“Good morning. Andrew here. Anthropic is now more valuable than OpenAI, something Silicon Valley wouldn’t have expected a year ago. We go behind the numbers — and outline the risks the new leader in A.I. faces.
Also: Lauren Hirsch has a scoop on deal talks between iHeartMedia and SiriusXM, which were put on pause. And Dell’s stock is flying, raising new questions about President Trump’s investments in the computer giant.
The numbers behind Anthropic’s rise
A year ago, the biggest and buzziest artificial intelligence start-up around was undoubtedly OpenAI. That’s no longer the case.
Anthropic has taken that crown with its $65 billion fund-raising round. How the company went from also-ran to 900-pound gorilla is a reflection of the current state of the A.I. race, and it raises questions about what comes next.
Where things stand: Anthropic is valued at $900 billion, while OpenAI was last valued at $730 billion, not including the $122 billion it had just raised.
Behind Anthropic’s ascent: The company’s Claude models were consistently highly regarded. But by November, Anthropic’s Claude Code programming tool had become exponentially better at automating software development, driving a surge in adoption.
Its models have continued to improve rapidly — and its Mythos model has proved so adept at uncovering cybersecurity vulnerabilities that it has spooked governments and companies around the world.
Here are three data points that help explain Anthropic’s leapfrogging OpenAI in valuation:
Anthropic’s revenue run rate — an extrapolation of projected annual revenue, based on a month’s financial performance — has jumped from $4 billion last July to $9 billion at the end of last year to $30 billion last month to $47 billion now. By contrast, OpenAI implied in March that its revenue run rate was about $24 billion; and The Information reported this week, citing unnamed sources, that the figure had risen just north of $30 billion. (Of note: OpenAI has disputed Anthropic’s revenue calculations in the past.)
Anthropic’s market share on OpenRouter, a platform that allows customers to dynamically switch between A.I. models, has outpaced OpenAI’s for many of the past 18 months. (Note: Tokens are the basic units of A.I. use.)
Anthropic has just outpaced OpenAI in business adoption, too, per Ramp, a provider of business billing and expense services:
But Anthropic’s lead could easily evaporate:
Anthropic has been forced to spend billions buying up computing power — including renting some from SpaceX — to address outages on Claude that have led to user complaints.
It’s still in a court battle with the Pentagon over its designation as a supply-chain risk to national security as the Defense Department reportedly tests out alternative A.I. models. (That said, Dario Amodei, Anthropic’s C.E.O., is patching things up with the White House.)
And Anthropic, whose models require more and more tokens, could take a hit as companies increasingly clamp down on A.I. spending.
The stakes are huge. Anthropic and OpenAI are racing toward potentially trillion-dollar I.P.O.s this year, giving them the ability to raise billions from public-market investors after seemingly maxing out the private markets.
Both companies are worried that there’s a finite amount of investor money for those offerings — and each wants to be the one that people bet on.” [1]
What are the Chinese doing? Let us to compare these numbers with similar DeepSeek numbers.
DeepSeek operates on a drastically smaller financial scale than Anthropic and OpenAI.
This means that DeepSeek isn’t wasting sky-high sums of our, and investors’ money.
While Anthropic boasts a $47 billion revenue run rate and OpenAI sits at $30+ billion, DeepSeek’s annual run rate is estimated closer to $220 million to $300 million.
A direct comparison of these financial and market metrics highlights the key differences:
• Revenue Run Rate: DeepSeek's annualized revenue is estimated at around $220 million.
This is dwarfed by Anthropic's $47 billion and OpenAI's $30 billion figures, though DeepSeek makes up for it with hyper-efficient computational costs.
• Profitability Claims: DeepSeek does not share verified corporate revenue but claims immense structural efficiency. It reported a theoretical $562,027 in daily revenue against just $87,072 in inference costs, citing a theoretical daily margin of 545%.
• Corporate Market Adoption: While Anthropic and OpenAI battle for enterprise dominance (e.g., Anthropic’s 1,000+ enterprise clients spending $1M+ annually), DeepSeek claims over 26,000 corporate accounts globally, but many lean on its inexpensive, open-source-leaning APIs. They started, and they will keep working in DeepSeek ecosystem. Anthropic and OpenAI belong in history’s trash pile. They are like bumble bees – they are too heavy to fly, they don’t know it, so they keep flying.
1. DealBook: How Anthropic got so big. New York Times (Online) New York Times Company. May 29, 2026.
Komentarų nėra:
Rašyti komentarą