Alphabet spent $5.9 billion more than it took in last quarter, its first cash burn on record, on $44.9 billion in capital spending, and added another $15 billion to its 2026 plan.
For a decade, the four biggest US technology companies funded their growth out of their own cash engines. After Alphabet's first-ever quarterly cash burn of $5.9 billion, paired with a fresh $15 billion step-up in 2026 capital spending signaled on the same earnings call, that financing model is breaking. The way Big Tech pays for the AI build-out is now the story.
The $5.9 billion shortfall is small relative to Alphabet's $119.8 billion in second-quarter revenue, up 24% year over year. It is large relative to the $44.9 billion in capital spending the company absorbed last quarter, most of it directed at the data centers, custom chips, and grid capacity that feed Google's AI products (Alphabet Q2 2026 earnings release). The non-GAAP free cash flow line, the measure that strips out a roughly $99 billion mark-to-market gain on Alphabet's equity investments, is the first time on record that figure has been negative (StockTitan summary of GOOG 8-K).
The cash flow reversal shows up at Amazon, Meta, and Microsoft too. Industry-wide AI capex is on track to clear $700 billion in 2026, and consensus projections cited in coverage of the print have Amazon burning cash this year, Meta's free cash flow projected to shrink 95.7% to roughly $1.85 billion, and Microsoft's current-fiscal-year cash generation landing around $25.39 billion (Brazil Sun, Anabelle Colaco). On July 23, the day of the print, Alphabet traded down about 6%, Meta and Amazon roughly 3.5% each, and Microsoft little changed, pricing in a financing-model shift rather than a single quarter's miss.
Capital spending is paid in cash, denominated in chips, land, and long-dated power purchase agreements. Revenue, even at the AI-exposed end of the business, is still trailing. Google Cloud, the segment most exposed to the AI infrastructure cycle, posted $24.8 billion in revenue for the quarter, up 82% year over year, and the company's Cloud backlog now stands at $514 billion (Alphabet Q2 2026 earnings release). CEO Sundar Pichai framed the demand as broad, citing Gemini Enterprise adoption by roughly 90% of the Fortune 100 and the launch of Gemini 3.6 Flash and Gemini 3.5 Flash-Lite during the quarter (Alphabet CEO Q2 2026 earnings call remarks). That backlog is the asset on which the financing model is being extended. Cloud backlog grows, capital spending follows, and the gap between the two has to be closed somewhere.
The "somewhere" is debt and equity. In June, Alphabet raised $49.6 billion in equity, including Class A, Class C, and a mandatory convertible preferred, plus $20.3 billion in net proceeds from senior unsecured notes, and authorized a new at-the-market equity program (StockTitan summary of GOOG 8-K). That is a deliberate move from self-funding to capital-markets funding, and it arrived before the cash flow line turned negative for the first time. The company did not wait until it ran out of options. It raised the financing while it still had the credit standing to do so at the lowest immediate cost.
The legitimate worry, voiced by Saxo Markets chief investment strategist Charu Chanana in the wake of the print, is whether AI revenue can grow faster than capital expenditure, depreciation, and operating costs (Brazil Sun, Anabelle Colaco). That is the falsifier for the financing-model thesis. If Google Cloud's 82% growth compounds fast enough to outrun the $44.9 billion quarterly capex run-rate, the cash burn reverses. If it does not, the debt and equity tap stays open, and shareholders who treated Big Tech as a cash compounder for ten years will see the cash component of their return equation shrink.
The watch item for the next two quarters is therefore narrow: AI-revenue growth against capex plus depreciation, on a per-quarter basis, at Alphabet, then again when Microsoft, Meta, and Amazon print their own numbers next week. The shape of the cross-company capex table in those reports will decide whether the $5.9 billion burn is a one-quarter anomaly or the first line of a new chapter.