SpaceX, OpenAI, Anthropic and their giga-IPO dreams

MORE than a decade ago, before venture capitalists and buy-out barons began whipping out their cheque books, going public was the obvious choice for ambitious businesses.

A stock market listing offered startups both cash, courtesy of deep pools of capital, and cachet, conferred by a willingness to subject themselves to the scrutiny of millions of investors.

It was possible to attain a US$100 billion-plus valuation while staying out of the stock market spotlight. Lidl may have pulled it off with cheap groceries, Mars with confectionery, Cargill peddling the sort of stuff that goes into Mars bars, Gulf and Chinese natural resource firms extracting less-digestible commodities, Vitol and Trafigura trading these.

But if it was explosive growth you were after, the real rocket fuel was to be found in public markets.

Maybe it still is. Reports have recently surfaced that SpaceX, a rocketry and satellite firm founded by Elon Musk, was sounding out investment bankers about an initial public offering (IPO) as soon as next year at a valuation as high as US$1.5 trillion.

Earlier this month Anthropic, an artificial intelligence (AI) darling valued at roughly US$180 billion in a recent funding round, was also said to be exploring a listing. It may want to pip OpenAI, the US$500 billion byword for the AI boom, which is likewise rumoured to be increasingly IPO-curious.

Any of these deals could be the biggest in America since Alibaba, a Chinese e-commerce titan, sold US$25 billion of shares on the New York Stock Exchange in 2014. It could be the largest anywhere since Saudi Aramco, the world’s oil colossus, raised nearly US$30 billion in 2019 at a valuation of nearly US$2 trillion in Riyadh.

This sudden interest in public markets seems out of character for the trio. None has struggled to entice private backers. They have so far secured a total of nearly US$120 billion in funding.

Anthropic, which turns five in January, and even 10-year-old OpenAI look youthful by contemporary IPO standards; the typical American company now goes public at 16, four years later than a decade ago. SpaceX, which has shunned the stock market for 23 years, looks ancient.

In each case, a listing also presents thorny dilemmas.

The trade-off

The obvious reason for all three to pursue an IPO is access to ever more capital.

SpaceX is developing Starship, a reusable vessel capable of lifting up to 150 tonnes into orbit, roughly double the payload of any rocket currently in operation (including the company’s own Falcon Heavy). OpenAI has said it plans to invest perhaps US$1.4 trillion in computing power over the coming years. If the younger and smaller Anthropic wants to keep up, it, too, must splurge on data centres.

Even as the trio’s capital needs grow, the private market may stop expanding.

After swelling at a compound rate of 10 per cent a year between 2012 and 2021, global private assets under management have since plateaued at just over US$20 trillion. Limited partners are demanding that venture capital and private equity funds disburse some proceeds before asking for fresh cash.

Moreover, as the unlisted giants’ successive funding rounds balloon in size – OpenAI’s latest came in at US$40 billion, larger than any IPO ever – so do the typical contributions from what is by public market standards a puny number of participants.

At some point, that may become an unacceptable concentration of risk for all but the most fearless, or reckless, adventure capitalists out there. The public equity market, with a global market capitalisation of some US$130 trillion, is far larger and its investor base far less concentrated.

In its aggregated wisdom, however, it is also far more discerning. This discernment is what SpaceX, OpenAI and Anthropic would be trading off for cash. In none of their cases is the trade-off comfortable.

For SpaceX, the main headache would relate to corporate governance – or rather, if you treat Tesla as a listed proxy for Musk’s assorted ventures, a troubling lack of it.

At Tesla, he has had run-ins with regulators (who fined him and Tesla US$20 million apiece in 2018 for his remark on Twitter that he had “funding secured” to take the carmaker private) and judges (last year, one in Delaware voided his pay deal after concluding that Tesla’s board was “beholden” to him).

His growing fortune, which soared from US$470 billion to US$638 billion on Monday (Dec 15) after SpaceX was valued at US$800 billion in a secondary share sale, seems likely if anything to make him feel more emboldened.

At least SpaceX makes money. It is thought to be generating cash, and turned a profit for the first time in 2023.

For their part, OpenAI and Anthropic are burning enough to make a Starship launch pale by comparison. OpenAI is likely to make a net loss of around US$12 billion this year on revenue of a comparable size. It reportedly expects to torch another US$115 billion in cash before becoming profitable in 2030. Anthropic thinks it may break even a couple of years earlier, but will still bleed billions beforehand.

The slings and arrows of outrageous fortune

It will be a while before OpenAI, Anthropic and even SpaceX generate net profits comparable to those of Alibaba, which was US$4 billion in the black in the 12 months before its IPO, let alone Saudi Aramco, which booked US$111 billion, more black than which none has ever been.

The last time large profitless tech darlings went public, in the IPO wave of 2019, Uber needed four years and positive cash flow for its shares to rise reliably above their IPO price. You would be better off today having invested in the S&P 500 index.

The tech trio may think they have no choice but to list before big profits arrive.

SpaceX will look less exceptional once rival space firms such as Blue Origin narrow the technological gap. OpenAI and Anthropic face fierce competition already.

They could remain private and risk running out of the capital needed to stay ahead. Or they can go public and risk markets’ impatience over profits. Either way, their lofty valuations will soon be put to the test.

©2025 The Economist Newspaper Limited. All rights reserved