Anthropic investors have been kicking the tires on what could be the most valuable initial public offering in history. A handful of the frontier lab’s backers confirmed to the Financial Timesthis week that they expect privately held Anthropic to go public in October with a targeted valuation of $2 trillion or higher, which easily eclipses SpaceX’s record-breaking $1.77 trillion IPO in June.
That valuation would more than double the $965 billion the company was worth when it reported a Series H funding round in May. Bloomberg, meanwhile, has reported that Anthropic is also in talks to buy the AI startup Decart AI for $6 billion. Anthropic filed for an IPO confidentially with the Securities and Exchange Commission in June, but has not publicly set a timeline. Rival frontier lab OpenAI followed suit shortly after Anthropic, but is not expected to IPO until 2027.
The awkward part of all this, though, is that Anthropic isn’t making money yet. Across the Nasdaq 100 universe, the index of large-cap tech companies Anthropic would join post-IPO, the average company trades at roughly 34 times trailing earnings and 25 times forward earnings. At those multiples, a $2 trillion Anthropic would need to post annual profits in the neighborhood of $59 billion to $79 billion to keep pace.
It could be getting closer but the Claude chatbot purveyor led by Dario Amodei still has a long way to go. The Wall Street Journal reported that Anthropic’s second-quarter 2026 revenue would more than double to $10.9 billion, while the company would for the first time post an operating profit. But operating profit is not the same as net income. Operating profit tells investors whether the business is covering costs like salaries, compute, and research, but it doesn’t account for interest on debt or taxes. Net income is what’s leftover after all of that is subtracted out. And for a company like Anthropic, with all the needs that go along with sustaining a bleeding-edge frontier lab, the distance between operating profit and actual bottom-line profit could be substantial.
Avery Marquez, director of investment strategies at Renaissance Capital, said approaching that threshold of a profitable bottom line will be key to make Anthropic’s valuation palatable to public investors.
“Just seeing the [$2 trillion] number, it’s definitely jolting,” she said. “Reaching near operating profitability will at least be something that in my mind makes this very large valuation maybe not seem so crazy.”
At $2 trillion, Anthropic would be keeping company with six other businesses in the world with valuations that size or more plus Broadcom, which has been floating near the $2 trillion mark since first crossing it earlier this year. But just look at the profits of those six firms.
Nvidia’s valuation is more than $5 trillion and it earned $120.1 billion in net income last fiscal year on $215.9 billion in revenue. Alphabet, at $4.55 trillion, made $132 billion on $403 billion in revenue. Apple, at $4.49 trillion, earned $112 billion on $416 billion in revenue. Microsoft, at $3.7 trillion, posted $133.7 billion of net income in the year ended June 30. Chipmaker TSMC, one of the most valuable companies outside the U.S., rounds out the group at $2 trillion.
Anthropic would be closest to Amazon, which booked $77.7 billion in net income in its most recent fiscal year, although a portion of its own profits are a function of Anthropic’s valuation. (Amazon’s most recent second quarter earnings show $62.6 billion of net income, and $53.4 billion of that was non-operating pre-tax income “primarily from our investments in Anthropic,” its earnings release states.)
What’s Going Right
Anthropic’s run-rate revenue went from about $9 billion at the end of 2025 to $47 billion by mid-May. Outside data shared by Salesforce CEO Marc Benioff estimated Anthropic’s run rate had reached $74.1 billion, surpassing OpenAI’s $41.3 billion. (Salesforce is an early investor and customer of Anthropic; neither company has confirmed the figures and Benioff shared data from TickerTrends.)
“What most impresses me about Anthropic (besides unprecedented revenue growth) is their enterprise hat trick,” posted Benioff. “The best model (Claude), the best coding agents (Claude Code), & the best productivity tool (Cowork).”
