Elon Musk managed it with SpaceX. The question now is can OpenAI and Anthropic pull off the same trick when their long-anticipated IPOs finally happen at a time when the precariously thin skin of the AI bubble is finally seeping into the consciousness of commentators on Wall Street and beyond.

SpaceX raised $86 billion through its Nasdaq IPO earlier this year, rendering Musk a trillionaire in the process, something he might have mentioned once or twice since then. OpenAI’s estimated IPO value is reckoned to be somewhere between $730 billion and $850 billion, depending on how optimistic/delusional - delete as applicable - the valuer is. Meanwhile Anthropic’s comparable figure comes in north of $1trillion. Not bad for firms that have yet to turn a dollar of profit!

It’s an amazing/insane - again, delete according to prejudice - example of capitalism in action. But, and whisper this softly, what if something were to go wrong? What if those big numbers were just a Wall Street wet dream and the reality is a lot more grounded? How much would it take for that AI bubble to finally burst? And how badly would the rest of us be caught up as the mythical foundations come tumbling around the ears of the Tech Bros?

Inevitably that’s a lot of ‘what if’s, however valid the questions. For the moment, the appetite for such huge valuations remains high and investors remain ready to bankroll the sector,  As luri Struta,  Senior Research Associate at S&P Global Market Intelligence, notes, funding for gen AI companies has already shattered last year's record with the first six months of this year being the strongest ever:

These numbers, they are largely due to a massive concentration of funding into top-tier foundation model providers, like those provided by OpenAI and Anthropic. But we believe this is very likely to be the peak in transaction value for gen AI in private markets, at least for a period because we think a lot of these companies have already put to test the private markets in terms of how much they can raise and now are increasingly seeking to move their fundraising activity to public markets.

In fact such is the long shadow cast by OpenAI and Anthropic’s potential plans, that S&P conducts some research where it deliberately excludes them because they skew the numbers upwards - and this brings to the fore more of an alarm bell, albeit one not ringing that loudly just yet,  Looking at gen AI companies that don't own a frontier foundational model, there has been a sequential increase in funding in Q2, but an underlying annual decline. Struta says:

We think some of these declines can be attributed to factors like the largest foundational models have increasingly moved into different verticals, like enterprise, legal, coding and so on. This does make some VC investors avoid, or at least pause, funding for AI companies that are not at the frontier and within this environment, a foundation model, like we saw with Anthropic launching Cowork, could just come and crush it by attacking the specific verticals, like coding and legal. 

But it’s also a simple case of finite resources at play as well, Struta suggests:

A lot of the capital is being sucked in private markets by top foundation model players. They are growing at a never seen before speed in private markets.  As a result, I think investors are sort of pausing. They're not thinking to stop their investing activities in non-frontier foundational models, but they're definitely looking for a pause and trying to understand where this is leading before putting more money to work.

So what are investors looking for when they contemplate what horse to back? Is it more sophisticated that a Gordon Gecko-esque ‘greed is good’ philosophy updated from the 1980s? Struta posits:

They're looking for technology that is unique and cannot be replicated easily. This is increasingly harder and harder to do, but still can be done, especially if your focus is small enough that it doesn't become an interest for a larger company or a large premier foundation model is not interested in it.  We've seen, for example, in the legal space, we do have two companies, one in Europe and one in America,  that are sort of the first two leaders in legal AI.

But even that sort of focus isn’t a safe bet, it seems:

We also saw Anthropic moving into a lot of vertical markets, including legal with the Cowork product, so this is definitely becoming a risk for those investors. But I would say it also becomes this very old debate of best-of-breed or best-of-suite. So if you are a client and you need the best legal AI, then you're probably going to go to a specialized model. If you are happy with one that maybe it's not the best, but it's okay for you because it's also cheaper, then you're probably going to use sort of a model that's not specialized.

There’s precedent to be drawn from elsewhere here, Struta suggests, pointing to the tussle between Microsoft and Lotus back in the 1990s:

Nobody remembers about Lotus, but they were the first one to produce this office suite and they're basically killed by Microsoft.  The same thing happened with Internet Explorer and Mosaic. Mosaic was also lost in the end and Internet Explorer won. More recently, we see this, for example,  in Microsoft Teams competing with Zoom . Zoom is still here. Zoom is more expensive, but as a customer, if you need better features, then you probably choose Zoom. If you're happy with best-of-suite, if you don't need to make presentations all the time, then you’re probably fine with Microsoft Teams.

Whatever does happen in the coming months with OpenAI and Anthropic’s IPO intentions, there will inevitably be a knock-on effect on the wider market. Wall Street’s twitchiness and volatility around tech stock prices being impacted by the smallest piece of AI-related news or hyped-up claim, is a regular feature of 2026 - see diginomica passim. 

The mega IPOs are important signals, admits Shari Mager, National Capital Markets Leader & Partner at KPMG, but they’re not perfect proxies for the entire market:

They do act as market bellwethers. They are providing very valuable data points. So think investor appetite, valuation discipline and risk tolerance. Companies like SpaceX and leading AI firms, they do attract outsized attention, but their scale and their market position make them unique. It's important for companies to view them as indicators of sentiment rather than direct templates for the broader IPO market.

And in terms of the health of the market, that is still strong, says Mager, albeit extremely selective:

The thing that companies in the space need to remember is there is extremely strong demand, but it's concentrated around AI, digital infrastructure, certain health care sectors and fintech with other areas we're seeing secular growth drivers. So I guess what I would say is the mega IPOs are definitely telling us the market is open, but it's not open for everyone because investors are being highly selective, and they're rewarding companies that can combine scale, strong fundamentals and really a compelling long-term growth story.

Of course, assuming that OpenAI and Anthropic do replicate Musk’s success in coming to market on such high valuations, and that is still the most likely outcome, then life changes for them as public companies., particularly when it comes to transparency and accountability. Mager explains:

The level of scrutiny changes overnight once you go from private to public. Private companies generally have much greater flexibility in how they communicate performance and strategy. But once they're public, management teams are often facing quarter-by-quarter accountability from the shareholders, the analysts, the regulators, the media. There are so many stakeholders that are going to be focused on them,  it really requires a much higher level of operational rigor. So if you think about expectations around forecasting and execution, they become much more demanding. 

And from a Wall Street that is already prone to shout ‘show me the money’ when another tech firm gives it the tough love message that CapEx budgets are set to soar to fund AI expansion, that means:

There's definitely going to be heightened pressure for AI companies to prove the ROI. I think governance becomes a very strategic differentiator. And honestly, maintaining growth while they're public is probably going to be one of the more difficult asks of a newly public company compared to when it was private.

One last piece of advice that Messrs Altman and Amodei might want to take on board as they contemplate making the leap from the shadows of the private markets:

I would say what they need to remember is going public is not the finish line. It's really the beginning of a new level of accountability. The ones that are going to be successful are the ones that compare their innovation and growth with strong governance, transparency and frankly, consistent execution.

The East India Tea Company and the South Sea Bubble.

I don’t know why that popped into my head, but...