Google Broke 22 Years of Cash Flow to Own the Answer

Alphabet has been a public company since 2004. In twenty-two years it never once posted a negative free cash flow quarter. On July 22, 2026, it did.

Negative $5.9 billion. In the same quarter it delivered record operating income — $119.8 billion in revenue, up 24%, with income from operations up 30% to $40.8 billion at a 34% margin.

The stock fell 7%. The coverage called it a spending problem.

It isn't. It's a disclosure.

First, an Asterisk Nobody Put on the Headline

You will see it reported that Alphabet's earnings per share rose 294% to $9.11. That number is not what it looks like.

Other income for the quarter carried a net gain of $98.0 billion, almost all of it unrealized gains on Alphabet's equity holdings. The company disclosed that the $99.0 billion equity-securities gain by itself added $77.1 billion to net income and $6.26 to diluted EPS. Take it out and EPS is around $2.85.

I am flagging it because it cuts against my own argument if I leave it unsaid. The record worth holding against $205 billion of spending is the $40.8 billion of operating income, up 30%. That is the business. The EPS figure is a mark-to-market entry that can reverse next quarter, and Alphabet says so in the filing.

Read the Whole Sentence

Operating cash flow was $39.1 billion, up 41% year over year. Capital expenditure was $44.9 billion, roughly double what it was a year earlier. Full-year guidance went up to $195 billion to $205 billion, the second raise in three months and the third distinct 2026 number in three quarters. Trailing-twelve-month capex is $132.4 billion, up 98%. Capital spending now consumes 37.5 cents of every revenue dollar, the highest share in the company's public life.

Put that in perspective. Alphabet's entire capital expenditure for full-year 2022 was about $31 billion. The company now spends more than that in ninety days.

It has bought back zero shares for two consecutive quarters, against $13.2 billion in the year-ago period. Its CFO told analysts free cash flow will remain under pressure and that 2027 spending rises significantly. She declined, for the second quarter running, to say by how much.

None of that is a company under stress. Alphabet holds $242.5 billion in cash and marketable securities and still generated $53.3 billion in free cash flow over the trailing twelve months.

This is a company that chose to outspend its own cash generation. Deliberately. At scale. For the first time since the Bush administration.

The only question worth asking is what it's buying.

Get the Funding Number Right

The figure circulating is $85 billion. It comes from Alphabet's own June 2 pricing release, which headlined a total equity raise of $84.75 billion. That is a program size, not money collected. It bundles the gross June offerings, a $10 billion private placement to a Berkshire Hathaway affiliate, and the entire $40 billion capacity of a new at-the-market equity facility that had sold nothing as of June 30.

Here is what actually landed, per the cash flow statement: $30.5 billion from Class A and Class C stock, $19.1 billion from 6.25% mandatory convertible preferred stock, and $20.3 billion from senior unsecured notes. Call it $70 billion of new capital in one quarter.

The balance sheet moved more than the headline suggests either way. Long-term debt more than doubled, $46.5 billion to $98.2 billion. Interest expense went from $261 million to $1.28 billion, nearly five times, in a single year. The cleanest balance sheet in technology now carries preferred stock ranking ahead of its own common, converting around May 2029.

It's Buying the Answer

Roughly 60% of that spend is servers. The rest is data centers and networking. What runs on it is Gemini, Google AI Overviews, and AI Mode.

The usage numbers Pichai put on the quarter make the purpose plain: 950 million monthly active users on the Gemini app, 22 billion API tokens processed per minute, nearly 90% of the Fortune 100 on Gemini Enterprise.

Google is not defending ten blue links. Ten blue links do not require $205 billion a year. Google is defending its position as the place a buyer asks the question, as the act of asking migrates from a search box that returns a list to an engine that returns an answer.

That is the whole strategy, and Alphabet just published its price.

I have spent decades in a business built on getting brands into other people's coverage. The mechanics changed underneath all of us, and most of the industry is still describing the change as a channel. It isn't a channel. The retrieval layer is where discovery now happens, and the incumbent just posted the first negative cash flow quarter in twenty-two years to make sure it owns that layer.

