Alphabet Isn't a Conglomerate. It's One Engine Towing Everything Else.

Open Alphabet's annual report and you find four neat boxes, the way a holding company is supposed to look: Google Services, Google Cloud, Other Bets, and a catch-all for corporate. It reads like a portfolio - search here, cloud there, self-driving cars and life-sciences bets off to the side. Then you look at where the money actually comes from, and three of the four boxes turn out to be passengers. One box earned $121.3 billion in operating profit last year. The whole company earned $112.4 billion.¹ Read that twice. The single biggest segment out-earned the entire corporation - because it was busy paying for everyone else.

The official story is that Alphabet is a diversified technology conglomerate with multiple profitable businesses. The real story is that it is a single-engine aircraft wearing the paint job of a fleet. The engine is Google Search, and almost everything else in the building is either being carried by it, just learning to carry itself, or actively burning the fuel it produces.

One segment earns more than the whole company

Here is the arithmetic that the four-box layout hides. In FY2024, Google Services posted $121.3 billion in operating income. Google Cloud added $6.1 billion. Then the subtractions begin: Other Bets lost $4.4 billion, and a line called 'Alphabet-level activities' lost another $10.5 billion. Net it all out and you land at $112.4 billion in total operating income.¹ The math only works one way. Search profit isn't one contribution among several - it is the source pool that the losses are drawn against. Cloud's modest profit barely dents the deficit that the moonshots and the corporate center create. The treasury has exactly one funded account.

Drill into Google Services and the concentration gets sharper still. Google Search & Other generated $198.1 billion of revenue in FY2024 - about 57% of Alphabet's entire top line.² YouTube advertising, the business everyone calls Alphabet's crown jewel number two, brought in $40.4 billion, roughly the same as Google Cloud's $43.2 billion and the Subscriptions/Devices line at $40.3 billion.² YouTube doesn't even get its own segment; its profit is folded inside Services, invisible in the filings. The story we tell ourselves is of a multi-pillar empire. The filing describes one pillar and several lean-tos.

What the engine is quietly paying for

Follow the fuel and you see the cross-subsidy in motion. Other Bets - the home of long-horizon ventures like Waymo - produced just $1.648 billion of revenue in FY2024 against a $4.444 billion operating loss.⁸ It loses nearly three dollars for every dollar it earns, and it isn't trending toward rescue: in the fourth quarter, Other Bets revenue actually fell to $400 million from $657 million a year earlier, while quarterly losses widened.⁸ The popular line that the moonshots are 'close to breakeven' is simply not in the numbers. They are funded experiments, and the funder is Search.

Then there is the $10.5 billion 'Alphabet-level' loss - the least-examined line in the whole report. It isn't a business at all. It is the shared cost of being Alphabet: general AI model R&D, philanthropy, and corporate finance, HR, and legal - the last of which now includes fines and settlements.⁴ In other words, the bill for the company's most expensive ambition, frontier AI, and the bill for fighting its antitrust wars both land here, unallocated, and both get paid out of the same account. Search doesn't just fund the future Alphabet wants to build; it funds the legal defense of the present it already has.

Even Google Cloud, the one story you might reach for to argue diversification, undercuts it on inspection. Cloud was loss-making as recently as 2022 and only turned consistently profitable in 2023. Its $6.1 billion of operating income in 2024 is real and growing - but it spent years drawing on Search's profit to get to the point where it could finally pay its own way.¹ The cross-subsidy didn't just fund the moonshots. It funded the segment now being marketed as proof Alphabet doesn't need the subsidy.

The one engine is now in a courtroom

Here is why the single-engine design suddenly matters. In August 2024, a federal court held that Google unlawfully monopolizes the markets for general search and search text ads - the exact business that produces the profit funding everything else - through exclusive contracts with browser makers, device manufacturers, and carriers.⁵ The court found Google held almost 90% of desktop search and nearly 95% of smartphone search at the time of the suit.⁶ The very distribution that makes Search the most profitable machine in the company is what the court declared illegal. The flywheel and the violation are the same object.

And it isn't a single front. In April 2025, a separate court found Google liable on two counts in the DOJ's adtech case, ruling its conduct 'substantially harmed' publishers - a second antitrust loss, distinct from the search ruling.⁷ The engine isn't being challenged once; it's being challenged in two courtrooms at the same time.

Isn't this just what a healthy company does?

The fair objection is that funding new businesses from a dominant cash machine is exactly how great companies are supposed to compound. Amazon used retail to seed AWS; Apple used the iPhone to fund everything that followed. A profit engine subsidizing the next engine isn't a flaw - it's strategy. True. And the genuine counter to the alarm is that the engine just got stronger, not weaker: Google Services' operating margin recovered to 39.77% in FY2024, up from a 34.15% trough two years earlier.³ The machine is humming. Cloud has crossed into self-funding. By that read, the cross-subsidy is working as designed.

But there's a difference between cross-subsidy by choice and cross-subsidy by necessity. Amazon could fund AWS because retail was a durable, uncontested business. Alphabet funds its future from a business a federal court has ruled an illegal monopoly, sustained by distribution deals a judge has now ordered it to end.⁶ The remedies stopped short of breaking up Chrome or Android, and Google plans to appeal - so the engine isn't seized, only constrained, and a final resolution could take years.⁶ That's the honest middle: not collapse, but a forced loosening of the exact mechanism that made the one funded account so reliably full. When the source of every subsidy is the thing under legal pressure, 'diversified' is the wrong word for it. 'Concentrated, and aware of it' is closer.

Strip the holding-company packaging away and Alphabet is not a fleet. It is one engine - Search - towing a cloud that just learned to fly on its own, a lab full of expensive experiments, and a corporate center carrying the AI bills and the legal ones. For a decade that design looked like ambition diversifying into the future. In 2024 it started to look like exposure: nearly every dollar of profit, and therefore nearly every bet on what comes next, depends on a product that a court has now declared illegal. The conglomerate was always a costume. The real company is a single magnificent engine - and someone with a gavel is finally reaching for the throttle.