Alphabet's $112 Billion Profit Is a Mirage. The Cloud Numbers Aren't.

The headline number is a ghost. Alphabet reported $112 billion in net income for the second quarter of 2026. Almost $99 billion of it came from unrealized gains on equity holdings — mostly its stakes in Anthropic and SpaceX. Strip that out and the company earned roughly $13 billion, perfectly ordinary for a business pulling in $120 billion in revenue.

The real story isn't in that distorted bottom line. It's in what happened above it — and why the stock still fell 7 percent on news that looked good.

The operating engine didn't miss a beat.

Revenue was $119.8 billion, up 24 percent, beating the $116.9 billion consensus. Operating income came in at $40.8 billion, up 30 percent, with operating margin expanding to 34 percent from 32 percent a year ago. Search grew 17 percent to $63.3 billion. YouTube ads grew 13 percent to $11.1 billion. The ad machine is humming at a clip that's hard to dismiss.

But the piece that changes the narrative is Google Cloud. Revenue surged 82 percent to $24.8 billion, well above estimates. More importantly, cloud operating income jumped to $8.8 billion from $2.8 billion a year earlier. That's a threefold increase in actual operating profit from the AI infrastructure business in twelve months. The cloud unit was long praised as the growth engine; now it's starting to look like the margin engine, too.

Nearly 90 percent of the Fortune 100 are using Gemini Enterprise. The company launched three cheaper Gemini models to compete on cost and latency. Cloud is the one part of Alphabet's business where the AI investment cycle is showing real operating leverage.

So why did the stock sell off?

Capex guidance. Alphabet raised its estimated capital expenditures to a new range that extends past $200 billion, up from $180 billion–$190 billion. The CFO said 2027 spending would "significantly increase" from there. Analysts at Truist projected 2027 capex could reach $270 billion–$280 billion.

Free cash flow flipped negative in the quarter at $5.9 billion. Trailing-twelve-month free cash flow is $53.3 billion but declining 20 percent year over year. The market has a low tolerance for spending it can't see the return on, and $200 billion-plus of annual capex with no near-term FCF payoff is exactly the kind of uncertainty that makes investors step back.

This isn't a new fear. The big-tech AI capex cycle has been building for two years. What changed in July is that Alphabet became the first major US tech company to report this earnings season, and the guidance increase set the tone. Barron's called it a bad omen for the sector. Alphabet Falls as $205 Billion Spending Plan Fuels AI Cost Fear.

The old story is getting stale, even if the risk is real.

The market is still pricing Alphabet as an ad company writing a blank check to compete in AI — burning cash with an uncertain payoff. That story was valid a year ago. It's becoming stale now that cloud operating income has tripled and the backlog sits at $514 billion.

But I won't pretend the free cash flow path is clean. It isn't. FCF turned negative in Q2, and management gave no reason to think it would recover before capex accelerates further in 2027. Simple forward multiples beat complex DCF models, but there's no clean multiple argument when the cash conversion is deteriorating. Alphabet trades at 34.5 times forward earnings — not cheap, not absurd, but not a discount that forgives execution risk.

Berkshire Hathaway invested $10 billion in Alphabet in June at roughly $329 a share. The stock has drifted below and above that level since, currently trading around $354, well down from its 52-week high of $409. Berkshire's stamp of approval is real — but it doesn't eliminate the capex question. It just means one major investor has decided the risk-reward works at that entry.

AInvest's aggregate signal still labels Alphabet a Buy, with a composite analysis rating of 4.8 out of 5. The consensus view hasn't budged. That matters less than the cash-flow evidence, but it does show the professional analyst crowd hasn't rewritten its view despite the 7 percent post-earnings selloff.

What would make this a buy setup?

The proof point is cloud margin expansion sustaining through the spending cycle. If cloud operating income keeps growing at a pace that outpaces the capex burn — for example, reaching $15 billion to $20 billion annually as the backlog converts to revenue — the math starts to support the spending. The $514 billion in contracted work gives that trajectory room to play out.

Alternatively, the stock could simply find a floor around the Berkshire entry zone ($320–$340) while the capex cycle matures, letting patient investors accumulate at a discount to the 52-week highs without needing perfect FCF timing.

What breaks the setup.

Cloud growth decelerates below 50 percent while capex keeps climbing. The backlog stops expanding or starts slipping. Free cash flow stays negative for three or four consecutive quarters and management doesn't adjust the spending plan. Any of those would tell me the spending isn't translating into the operating leverage the backlog promises.

Discipline over ego. The inflection is visible in the cloud numbers, but the free cash flow bridge isn't built yet. This is a watch position, not a full entry — unless the stock gives you a Berkshire-level price and the cloud margin trajectory holds through Q3.

The market is still pricing the capex fear. The cloud operating income says the old story doesn't fit anymore. The free cash flow says wait and see which one wins.