Alphabet reported a Q2 2026 earnings release that shows Google negative cash flow for the first time since the company went public, according to Ars Technica. The odd part is that this happened in a quarter when Alphabet still printed money: $119.8 billion in revenue, ahead of analyst expectations.
The problem was not demand. It was spending. Alphabet said Google spent $44.9 billion in the quarter on capital expenditures tied to expanding its AI infrastructure. After excluding non-cash investment gains, the company’s operating cash flow was about $39.1 billion, according to the earnings figures cited by Ars Technica. That left free cash flow at negative $5.8 billion.
Why did Google have negative cash flow?
Google had negative free cash flow because its cash generated from operations did not cover its capital spending in the quarter. Free cash flow is the cash left after a company pays for the assets needed to run and expand the business, so it matters even when the income statement still looks fat.
Alphabet remains profitable, and Ars Technica noted that the company still has more than $100 billion available. The negative free cash flow figure is still a marker investors watch because it shows how much actual cash a business can keep after funding its buildout without leaning on asset sales or new borrowing.
Search remained Alphabet’s biggest engine, bringing in $63.3 billion in the quarter. Google Cloud reported $24.8 billion, up 23.8 percent from the first quarter, a sign Alphabet is seeing strong demand for the cloud and AI services it sells to other companies. Subscriptions, platforms, and devices produced $12.9 billion, while YouTube ads brought in $11.1 billion. Ars Technica tied the YouTube ad growth, up more than 12 percent from the previous quarter, to longer ad loads on the platform.
How much is Google spending on AI infrastructure?
Alphabet had previously told investors to expect $180 billion to $190 billion in capital expenditures for 2026. The company now says that total could reach $205 billion, according to Ars Technica. That would be far above the $91 billion Google spent in 2025 and roughly six times the $22 billion it spent in 2022, before the current AI buildout accelerated.
The spending is aimed at the data centers and related infrastructure needed to train and run AI models. Google also designs its own AI chips, and Ars Technica reported that the company has pitched its Tensor 8i and 8t chips as more efficient hardware for AI data centers. That claim is Alphabet’s, and investors will care less about the branding than whether the chips lower the bill fast enough.
The market reaction was cold. Google’s stock fell about 4.5 percent overnight after the earnings update and continued moving lower, according to Ars Technica. The reaction fits a broader investor concern: AI spending across the industry is expected to exceed $700 billion this year, while the path from model demos to durable profit remains uneven.
Alphabet has also signaled that higher spending will continue next year. At the same time, Google’s AI roadmap has some visible friction. The company recently delayed its flagship Gemini 3.5 Pro model, saying it remains in testing with a small group of partners. Ars Technica also cited reports that Google has not yet seen the performance gains it needs to keep pace with rivals such as GPT 5.6 and Claude Mythos, while the company has dealt with resignations among senior AI researchers.
For now, Alphabet can afford an ugly cash-flow quarter better than most companies. The sharper question for investors is whether Google’s AI infrastructure bill starts producing returns before negative free cash flow becomes more than a one-quarter accounting bruise.
This story draws on original reporting from Ars Technica.
