Alphabet Earned $112 Billion and Raised $105 Billion. That Was Strange.

Alphabet reported $112 billion in profit for the second quarter of 2026. That is more profit than most countries' GDPs. Google's parent company has been around since 1998, and if you had told its founders in 2004, when it went public, that the company would one day report a quarterly profit larger than the entire federal budget of Portugal, they would have had a complicated conversation about what they actually built.

Then you look at what else happened that quarter. Alphabet raised $105 billion from outside investors. It sold common stock, mandatory convertible preferred stock (a hybrid security that pays interest now and turns into regular shares later), and senior unsecured debt notes. It was the largest equity issuance in history. The stock fell about 5% after hours.

A company earning $112 billion in profit raising $105 billion in outside capital. That is the sort of sentence that usually means someone is hiding something. In this case, the sentence is true but misleading in both directions.

The $112 billion is real. It is also almost entirely unrealized.

The basic point is that Alphabet's $112 billion in net income contains $98 billion of unrealized gains on equity securities. Alphabet holds minority stakes in private companies that have gone public or gotten closer to public — most notably SpaceX, which did what Bloomberg called the largest stock-market debut in history last month. When SpaceX's public price rose, Alphabet's share of that value jumped, and the jump got booked as profit.

The accounting rule behind this is ASC 321 (the US standard that took full effect in fiscal 2018). It says that when a company holds passive minority stakes in other businesses whose shares have a "readily determinable fair value," those stakes are marked to market every quarter, and the change flows straight through the income statement into net income. Alphabet has been a public company for over two decades and has been marking these investments through its earnings since 2018. This is not a new trick. It is a feature of the accounting rules that most people have been quietly ignoring until SpaceX went public and made the number visible.

Before this quarter, Alphabet's other income was $1.2 billion. It is now $98 billion. The difference between those two numbers is SpaceX's IPO price appreciation, with some help from other portfolio holdings — Alphabet is also a major investor in Anthropic, the AI startup that may go public as soon as October.

So the $112 billion headline profit is genuine under the rules. It is also, for cash-flow purposes, entirely imaginary. Alphabet cannot spend $98 billion of paper gains. It can sell the shares someday and convert the gains into cash, but the gains are a mark-to-market measurement, not a collection of cash receipts.

The part that actually matters is the cash. Alphabet generated $39.1 billion in operating cash flow in the quarter. That sounds like a lot. It then spent $44.9 billion on capital expenditures — mostly AI data centers, servers, and network infrastructure. The difference is negative $5.9 billion. Alphabet's free cash flow turned negative for the first time in its 22-year public history.

Capex more than doubled from the same quarter last year, when it was $22.4 billion. Management raised full-year capex guidance from $180–190 billion to $195–205 billion. And they said spending will increase again in 2027.

Here is a tiny dialogue between two people who are reading different parts of the same report:

Investor: You earned $112 billion this quarter. Accountant: Yes. Investor: Great, I'm not worried about your spending. Accountant: I wouldn't be you.

The investor is reading net income. The accountant is reading cash flow. They are looking at different machines.

The cash-flow deficit is why Alphabet raised $105 billion. The equity raise was $84.75 billion, upsized from $80 billion, and it came in several flavors:

• $18 billion in underwritten common stock
• $16.75 billion in mandatory convertible preferred stock at a 6.25% coupon, converting to common shares in May 2029
• $10 billion in a private placement from Berkshire Hathaway, bought at a 6.5% discount to market price
• $40 billion authorized under an at-the-market program for future sales

Plus $20.3 billion in senior unsecured notes.

The preferred stock detail is worth sitting with for a moment. It pays 6.25% interest and converts to regular shares three years from now. In practice, this is a bridge instrument: it brings cash in now while postponing the dilution that would happen if Alphabet sold the same amount of common stock today. But the 6.25% coupon is expensive in today's rate environment, and the conversion means existing common shareholders will see more shares outstanding when it happens. The preferred stock is basically a loan that becomes equity. It is debt with a maturity date that the company doesn't have to repay if it prefers to issue shares instead.

Berkshire's $10 billion stake at a 6.5% discount is a separate signal. Warren Buffett's firm does not routinely buy positions in companies at a discount unless it thinks the market is mispricing something. Or it thinks the company needs the capital urgently enough to pay a discount premium. Or both.

Buybacks, which were $13.2 billion in the same quarter last year, dropped to zero. Alphabet is no longer returning capital to shareholders. It is raising it.

At the end of the quarter, Alphabet held $242 billion in cash and marketable securities. That number is high. But part of it is the $98 billion in unrealized gains that just flowed through the income statement. If those gains were reversed — if SpaceX's price dropped — the cash-and-securities number would shrink dramatically.

The cloud numbers are genuinely strong. Revenue of $24.8 billion, up 82% year over year. Operating income of $8.8 billion, up from $2.8 billion a year ago. Nearly 90% of the Fortune 100 now use Gemini Enterprise. Gemini itself has 950 million monthly active users.

The business is growing fast enough to justify massive spending if the spending actually generates returns. That is the question the market answered with a 5% selloff. Not "is cloud growing" but "is the cash burn sustainable at this scale?" If capex is $200 billion this year and $200-plus billion next year, operating cash flow needs to accelerate to keep up. Alphabet's adjusted operating cash flow (adding back about $11.7 billion in working capital consumption, mostly receivables and prepayments for third-party data center capacity) was closer to $50.8 billion. That covers the quarterly capex of $44.9 billion by a thin margin.

On an annual basis, the math gets tighter, not looser. The gap between operating cash generation and capex burn is the structural question.

The simplest model is this: Alphabet is running an AI infrastructure buildout that costs roughly $50 billion per quarter and generates roughly $39 billion per quarter in operating cash. The shortfall is being funded by external capital. The $98 billion in unrealized gains masks the shortfall on the income statement but does not fix it on the balance sheet.

This is not, strictly speaking, unusual. Companies build infrastructure ahead of revenue all the time. Amazon spent years at negative free cash flow while building AWS. The difference is scale: $200 billion of annual capex is an extraordinary number for any company, and raising $105 billion in a single quarter is extraordinary on top of that.

The interesting part is not that Alphabet is spending a lot. It is that the accounting rules make the spending look covered when it isn't. Net income says $112 billion profit. Cash flow says negative $5.9 billion. The $118 billion gap between those two numbers is almost entirely the unrealized gains. Anyone reading the income statement without also reading the cash flow statement would walk away with the wrong picture.

The old-finance analogy is a bank that marks its loan portfolio to fair value on its earnings call, then borrows from the Federal Reserve to pay its bills. The two numbers are both real. They just measure different things, and the one that matters depends on which question you're trying to answer. If the question is "is this company valuable?" then the paper gains matter. If the question is "can this company pay for what it is building?" then the paper gains are decoration.

The stock is at about $354, down from its 52-week high of $408 and up roughly 13% year to date. The market has not decided whether to treat the AI buildout as an investable thesis or a funding problem. That uncertainty is baked into the price.

The catch in the $112 billion profit headline is not that the number is fake. It is that the number is real, cashless, and structurally disconnected from the machine that actually needs funding. Alphabet built a cash-generating monopoly in search and advertising, then decided to turn that cash flow engine into an AI infrastructure company that costs more to build than it currently generates. The accounting rules let it look like the transition was seamless. The cash flow says otherwise.