A backlog is money a company has already locked in but has not collected yet. Customers have signed real contracts promising to pay. The company just has not delivered the work, so it cannot count the money as revenue yet.
Think of it like a barber with a fully booked appointment book for the next two years. The chairs are not filled yet and no one has paid, but those appointments are signed commitments, not hopes. That appointment book is the backlog.
Google does not call it “backlog” in its reports. It calls it remaining performance obligations, usually shortened to RPO.
You do not need a finance degree or a paid data service. Do this:
That is the whole trick. The number analysts argue about on TV is sitting in a free document, in plain sight, every single quarter.
A backlog total by itself is only half the story. What matters is the conversion schedule: how fast those signed contracts turn into revenue. A contract that pays next quarter is worth more today than one that pays in five years, because the near money is more certain and arrives while the business is spending now.
Here is Google’s picture, from its reported numbers:
Why this matters: just over half of Google’s backlog converts within about 24 months. That is fast. It means the revenue is not some distant promise, it is already scheduled to arrive while Google is spending roughly $200 billion a year building the data centers to serve it. When you check any company, always ask the same two questions: how big is the backlog, and how much of it converts in the next 12 to 24 months?
This is where companies try to blur the line, and where most beginners get fooled.
A signed backlog is real. RPO only counts contracts customers have legally signed. Accounting rules back it up, auditors check it, and the company can get in serious trouble for inflating it.
“Pipeline” is talk. When a company says things like “strong pipeline,” “growing customer interest,” or “engagement is up,” none of that is signed. It means people are looking, not buying. It costs a company nothing to say, and it can vanish in one bad quarter.
The rule is simple. If the number has the words “remaining performance obligations” behind it, it is signed. If it is wrapped in adjectives, it is marketing.
A big backlog is not automatically a healthy backlog. Run these two checks before you get impressed.
If a huge share of the backlog comes from one or two customers, the backlog is only as strong as those customers. Google is the live example. Anthropic, an AI company, committed roughly $200 billion to Google Cloud over five years. That is about 40% of Google’s entire $514 billion backlog riding on one customer. It does not make the backlog fake, Anthropic’s revenue is exploding and it recently turned profitable. But it means you should be watching that one customer’s health almost as closely as Google’s own.
Compare two lines quarter to quarter. If the backlog is going down while the company’s spending is going up, the company is burning through its signed work faster than it is signing new work, and paying more to do it. That is a business eating its own future. The healthy pattern is the opposite: backlog growing alongside the spending, which is exactly Google’s setup right now.
On October 20, the Wall Street Journal published numbers from a leaked draft of Anthropic’s IPO filing. An S-1 is the document a company files with the SEC before selling stock to the public. Anthropic’s was still confidential, so treat these numbers as reported, not final. It is the first real look inside the one customer holding up roughly 40% of Google’s backlog. Here is what it showed:
Why this matters for the checklist: concentration is still the flag to watch, but the terms decide how dangerous it is. A concentrated backlog with pay-anyway terms, behind a customer whose sales just grew 45% in one quarter, is a very different risk from a concentrated backlog built on a handshake. When you run the concentration check, always read the terms next.
Run this on any company in about five minutes:
Open the latest earnings report, search “remaining performance obligations,” and write down the total backlog.
How much of it converts in the next 12 to 24 months? Near money beats far money. Just over half within two years is strong.
Does one customer make up a giant slice, like Anthropic’s roughly 40% of Google’s book? If so, that customer’s health is part of your homework. Then read the terms: does the customer pay even if it spends less than promised? Pay-anyway clauses make a concentrated backlog much safer.
Is the backlog growing while spending grows? Healthy. Shrinking backlog plus rising spending is the red flag. Compare against last quarter, not just the headline.
Four steps, free documents, no guru required. That is how you read a backlog like an owner instead of a spectator.
This breakdown teaches the skill. Inside the ThinkinGenWealth Discord, the Spilled It case-study tab has my complete Google thesis, the full numbers, the bull case, the bear case, and what would have to change for me to change my mind. Come learn it with us.