Google might be the most important signal in AI right now. This is the only company that owns the whole stack: the apps people actually use, the AI models underneath them, the data centres those models run on, and the chips inside the data centres. Everyone else rents at least one of those layers from somebody, but Google owns all four. So when Google reported earnings last week, the whole world was watching in the hope of catching signs of what’s happening in the industry.

The headline number is that Google's cloud business grew 82% compared to a year ago. And the more interesting detail is buried in the commentary: Google is still turning customers away. It cannot physically build capacity fast enough to serve the demand already sitting in front of it. That's a very unusual problem to have.

The fix is to build, and fast. Google raised its spending plans for the rest of 2026 and warned that 2027 will step up "significantly" from there.

The market hated to hear this, and Google’s stock fell hard (now 20% down from all-time highs). The fear is legible enough. Free cash flow turned negative this quarter, the first time since 2004. Between the spending already guided and the land grab underway across the industry, that isn't reversing for several quarters at minimum. Google has gone further and funded the investment ramp with debt this year, and lately it even started selling new shares to the public. For a company long celebrated for the strongest balance sheet in the world, asset-light economics and steady buybacks, that's a real change of character… and plenty of holders aren't waiting around to see how it resolves.

There are plenty of reasons to stick around, though. First of all, Cloud is becoming a larger and larger portion of the overall business. As this segment scales, it gains operating leverage, and Google’s margins are steadily increasing. In other words, Google is becoming more profitable over time.

Second, as I mentioned in the intro, Google is probably one of the purest AI plays in the market right now: present across all verticals, with no balance sheet risk (Google actually has the capacity to borrow a lot more than what it’s currently doing) and no dependence on any one customer for future “promised” revenue.

Third, the valuation looks pretty compelling here. The P/E ratio of 16 can be misleading, since last quarter Google reported a massive accounting gain thanks to its stakes in SpaceX and Anthropic. The actual valuation multiple, if we look at operating income, is closer to 26, which in my opinion is still pretty cheap for a business with such a bright future ahead of it.

Right now, the big money is walking away because the cash flow looks ugly and nobody can tell you what all this spending will earn, or how long the cycle even runs.

But that's kind of the deal. You get one of the best businesses ever assembled at a reasonable multiple because the next 4-6 quarters are very difficult to model. Sure, if you need the stock to work this year (like a lot of professional funds), it's a hard pass. But if your horizon is 5+ years, you're being paid for patience, which is the one edge that retail investors like you actually have over funds that get fired for a bad quarter.

Upgrade to unlock full access to the RatedA Portfolio (with weekly updates on every buy/sell and rationale), and the complete database of companies + ratings.