Frontier Model Competition Is Wide Open as Anthropic Closes the Gap
Plus, Google gets off easy in antitrust ruling, but startups get thrown a bone
The Week In Short
Anthropic’s $13 billion raise at a $183 billion valuation amps up its rivalry with OpenAI. Eric and Nayeema Raza lay out what to expect at Deus Ex Medicina next week.
Behind the paywall: Judge Amit Mehta ruled that Google can keep Chrome, but it must share its search data with smaller competitors. VCs slowed down dealmaking in August. OpenAI and Atlassian made big AI application acquisitions. The IPO market is wide open for September. Paramount hones in on The Free Press. Nick Clegg warns that Section 230 won’t cover AI chats. RFK Jr. gets grilled by the Senate.
The Main Item
Anthropic Shows Spectacular Revenue Growth But Still Has to Prove Its Consumer Appeal
Anthropic is on a tear.
The news this week that it had raised $13 billion at a $183 billion valuation marks a 10X increase from what the company was worth early last year — a reflection of revenue growth that’s extraordinary even by AI standards.
The company is closing the gap with market leader OpenAI and reshaping the competitive landscape of the AI industry.
As recently as a year ago, the LLM competition at the top of the AI food chain could have been described as Google and OpenAI/ Microsoft fighting it out, with Meta and its open-source approach, and xAI, as wildcards, and then a bunch of smaller players including Anthropic and Mistral and Cohere.
Now it looks more like a wide-open competition among Google, OpenAI, Anthropic, xAI, and Meta (assuming Mark Zuckerberg can get his expensive new hires working together), with Mistral and Cohere persevering as national champions in Europe and Canada, respectively, while Microsoft, Apple, and Amazon continue to work out their strategies.
Investors far and wide were scrambling for a piece of Anthropic: the company announced no fewer than 19 names as participants in the latest round, some of which were investing from multiple funds. A $5 billion fundraise became $13 billion, and could probably have been even bigger. It’s no wonder Anthropic was eager to limit SPVs that would bring even more people into the cap table.
The basic numbers explain the enthusiasm. Anthropic’s annualized revenue run-rate sat at about $1 billion at the beginning of the year, and the company is projecting a startling nine-fold increase over the course of 2025 to $9 billion by year-end. OpenAI, by comparison, had ARR of $5.5 billion at the end of last year, and while it hasn’t given a forecast, outside estimates peg it at $15-$20 billion by the end of this year.
By any normal standard, OpenAI’s revenue growth is pretty stellar — but appears to be only about half the rate of Anthropic’s, albeit from a larger base.
On top of that, for any investor comparing the two companies, Anthropic doesn’t face the overhang of a problematic non-profit conversion, which is shaping up as a real threat to OpenAI. OpenAI’s latest fundraise in March valued it at $300 billion and more recent secondaries put it at $500 billion, but much is contingent on the conversion.
Anthropic CEO Dario Amodei may not have the same sort of smooth moves as OpenAI CEO Sam Altman, who’s done an excellent job establishing himself as the public face of the AI revolution and played his political cards aggressively as a globe-trotting ambassador; Altman is positioned as both a bold creator of the future and a trustworthy steward of powerful technologies.
Amodei has stuck closer to the knitting. He’s offered stark warnings about job-destruction and cast Anthropic — founded in part because he and others working at OpenAI felt Altman wasn’t sufficiently attuned to safety — as the more responsible actor, but also focusing on how AI can help businesses.
That partly reflects a major difference in the trajectories of the two companies. OpenAI is first and foremost a consumer play, with individuals accounting for 70% of revenues. Anthropic is the inverse, with 85% of revenues coming from business customers rather than the consumer-facing Claude chatbot.
OpenAI is chasing the enterprise market hard, especially the red-hot coding sector, which is dominated by Anthropic. It’s less clear what Anthropic’s intentions are with its consumer offerings; its weak spot is the laggardly performance of Claude as a competitor to ChatGPT.
The enterprise market for AI is certainly very large, but it’s worth remembering that the internet-era companies that defined Big Tech — Apple, Google, Facebook, Microsoft, and Amazon — were all consumer-first. It’s hard to see Anthropic as a trillion-dollar SaaS vendor.
We don’t have much sense as to whether business customers are influenced by Anthropic’s commitments to ethical AI, though it does seem to have helped with employee retention; the company’s models are very good, especially for coding.
Amodei has also run into exactly the problem that OpenAI had as a mission-driven non-profit, namely that it takes many billions to compete in the LLM game, and you can’t always stand on principle. Anthropic’s latest round includes money from the Qatar investment authority — a climb-down for Amodei, who’d previously rejected taking Middle Eastern money.
At the risk of irritating everyone involved, one can think of the major LLM rivals as sitting along a metaphorical political spectrum, according to how they engage on “responsible AI.” Google sits as the dead-center of establishment thinking, with Altman veering center-right to be with the times, and Elon Musk’s xAI out on the far-right with explicit efforts to make Grok reflect his own worldview. Amodei, a San Francisco native, is the center-left contender, with more concern for the social consequences of unfettered techno-capitalism.
Making that approach a winner in the consumer market is the company’s next big test.
Newcomer Podcast
What To Expect at Deus Ex Medicina
Eric and journalist Nayeema Raza preview everything coming up at Deus Ex Medicina next week in San Francisco.
Some themes to keep an eye on: how longevity went mainstream, why precision medicine and novel bio are finally feeling real, whether or not AI can discover a new drug, and who wins the race to own the patient.
VC Chatter
Google's Mild Antitrust Penalty Still a Help for Startups, Investors Say
Google dodged its worst-case antitrust scenario and won’t have to sell off any products, a federal judge ruled, sending the search giant’s stock up 8%.