Big Tech AI Capex Tracks $770B as Alphabet Burns Cash

Hyperscaler capex is on pace to hit $770 billion in 2026 as Alphabet posts its first-ever negative free cash flow quarter, reshaping corporate bond markets and competing with U.S. Treasury issuance for institutional capital.

Lead

The AI infrastructure arms race among the world's largest technology companies crossed a historic threshold in the second quarter of 2026, when Alphabet reported negative free cash flow for the first time in its 22-year history as a public company. Spending $44.9 billion on capital expenditures in a single quarter - absorbing 37.5 cents of every dollar in revenue - drove free cash flow to -$5.85 billion, even as revenue surged 24% year over year to $119.8 billion. The result crystallized the defining tension of the current technology cycle: companies that once generated inexhaustible cash are now consuming it at a pace that is simultaneously reshaping global capital markets.

AI Stocks and the $770 Billion Spending Wave

Combined capital expenditures from the five largest cloud and AI infrastructure providers - Microsoft (MSFT), Alphabet (GOOG), Amazon (AMZN), Meta (META), and Oracle (ORCL) - are now tracking between $750 billion and $770 billion for 2026, according to UBS credit strategists, a 67% increase over 2025 levels and roughly double what the group spent two years earlier. Microsoft has guided to approximately $190 billion in calendar-year capital spending, including $145 billion reported in fiscal year 2026 ending June 30, with its Azure business crossing a $100 billion annual revenue run rate and recording 43% year-over-year growth in the final fiscal quarter. Amazon is targeting approximately $200 billion. Alphabet raised its full-year guidance for the second consecutive quarter, lifting the ceiling to $205 billion. Meta, which has raised its own guidance twice since January, is now guiding to $125 billion to $145 billion. Nvidia (NVDA), whose GPU platforms underpin the majority of this compute build, remains the concentrated beneficiary of the cycle.

Why Did Alphabet Post Negative Free Cash Flow?

The mechanics behind Alphabet's historic quarter are straightforward: operating cash flow of $39.1 billion was exceeded by capital expenditures of $44.9 billion, producing a -$5.85 billion free cash flow result. Operating income itself grew 30% to $40.8 billion, and operating margin expanded two percentage points to 34% - a strong underlying business performance by any historical standard. The capital outflow reflects the cost of building compute infrastructure at the scale required to train and serve Gemini AI models, support agentic AI deployments, and sustain Google Cloud, which grew 82% in Q2 to $24.8 billion in quarterly revenue, with Cloud operating income tripling to $8.8 billion. Management has now raised full-year capex guidance twice within 90 days.

How Is This Reshaping Corporate Bond Markets?

The scale of hyperscaler borrowing is reordering the U.S. investment-grade bond market in ways that have no recent precedent. Hyperscalers and adjacent AI companies issued approximately $225 billion in bonds through the first half of 2026 - a near-1,000% surge from the same period a year earlier - and are on pace to reach $400 billion for the full year, according to Morgan Stanley estimates. Amazon completed the largest issuance cycle, raising $53 billion in a combination of a $37 billion U.S. bond offering and $16 billion in euro-denominated notes. Alphabet tapped global markets in February, raising $31.5 billion across dollar, sterling, and Swiss franc tranches. UBS estimates that if capex continues at current pace, hyperscalers will need to raise an additional $40 billion to $50 billion beyond prior projections. The supply pressure is measurable in spread dynamics: median spreads on two-to-four-year hyperscaler bonds have risen to 40 basis points above comparable Treasuries, from 30 basis points in 2025, while five-to-seven-year spreads have moved to 60 basis points from 50.

What Does Hyperscaler Debt Mean for Treasury Markets?

At $400 billion in projected annual bond issuance, hyperscalers are introducing a supply dynamic that competes directly with U.S. Treasury offerings for institutional capital allocation. New-issue concessions, historically negligible for investment-grade megacap technology borrowers, have averaged 12 basis points in 2026 as underwriters price in growing supply fatigue. Deals remain oversubscribed - often by four times or more - but market participants have flagged rising leverage from companies previously considered near-immune to balance-sheet risk. Goldman Sachs projects that by 2027, more than a third of Big Tech's AI investments will be funded with borrowed capital, a structural shift from the free-cash-flow self-financing model that defined the prior decade.

Outlook

The current trajectory implies aggregate hyperscaler capex surpassing $1 trillion in 2027. Near term, the critical variable is whether AI workload revenue growth - measured in Azure acceleration, Google Cloud momentum, and Amazon Web Services throughput - continues to justify the investment pace, and whether bond markets sustain appetite for issuance at a rate that has no historical analog among investment-grade issuers. Alphabet's negative free cash flow quarter marks not a sign of financial distress but the moment the AI infrastructure cycle entered its most capital-intensive phase. The duration of that phase, and the returns it eventually produces, will determine whether the debt accumulated in 2026 proves transformative or excessive.