Alphabet’s $85 billion raise signals a debt-funded era for Big Tech’s AI buildout

The five platforms that power most of the world’s cloud computing — Amazon, Alphabet, Meta, Microsoft, and Oracle — are collectively burning through capital faster than they can earn it. Capital expenditure as a share of revenue now runs between 45 and 57 percent across the group; as recently as 2024, that share sat in the low teens. The crossover point, where AI infrastructure spending began exceeding the operating cash flow that sustained it, arrived this quarter.

For any worker drawing a salary from one of these companies, or any consumer relying on services they run, the shift carries weight. The AI features landing in productivity software, cloud tools, and the recommendation algorithms that route news and shopping are no longer funded by profits. They are funded by borrowed money — and by an equity market that has, so far, chosen to believe the returns will arrive.

Alphabet moved first and most aggressively. The parent company of Google raised $84.75 billion in new equity, among the largest single capital raises in US corporate history. Some $44.75 billion, including a $10 billion private placement with Berkshire Hathaway, is directed at AI infrastructure and related corporate purposes. Amazon followed with a $25 billion bond. Oracle has since announced a combined $40 billion debt-and-equity plan for its next fiscal year.

Taken together, the five hyperscalers are projected to spend roughly $800 billion in capital expenditure in 2026, more than 80 percent above the prior year. That figure exceeds the annual GDP of Switzerland. The share of capex financed by external debt has risen from 9 percent in 2024 to 32 percent in the eighteen months to June 2026, with Morgan Stanley projecting that hyperscaler borrowing alone will top $400 billion this year — more than double the $165 billion raised in 2025.

The skeptic’s position is not hard to state. AI infrastructure carries heavy upfront costs and a long payback horizon. The revenue the new data centres are meant to generate — from cloud AI services, API access, and enterprise software subscriptions — is not yet visible at the scale required. At Amazon, Meta, and Oracle in particular, free cash flow is under pressure; FactSet’s analysis suggests some hyperscalers could approach zero or turn negative in 2026. The Bank for International Settlements has warned that the technology sector will need to issue roughly $1.5 trillion in new debt over the next three years to sustain the current pace — a debt wave with no precedent in the industry’s history.

The full picture will come into sharper focus when the hyperscalers report third-quarter results in mid-October. Alphabet, Meta, and Microsoft are scheduled within days of each other, followed by Oracle’s fiscal second-quarter report in December.

Tags: artificial intelligence, AI infrastructure, amazon, Alphabet, hyperscalers