Can Anthropic 10x Again?
For questions about this article or underlying research, email help@trefis.com.
Here’s a sentence nobody would have believed two years ago. Anthropic already did a 10x this year.
At the end of 2025 it was running at about $9 billion of revenue a year. By April, $30 billion. By May, $47 billion. By late July, $65 billion. In August, investors expected $100 to $120 billion by December. That’s 11 to 13 times in twelve months.
Its IPO papers show the same thing in plain accounting. Revenue in the first half of 2026 was $16.3 billion, up from $1.2 billion a year earlier.
So can it do it again?
What Would A 10x Look Like?
First, which 10x do you mean? There are two, and they’re very different.
A 10x in revenue. Say Anthropic finishes 2026 with about $60 billion of revenue for the year. Ten times that is $600 billion in 2027. Alphabet, all of it, Search, YouTube, Cloud, everything, is running at about $510 billion a year. So a revenue 10x means Anthropic out-sells Alphabet in 2027. In one year.
A 10x in the stock. If Anthropic lists at $1.5 to $2 trillion, a 10x makes it worth $15 to $20 trillion. That’s about four to six Alphabets.
Neither is impossible in a world where AI does most office work. But neither is a forecast. It’s a wish.
What Anthropic Itself Expects
Now the most useful number in the whole IPO filing. It isn’t revenue. It’s how much computing power Anthropic has already promised to buy.
About $417 billion signed in just the first five months of 2026, with contracts starting in 2026 and 2027. And up to $120.5 billion with SpaceX, at about $3.1 billion a month. Nobody signs those contracts for fun. They’re Anthropic’s own forecast, written in money.
So read them like one. Today compute costs Anthropic about 55% of revenue. If the $417 billion is used over roughly seven years, that’s about $60 billion a year. At 55%, that much compute supports about $110 billion of revenue a year. Roughly where investors expected Anthropic to end 2026.
Anthropic isn’t betting on another 10x. It’s betting on about 2x to 3x. That would still be one of the fastest growth stories ever told.
Could the stock still 10x? History says even the best rarely do it fast. Google went public at $85 in August 2004. It peaked at $741 in November 2007, about 8.7 times in three years. That was Google, starting at $23 billion. Anthropic starts about 65 to 87 times bigger.
The 10x already happened. The question for 2027 is 3x. Revenue of $120 billion to $180 billion in 2027, 2x to 3x this year, would be a triumph, and it’s roughly what the compute contracts imply. A 10x would need AI to replace a meaningful slice of the world’s office work in a single year.
What To Watch
What would tell you it’s working? Two numbers, every quarter. Customers’ spending growth (now about 9x a year; it can’t stay there, but how slowly it falls matters). And compute as a share of revenue. If that keeps falling from 55%, Anthropic is getting the AI equivalent of Google’s cheap servers, and the bull case gets a lot stronger.
And the hopeful version? The 10x that matters may not be in dollars at all. The price of AI keeps falling. If a task costs a tenth as much next year, Anthropic can do 30 times the work for 3 times the revenue. That’s how computing has always grown: cheaper, then everywhere.
For the side-by-side with Google’s IPO, see Anthropic IPO vs Google IPO.
What A 10x Can Cost You
A 10x helps. Losing half along the way doesn’t. Nobody can call where Fed rates go in the next 6 to 12 months, or whether 2022 repeats. That year the S&P 500 fell 25% from its peak, Nvidia 63% and Meta 74%. For families with $1 million or more to invest, splitting money wisely between growth and defensive or value stocks can lead to a far better outcome than riding one winner. Talk to a Trefis partner firm in a free 15-minute call.
Spot an error or want more like this? Tell us in 10 seconds.