The Demand Side Moved First

Capex is Google's answer to a question the market already answered.

Similarweb clickstream data analyzed by SparkToro found that 68.01% of Google searches in the first four months of 2026 ended without a click. In 2024 that figure was 60.45%. A randomized field experiment published on SSRN this year found that showing an AI Overview caused outbound organic clicks to drop 39.8% — causal evidence, not a correlation. An Ahrefs study across 300,000 keywords put the click-through decline for top-ranking pages at 58%.

Chartbeat data cited by Press Gazette shows publishers worldwide receiving roughly a third fewer search referrals since AI Overviews rolled out broadly. And inside Alphabet's own numbers, Google Network — the third-party publisher line — is the single revenue line not growing. It came in at $7.30 billion this quarter against $7.35 billion a year ago, after falling 4% in Q1.

AI Overviews now reach about two billion people a month. ChatGPT reported 900 million weekly active users in February. The behavior did not scatter across a dozen new destinations. It consolidated into a handful of answer surfaces, and Google is spending a fifth of a trillion dollars a year to make sure one of them stays its own.

Three Quarters of a Trillion Dollars

Alphabet isn't alone. Meta guided 2026 capital expenditure to $125 billion to $145 billion and watched its stock drop more than 9% on the news. Microsoft is tracking toward roughly $190 billion for the calendar year. Amazon is guiding near $200 billion.

Combined hyperscaler capital expenditure is expected to reach approximately $725 billion in 2026, up about 77% from roughly $410 billion in 2025. Goldman Sachs now models $5.3 trillion across the four from fiscal 2025 through 2030, revised up from $4.5 trillion before first-quarter earnings.

Nobody spends $725 billion in one year to serve a use case they think is a phase.

Where I'd Push Back on Myself

The bear case is not stupid, and I'd rather state it than pretend it away.

One quarter is one quarter. Capital expenditure is lumpy by design, and a single negative print proves less than a trend. Trailing free cash flow is still positive at $53.3 billion. The balance sheet is enormous. Nothing here says distress.

The sharper risk isn't cash at all. It's depreciation. Alphabet booked $7.1 billion of depreciation this quarter against $44.9 billion of capital expenditure. Servers depreciate over five or six years, buildings over decades. That wave lands on schedule whether or not AI revenue shows up to meet it, and it could hurt reported margins badly in 2027 and 2028.

Here's the thing. Both can be true. The spending can prove badly timed for shareholders and still be a completely accurate signal about where buyer attention went. Those are two different questions, and if you run a brand you only need the answer to the second one.

What This Means If You Run a Brand

The narrowing is the part to understand.

A search results page gave ten companies a position on it. A generated answer names three sources, sometimes one. As the compute behind those answers scales, and $205 billion a year is considerable scale, the answers get more confident and the set of cited sources gets tighter.

Citation Share is the scarcer asset every quarter this continues. Not impressions. Not rankings on a page the incumbent is spending a fifth of a trillion dollars to sit in front of.

Most companies cannot answer a simple question: when a buyer asks ChatGPT, Claude, Gemini, Perplexity, or Google AI Overviews the defining question in your category, how often does your name appear in the answer? In 2024 not knowing was forgivable. Against this capital commitment, it's a governance failure.

What I'd Do Monday

Get a baseline. Run your twenty highest-intent category prompts across all five engines and count how often you appear. You cannot manage a number you have never measured.
Separate rank from citation. They are different assets and they do not track each other. Reporting that conflates them is telling your board something untrue.
Audit what the engines can actually reach. Structured data, clean entity naming, and pages that are not sitting behind a wall. Retrieval cannot cite what it cannot read.
Fix the third-party record. The engines lean on Wikipedia, Reddit, wire services, and trade press far more than on your own site. That is where most of the work is.
Put one number on the board deck. Not traffic. Citation Share, tracked quarter over quarter, in the categories you intend to own.

The Line I'd Underline

Google spent more cash than it generated, for the first time in twenty-two years, to make sure it still owns the moment a buyer asks a question.

If the company that built search is willing to break its own cash flow to own the answer, the answer is worth owning.

Build the infrastructure before the crisis — not during it.