Hyperscalers Turn to External Financing as AI Capex Hits $690 Billion and Free Cash Flows Collapse

The five largest cloud infrastructure companies have shifted from self-funding their capital expenditure to raising external capital at scale, according to FactSet Insight analysis published July 23. Aggregate capex for Alphabet, Amazon, Meta, Microsoft, and Oracle is expected to exceed $690 billion across their respective FY26 periods, an increase of more than 80% year over year.

That spending is outrunning the cash these companies generate. Free cash flows for FY26 are expected to move close to zero or turn negative for all five except Alphabet and Microsoft, according to the analysis. The result: incremental annual debt as a share of capex rose from 9% in FY24 to 32% by mid-2026, and equity has returned to the funding mix for the first time in years.

The Numbers

Alphabet priced an $84.75 billion equity raise in June 2026, the largest equity capital transaction for a listed corporation, according to FactSet. Of that total, $44.75 billion (including a $10 billion private placement with Berkshire Hathaway) is directed at general corporate purposes including AI capex. Oracle plans $40 billion of combined debt and equity for FY27. Amazon recently placed a $25 billion bond that was 2.5x oversubscribed.

Calendar 2026 capex guidance across the group points to close to $800 billion when including finance leases and customer pre-payments, FactSet reported. Higher component prices, particularly memory chips where high-bandwidth memory demand outstrips supply, are pushing costs further.

Aggregate total debt across the five hyperscalers has reached approximately $700 billion.

Credit Pressure Is Already Here

S&P downgraded Oracle on July 9 to BBB- from BBB, citing surging capex, negative free cash flow, and customer concentration on OpenAI and Anthropic. A matching move from Moody’s (currently Baa2 Negative, one notch above S&P’s new rating) would likely pressure Oracle’s bond prices and widen credit default swaps just as the company plans to raise additional funding.

The other four hyperscalers retain substantial headroom before any downgrade, with total debt-to-EBITDA sitting around 1x or below. But the underlying pressures, customer concentration and declining cash flows, are shared across the group.

Where the Money Goes

Compute (processors and servers) now represents approximately 60% of capex, up from 43% in 2022, according to FactSet. The shift toward shorter-lived compute assets means a growing share of capex becomes recurring replacement spending rather than one-off construction, making these spending programs harder to wind down.

Agentic AI workloads are a core driver. The analysis notes successive waves of larger models and multi-agent orchestration have pushed the expected peak of the capex cycle further out each time companies report earnings. Alphabet raised its 2026 capex guidance to $180 billion to $190 billion in Q1, up from $175 billion to $185 billion announced just one quarter earlier. Meta lifted its guidance to $125 billion to $145 billion from $115 billion to $135 billion.

The IPO Queue Adds Urgency

FactSet frames the current financing race as partly driven by anticipated IPOs from Anthropic and OpenAI in 2026 and 2027. Hyperscalers securing capital now are trying to lock in favorable terms before markets are asked to absorb even larger AI-related issuances.

Meta is reportedly planning a cloud business to sell excess AI compute to external customers, with CEO Mark Zuckerberg saying a cloud business was “definitely on the table,” per FactSet. Meta’s shares rose 17% between July 1 and July 16, suggesting investors welcome a second monetization route that would ease free cash flow pressure.

The Payback Timeline

The core question FactSet raises is when AI monetization catches up with investment. Estimates point to more than $900 billion in aggregate capex by FY28. AI costs are front-loaded with returns expected over a longer horizon, and the inflection point where AI profits overtake the build-out remains uncertain. For now, external financing will play a prominent role, alongside leasing and off-balance-sheet joint ventures for data center construction.