Alphabet spent more than it generated in cash this quarter: that is not the same as running out of money

On 22 July 2026, Alphabet reported second-quarter free cash flow of negative $5.855bn. The figure reflects $44.924bn of investment in data centres, servers and other assets, exceeding the $39.069bn of cash generated by operations. It is a meaningful sign of the financial scale of the artificial-intelligence race. It does not, by itself, mean Google has run out of cash.

The distinction matters because headlines often use “burning cash” for three different things: spending heavily on assets, reporting negative free cash flow, and lacking liquidity. Alphabet fits the second description this quarter. Its balance sheet reports $242.474bn in cash, cash equivalents and marketable securities at 30 June. Understanding those three terms makes any story about AI costs easier to read.

What happened in the quarter

In its official results, Alphabet reported revenue of $119.796bn, up 24% year on year, and operating income of $40.770bn. Google Cloud revenue was $24.768bn, up 82%, according to the company. At the same time, purchases of property and equipment reached $44.924bn, more than double the $22.446bn reported in the same quarter of 2025.

The company defines free cash flow as cash from operations minus capital expenditures. The arithmetic is straightforward: $39.069bn of operating cash flow minus $44.924bn of capex equals negative $5.855bn. The negative number did not arise because the operating business stopped generating cash. It arose because physical investment in the quarter was even larger.

That is the first step in reading the report. A business can be profitable on its income statement, generate cash from operations, and still report negative FCF if it is building assets faster than it generates cash. The reverse can also happen: a company can show positive FCF by cutting investment, without that guaranteeing future growth.

Capex is not an ordinary running cost

Capex, short for capital expenditures, is money spent on assets expected to serve for several years: buildings, networks, chips, servers or data centres. In AI infrastructure, those are the physical resources used to train models and serve requests. They are paid for today but used over time. That is why capex appears in cash-flow reporting rather than being fully treated as a single quarter’s operating cost.

This does not make it harmless. A large investment may deliver poor returns if demand fails to arrive, hardware becomes obsolete, or service prices do not cover the cost. But assessing that risk requires reading investment alongside revenue, operating cash flow, available capacity and financing, not just attaching the label “AI costs.” Alphabet attributes Cloud growth to demand for AI infrastructure and solutions; that is the company’s explanation, not a guarantee of future returns.

Free cash flow is a snapshot, not the entire balance sheet

FCF measures cash remaining after funding operations and capital investment. It is useful because it shows whether a business can pay dividends, repurchase shares, reduce debt or invest without external funding. But one negative quarter does not answer the solvency question. For that, readers need to look at available cash, marketable securities, debt and access to financing.

Alphabet ended June with $242.474bn in cash, equivalents and marketable securities. During the quarter it also reported $49.6bn in net proceeds from issuing common and convertible preferred shares, intended in part to scale AI infrastructure, and $20.3bn in net proceeds from senior unsecured notes. Those choices may raise questions about dilution or debt, but they contradict the literal claim that the company has no resources to finance its plan.

The release also gives a less dramatic perspective: trailing-twelve-month FCF was positive at $53.273bn. That does not remove the pressure of rising capex or predict 2027 results. It does prevent one quarterly column from becoming the whole story of Alphabet’s financial health.

An income statement does not replace a cash-flow statement

It is also important not to merge accounting profit with cash. The release reports $40.770 billion in operating income, while the cash-flow statement answers a different question: how much money came in and went out during the period. The two can diverge because payments and collections happen at different times, working capital changes and some accounting items do not involve cash. A careful reading therefore does not substitute profit for cash generation, or cash generation for profit. They describe different features of the same business.

Alphabet’s release offers another useful guardrail against a rushed interpretation. Quarterly net income includes a $98.227 billion unrealized gain related to marketable investments. The company reports that item in its income statement, while it defines free cash flow from operating cash flow and capital expenditures. Treating those lines as interchangeable can turn a change in a financial asset’s valuation into a claim about the company’s ability to build data centres.

Financing is not the same thing as return either. Issuing equity or debt can extend a company’s room to invest and meet commitments, but it does not prove that every additional server will earn an attractive return. It answers part of the liquidity question. Whether the investment is productive will show up later in revenue, margins and capacity use. That is why a reader should keep the four-line record and add a fifth, separate question: what evidence shows that the new infrastructure is earning its cost of capital?

There is a practical order to the check. Start with the cash-flow statement rather than a headline or a percentage change. Record the period beside every number, because a quarterly outlay and a trailing-twelve-month measure answer different questions. Then read the notes and financing disclosures before inferring distress from a negative FCF figure. Finally, separate what the company reports from what it predicts. The official release can establish the spending and the cash balance; it cannot settle the future demand, pricing or return on the infrastructure.

A record for reading the next headline

When a company announces a very large AI investment, fill in four lines. Operations: how much cash does the core business generate? Capex: how much is being invested in durable assets? FCF: what remains after subtracting one from the other? Liquidity: what cash, securities and financing are available to cover the gap?

For Alphabet, the record is clear: operations, $39.069bn; capex, $44.924bn; FCF, negative $5.855bn; liquidity, $242.474bn in cash and marketable securities, plus new financing during the quarter. It is a company that chose to invest more cash than it generated over three months. It is not a company whose official filing describes an inability to meet immediate obligations.

A fifth question remains: return. When, and at what margin, will those data centres generate revenue? The results show strong Cloud growth, but they do not separately establish the return on every dollar of AI capex. That uncertainty is exactly what markets are trying to price. A rigorous headline can say investment is pressuring FCF; it cannot declare the investment profitable or disastrous in advance.

The durable lesson

The AI race is turning software and advertising companies into infrastructure-intensive builders. That makes capex more important than it was a few years ago. It also demands more precise reading: investment is not a loss, negative FCF is not an absence of cash, and liquidity is not proof that a bet will pay off.

A reader can explain it in one sentence: Alphabet generated cash from its business, spent even more on infrastructure, and therefore reported negative quarterly FCF; its balance sheet and financing determine whether it can sustain that pace. That separation is more useful than choosing between AI euphoria and an alarmist headline.