are capital-intensive, long-cycle infrastructure plays pricing in sustained hyperscaler demand through 2027 or later.

Anthropic filed confidentially for an IPO. The timing sits alongside OpenAI and SpaceX dominating venture secondary markets. The model layer is going public while still unprofitable, which shifts the unit-economics debate from private venture boards to public equity analysts. When the S-1 disclosure comes, the inference-cost structure and gross margin per API call will be line-items, not projections.

The capex cycle is now financed for the next 24 months. The supply curve has not caught up. The model providers are moving toward mark-to-market pricing. The gap between what the infrastructure costs and what it returns is about to get measured in quarterly filings, not pitch decks.