OpenAI Races for $30 Billion Before Anthropic’s Debut

SAN FRANCISCO, Oct. 19 — OpenAIOpenAI is racing to raise about $30 billion before its rival Anthropic’s expected stock market debut in mid-November, Reuters reported on Monday. The round could value the maker of ChatGPT at more than $1.4 trillion, according to two people familiar with the matter, and SoftBank Group, Abu Dhabi’s MGX and other existing OpenAI backers are in talks to join it, one of them said.

The same day, Bloomberg News reported that Anthropic was in talks to rent 460 more megawatts of computing power in West Virginia from Nscale, a London-based data-center developer, in a deal worth about $45 billion.NscaleAn AI cloud company that builds its own data centers. It filed on Sept. 18 to list its shares on the New York Stock Exchange under the symbol NSCL.

The two frontier labs are increasingly tapping a similar base of investors, people familiar with the matter told Reuters, and OpenAI wants to finish before Anthropic’s financial disclosures invite comparisons between them, one of them said. Some investors see OpenAI’s push as an effort to lock in capital from funds that might otherwise put more into Anthropic’s offering, the people said.Shared BackersMGX was among the significant investors in Anthropic’s $65 billion round in May, by Anthropic’s account, and that round included $5 billion previously committed by Amazon. Amazon also agreed to invest up to $50 billion in OpenAI as part of its March round, Reuters reported.

Bloomberg News reported the size and the price on Sept. 29, describing the talks as early and subject to change. What Reuters added was the clock: the round is expected to move faster than is typical for one of its size, the people said. OpenAI held meetings with long-only funds and sovereign wealth funds last week, Reuters reported, while Anthropic was hosting its own investor day. OpenAI and SoftBank declined to comment to Reuters; Anthropic and MGX did not immediately respond.

“OpenAI could want to set its own valuation and not have to worry about a comp (comparison),” Edward Best, co-chair of capital markets at the law firm Willkie Farr & Gallagher, told Reuters. Mr. Best, who is not working on the deal, spoke hypothetically. “It’s really a two-horse race in terms of being public.”

The Numbers

OpenAI is telling investors that its growth quickened after the launch of its GPT-6 Astra model, that its annualized revenue has reached $50 billion and that it expects $70 billion by year-end, according to one of the people. Investors familiar with the company told Reuters that it also needs capital for its large infrastructure commitments, and to bolster its balance sheet should weaker IPO markets further delay its planned 2027 listing. Sam Altman, OpenAI’s chief executive, had ruled out a public offering this year “as it addresses concerns about AI safety,” Investing.com wrote in September, summarizing a Bloomberg report. In March, a record $122 billion round valued OpenAI at $852 billion, a figure that included the new money. Bloomberg described the $1.4 trillion target in September as a valuation before the new capital; Reuters did not say on which basis its figure stood, so the two numbers may not be directly comparable.

Anthropic told investors that its revenue rose to $16.8 billion in the third quarter from $11.6 billion in the second, and Reuters has reported that its annualized revenue reached roughly $65 billion by July. But slowing month-to-month growth has raised concerns among investors. Some existing and prospective investors have argued for a valuation closer to $1.5 trillion for its debut, instead of $2 trillion, after its third-quarter results fell short of expectations, Reuters reported last week.

The two companies’ revenue figures are not directly comparable, because they count revenue differently. OpenAI records revenue net, subtracting partners’ shares up front, and has accused Anthropic of inflating its revenue by counting the full, gross value of contracts sold through third-party cloud providers, Reuters reported. Anthropic has told Reuters that it follows established accounting practices. Anthropic’s disclosures could also prompt investors to press Sarah Friar, OpenAI’s chief financial officer, for more detail on OpenAI’s revenue growth and financial outlook, one person said.

It’s really a two-horse race in terms of being public.

The Landlord

On Aug. 25, Anthropic agreed to pay Nscale up to about $44.6 billion for computing on Nvidia’s Vera Rubin systems at its Monarch campus in West Virginia, according to the prospectus Nscale filed for its own listing.MonarchNscale bought the campus, about 2,250 acres in Mason County, W.Va., in March. It says the campus will generate its own power, independent of the local grid, and can deliver up to 8 gigawatts. The new talks cover a second building there, with 460 megawatts on Vera Rubin chips, and Anthropic holds a right of first refusal on an expected third, Bloomberg reported, citing people familiar with the negotiations. A deal could be completed within days, those people said, though talks are continuing and terms could change. A second contract with Nscale, worth about $45 billion, would roughly double what Anthropic has committed to the data-center developer. Both companies declined to comment to Bloomberg; Anthropic did not immediately respond to Reuters, which reported the talks citing one person familiar with them.

The first agreements leave the financing to Nscale. Its prospectus, filed Sept. 18, says it must use its best efforts to obtain financing for the chips and the data centers, that it had not then obtained binding commitments for any of that financing, and that Anthropic may end, without liability, any tranche not delivered within a specified period after its expected delivery date. Nscale’s total contracted value has grown from $100 million to more than $103 billion in two and a half years, by its own count. Microsoft has pulled out of a planned Nscale data center after power connection delays pushed its expected supply date into the 2030s, Reuters noted. Nscale wants a valuation of about $30 billion in its offering; Bloomberg has reported that some investors are pushing back, TechCrunch noted.

Anthropic aims to go public in mid-November, IFR reported on Oct. 15, citing one person familiar with the process. The person said Anthropic planned to file its documents publicly with the Securities and Exchange Commission on Oct. 26, start its roadshow on Nov. 9 and price the offering the following week. Those dates are 14 days apart, while the commission’s guidance asks companies that first submit drafts confidentially to file publicly at least 15 days before a roadshow, so the reported timetable may change. An early-Tuesday search of EDGAR’s public full-text index found no Anthropic IPO filing.

The offering has drawn objections. New York City’s comptroller, Mark Levine, wrote to Dario Amodei, Anthropic’s chief executive, last week, asking the company to pause its IPO until it “fully and transparently addresses” concerns over AI safety, governance and the pace of development, IFR reported. Anthropic’s draft offering documents warn that advanced AI could pose “catastrophic or existential risks to humanity,” IFR noted, citing Reuters, and Mr. Amodei argued in a September essay that “we must slow the pace at which we improve the capabilities of AI models.”

Daily Cortex is written with AI models, including Anthropic’s Claude, and fact-checked with OpenAI’s GPT models.