What most impresses me about Anthropic (besides unprecedented revenue growth) is their enterprise hat trick: the best model (Claude), the best coding agents (Claude Code), & the best productivity tool (Cowork). Salesforce is proud to be an early investor, customer, and partner. pic.twitter.com/PE1DIGTOnN
— Marc Benioff (@Benioff) July 23, 2026
The two rival frontier model developers, OpenAi and Anthropic, are comparable to one another, noted Marquez, which means whichever company files first sets the benchmarks that every company that follows has to measure up against.
Anthropic can tout its enterprise customer base, which is stickier and compounds more predictably than individual consumer subscriptions, which is where OpenAI’s ChatGPT has the name-brand recognition advantage.
Then there’s compute. Evan Schlossberg of Neostellar Capital Corp., whose fund holds a position in OpenAI, said the supply side of the business is the second thing he’ll turn to once he has an S-1 prospectus filing for Anthropic, right after he looks at their definitions for revenue and how it defines key financial metrics.
“The question is, what is Anthropic’s source over the next 18 months, 24 months, of how much compute they will be able to access at any given time,” said Schlossberg. “Do they own that? Are they leasing it? Is it short-term leases? Is it long-term leases?”
The answers will be revealing. A company that owns its servers or has locked-in, long-term leases has predictable costs and can squeeze performance out of its fleet of chips, making each dollar of revenue less expensive to deliver. Short-term leases can lead to spiking costs, scarce supply, and could leave Anthropic at the mercy of another company’s pricing.
“If you’re able to get slightly better margins out of the hardware you own, what is that showing in terms of overall margin,” added Schlossberg.
For its part, Anthropic has been locking in capacity. It has deals with Amazon, Google, Broadcom, and GPU access through SpaceX. If the Decart deal closes, it would also bring in software that helps chips run more efficiently, and an inference optimization team that could plug and play in Anthropic’s organization. Marquez said lining up an acquisition before a road show is pretty common in the tech-IPO world. Companies do it so the pro-forma financials already reflect the acquisition, even if the numbers describe a combined business that hasn’t actually operated together yet.
What this does to OpenAI
Schlossberg said the $2 trillion valuation for Anthropic is “exciting” news as an OpenAI investor.
“If you see strong, credible demand for investments in Anthropic and escalating premiums on that revenue, it would speak to a reasonable analogy that you’re seeing similar market trends for OpenAI,” he said. “It’s the same sort of bull or bear case.”
He’s also not worried about one lab slide-tackling the other.
“If everyone in the world wanted to switch over to OpenAI tomorrow, or Anthropic tomorrow, or Gemini tomorrow, I don’t believe those companies even have the compute to satiate that,” he said. “It seems less likely that you’re going to have one model intelligence company dominate the global demand for intelligence.”
Marquez sees Anthropic’s valuation turning up the heat for OpenAI. Whether it IPOs first or second barely matters for Anthropic, but it matters a lot for OpenAI, which will be priced against a live competitor if Anthropic goes first as planned. Anthropic’s enterprise revenues are flattering but hundreds of millions of people use ChatGPT. OpenAI will likely have to answer the strategic question as to whether it will continue pushing more deeply into enterprise where Anthropic is strong, or if it will lean into scaling more individual customers and monetizing advertising or paid conversions, she said.
But OpenAI doesn’t necessarily need to beat Anthropic at its own game, noted Marquez, it just has to arrive looking comparable with similar growth and a credible path to profitability on an EBITDA basis. The hurdle Anthropic will need to overcome is establishing what financial metrics make sense for the company.
“The big hangup for the valuation is, what metrics make sense for this company?” said Marquez. OpenAI will not have that problem, but it will have a very clear peer for investors to use for comparison.
“I don’t think that’s going to deter OpenAI at all,” said Marquez. “But I don’t think it helps OpenAI for Anthropic to go first.”
The post Anthropic needs to bring in Amazon-style earnings to justify its $2 trillion valuation—but it’s barely turned a profit appeared first on Fortune.
