The author is a Reuters Breakingviews columnist. The opinions expressed are her own.

By Karen Kwok

NEW YORK, May 21 (Reuters Breakingviews) - A charismatic founder can keep a scrappy young company afloat. A domineering one can push a maturing giant astray. Sam Altman led OpenAI to become the trailblazer of the artificial intelligence revolution. As it heads for an initial public offering amid revelations of apparent internal strife and newly reinvigorated competition, though, his potential missteps are growing more glaring.

Nearly four years after ChatGPT’s release sparked a multi-trillion-dollar technology arms race, Elon Musk’s rockets-to-chatbots venture SpaceX is set to test investor appetite for the industry’s famously big promises and scant profitability. Its IPO is set for as soon as June, narrowly ahead of OpenAI, which Reuters on Wednesday reported citing two sources that it could seek a listing as soon as September. Rival Anthropic may finish the 2026 trifecta.

Having already raised $186 billion as a private company, according to PitchBook data, public markets are the natural next step. HSBC analysts estimate that OpenAI’s revenue will reach $64 billion next year, up from $34 billion this year. Such explosive growth alone should catch any investor’s attention. The temptation, if SpaceX notches its aimed-for $1.75 trillion valuation, is to use it as a kind of yardstick for OpenAI’s worth. After all, when SpaceX merged with fellow Musk-run company xAI earlier this year, it valued the second-string lab at $250 billion.

That’s a multiple of 78 times xAI’s expected $3.2 billion of revenue in 2027, using HSBC's estimates. Apply that to OpenAI, and it implies a fantastical $5 trillion valuation. Better yet, OpenAI's ChatGPT commanded a 54% share of generative AI traffic in April, versus xAI’s Grok at 3%, per Similarweb data.

Such giddy highs are unlikely versus OpenAI’s current $852 billion valuation. Altman is charismatic, but whether he can distort reality like Musk remains to be seen. And the company still loses money: The Information reported that OpenAI projects it will burn $25 billion this year alone. But so too do its direct peers. What matters is whether it can realistically turn a profit in the near future. BNP Paribas analysts, at least, estimate that the company could become free-cash-flow positive by 2031.

To get there, Altman has made several key bets: to go hard for consumers; to promise gigantic investments in obtaining as much computing capacity as possible; and to spread across a dizzying array of projects. Each now faces challenges, primarily from the meteoric rise of Anthropic. The developer of chatbot Claude is led by Dario Amodei, a former OpenAI researcher who left in 2020 over disputes about strategy, governance, and safety.

OpenAI likes to say that Altman’s quest to lock up or build as much computing power as possible, including through his $500 billion Stargate joint venture, puts his company ahead now that supply constraints are biting. AI models require vast racks of chip-stuffed servers to train and serve up responses from models. Anthropic long trailed OpenAI’s resources on this score. In March, Claude briefly went down under unprecedented demand.

Yet that’s because the chatbot had proved its mettle for programming. When the Claude Code tool took off, it redefined the industry. Ramp data tracking spending across about 50,000 business customers showed that, in April, Anthropic usage narrowly overtook OpenAI for the first time. Admittedly, gleaning this through credit-card charges provides an incomplete picture. But Anthropic’s valuation, too, is also crossing over: the company is raising funds at a $900 billion valuation, a source familiar with the matter told Breakingviews.

Worse, Amodei is getting help from another of Altman’s old friends-turned-foe. Elon Musk co-founded OpenAI. Back then, the company was a non-profit with a bizarre governance structure, idealistically designed to restrain it from accidentally destroying the world with super-advanced silicon intelligence. Musk and Altman ultimately fell out over OpenAI’s direction and competing claims for control.

The spurned billionaire has proved a constant complication. OpenAI said in May last year that it would convert into a for-profit corporation, helping to uncork more funding from investors leery of its highly empowered non-profit board and limits on their potential profits. Musk sued, claiming that this was effectively the theft of a charitable organization. He lost the trial, but testimony and proffered evidence still proved bruising for OpenAI, surfacing awkward internal messages and fresh allegations from former insiders that Altman is, as Musk’s lead attorney Steven Molo told jurors during closing arguments, “a liar.”

Now, Musk is also tilting the scales in the battle with Anthropic. Earlier this month, SpaceX announced a deal granting Amodei’s company use of its Colossus 1 data center. This is a huge amount of computing grunt to gain immediately, and unlike in-the-pipeline projects, the switch can be flipped quickly.

The perilously short supply of AI chips is also keeping industry prices high. According to Silicon Data’s LLM Token Expenditure Index, the cost of 1 million tokens — the base unit of chatbot output — has risen by 61% since early March. The users best-equipped to eat such costs are resource-rich corporate customers: exactly the clients Anthropic focused on early, and where it now appears to be outgrowing OpenAI.

To be clear, Altman’s crew is chasing behind with the launch of coding agent Codex, which saw revenue double seven days after launch and pushed up overall internal sales projections this year, sources familiar with the matter told Breakingviews. The fresh competition still makes OpenAI’s predilection for “side quests,” or oddball projects with no clear immediate payoff, more vexing. Video creation and sharing platform Sora ate up scarce computing resources only to shut down in disappointment. A $6.5 billion deal for iPhone designer Jony Ive’s startup io pushes into the cutthroat world of hardware.

All the while, OpenAI is frequently left pushing back on talk of internal discord, even more so after the trial. This includes tensions between Altman and CFO Sarah Friar, as reported by the Wall Street Journal, which the company strenuously denies. Whatever the case, it is unwelcome noise as investors try to assess whether they can justify fueling the CEO’s promised $600 billion in infrastructure commitments.

Then there’s Altman’s odd position of owning investments in firms like chipmaker Cerebras, which has signed a major deal with OpenAI. These cross-holdings have drawn scrutiny from Republican politicians.

All of this leaves OpenAI in an awkward middle ground. It still defines the AI boom, but seems vulnerable to the industry’s rapid changes in ways that are specifically driven by Altman’s decisions. An IPO is eminently possible. But the company’s strange governance, apparently dominated by its CEO, invites additional concern. He has been essential to the company’s rise. He may also make its next step forward more difficult.

Follow Karen Kwok on LinkedIn and X.