Alphabet
GOOGL
Open
Platforms
- +2.2%unrealized, USD
- 333.18average cost, as supplied
- 340.67last · 2026-08-20
The record
Thesis
Written 2026-08-21.
Alphabet is the most complete company in the world, and I want to own it for a very long time. Almost nobody else holds every layer of the AI stack at once: its own silicon in TPUs, its own frontier models, its own data centres, its own distribution through Search, Chrome and Android, and its own demand through Cloud and YouTube. Most companies are renting at least one of those from somebody else. Alphabet owns the lot, and funds it from operations.
Every part of the company is growing
revenue growth versus the same quarter a year earlier, per cent
Scroll the chart sideways →
Show the numbers
| Measure | Value | Note |
|---|---|---|
| Google Cloud | 82% | $24.8bn of revenue |
| Search and other | 17% | $63.3bn of revenue |
| Subscriptions and devices | 15% | $12.9bn of revenue |
| YouTube ads | 13% | $11.1bn of revenue |
No single bar on that chart is holding the company up. That is the thing I actually want. Total revenue grew 24% to $119.8bn last quarter and operating income grew 30% to $40.8bn, but the reason I am comfortable holding this for a decade is that the growth arrives from four different places at once. A company with one engine has one thing that can break.
The moat around Search is the widest I know of. Search is the default verb for finding things, and the advertising system bolted to it is the finest cash-generating machine ever built. At $63.3bn in a single quarter it is still larger than Cloud, YouTube and subscriptions combined - and it is still growing at 17%. That is not a business being competed away; it is a habit, a distribution network and a two-sided market that took twenty-five years to assemble.
Cloud has stopped being the side business. Revenue grew 82% to $24.8bn, and that rate has risen every quarter for a year: 32%, then 34%, 48%, 63%, and now 82%. Operating income went from $2.8bn to $8.8bn, so it is scaling profitably rather than buying revenue. The backlog - work contracted but not yet recognised - stands at $514bn. That is the figure that makes the spending below look like a response to demand rather than a bet on it.
And it is still expanding. Gemini reached 950 million monthly users. Waymo, most recently valued at $126bn, is a genuine option on a category that does not exist yet, and Alphabet is one of very few companies that can fund something that ambitious out of petty cash.
Cash the business made, and cash it spent building
most recent quarter, billions of dollars
Scroll the chart sideways →
Show the numbers
| Measure | Value | Note |
|---|---|---|
| Cash from operations | $39.1bn | generated |
| Capital expenditure | $44.9bn | spent building |
The capital expenditure is the argument, not the objection. Alphabet is spending up to $205bn this year, raised from $180-190bn, and the chart above is what that looked like last quarter: more cash going into the ground than the business generated. I read it as a company converting cash into an owned asset base at exactly the moment that asset is scarcest, with a $514bn backlog telling it where the demand is. Compute is the input to everything it sells for the next decade, and it is buying rather than leasing it from a competitor. Businesses that compound for decades tend to be the ones that spent heavily when it was uncomfortable.
That is my view, and I hold it knowing the cost is real. Below is what I am watching to find out whether I am right.
What I am watching
I am bullish, not blind. These are the things that would change the picture, in the order I rank them.
Search query volume, as assistants absorb navigational queries. This is the one that matters most. Alphabet can win the AI race and still lose the economics if the query migrates to a surface that monetises worse than a page of search ads. Nothing in the numbers says this is happening yet - Search grew 17% last quarter - but it is the risk that would arrive quietly.
The antitrust remedies, which are live rather than theoretical. Judge Mehta's behavioural remedies took effect on 3 February 2026 - exclusive distribution contracts banned, limited search data sharing required. The court rejected structural relief, so there is no Chrome or Android divestiture, and both sides are appealing at the D.C. Circuit. I read the outcome so far as close to the good case. It is still a live matter rather than a closed one.
Whether the spending converts. One quarter of negative free cash flow is an investment decision. If it is still negative in two years with cloud growth and margin flat, the build-out was an expense rather than an asset. Right now cloud growth is going the other way, which is why I am relaxed about it.
Execution on models. Gemini 3.5 Pro was delayed for missing internal benchmarks. One slipped release is noise. A pattern of them, while rivals ship, would undercut the part of this thesis that says Alphabet wins AI rather than merely survives it.
Waymo's losses against its own targets. Other Bets is running about $8.2bn of operating loss for 2026, up from roughly $5.5bn, with paid rides short of the company's one-million-a-week goal. I am happy funding a real option. I want it to look like one.
What would make me wrong
- Search revenue per query falling for consecutive quarters, not just query mix shifting.
- Cloud growth decelerating while capital expenditure keeps rising - the backlog would have been a promise rather than a pipeline.
- The appeal going against Alphabet in a way that reaches default placements.
- Free cash flow still negative through 2027.
Latest movements
- 2026-07-30 — Opened at $333.18. Untouched since.