Anchors Aweigh for the SpaceX IPO

On 12th June, SpaceX went public on the Nasdaq stock exchange in the largest Initial Public Offering (IPO) in history. The rocket/social media/AI/compute company’s stock opened at $135 per share, closed the day at $161, and went on to hit a peak of $225.64 on 16th June (Yahoo! Finance). Since then, the stock has dropped to $159.72 (at the time of writing on 29th June), a 29% pull back. That still gives the firm a total market cap of $2.08 trillion, trading at a P/S multiple of around 111x.

The SpaceX IPO, these ensuing stock fluctuations, and in particular, that juicy multiple, have, like a Falcon 9 rocket, streaked across the collective consciousness. It’s generated takes as wildly varied as either proving the business case for the vast amounts of AI and compute expenditure that has defined the agentic AI era, or acted as a distilled example of dot-com-style irrational exuberance that precipitates a rocket-fuel driven stock market meltdown. The SpaceX IPO offers a lens into these two competing interpretations of the state of capital markets today; and I want to pick apart some of the hyperbole at these two poles. I do not believe we are living in the shadow of an AI bubble and imminent market crash of the scale that internet startups generated at the turn of the millennium. However, there is, undoubtedly, a galactic-sized bathtub full of froth, and the SpaceX IPO, and upcoming OpenAI and Anthropic public offerings, highlight market tendencies that echo the ‘irrational exuberance’ Robert Shiller identified in the early 2000s. However, there are concrete foundations beneath this froth that anchor the market in a way that was not present in the dot-com era. An oncoming but contained correction in AI stocks is likely. A total market crash à la 2000, hopefully, is not.

Once the scale of the SpaceX IPO became clear, and in particular the market capitalisation that the firm’s target price would generate, the one datapoint that made analysts blush was the multiple. At a $2.08 trillion market cap, with FY2025 revenue at $18.7 billion (WSJ), its current P/S multiple sits at around 111x. This is extremely stretched compared to other established tech stocks. Based on 2025 revenue, Alphabet trades at a P/S multiple of around 11.6x, Apple at around 11x, and even Nvidia, ‘only’ trades at around 24.2x. SpaceX makes these look quaint.

Using some solid projections, we can also estimate the multiples that OpenAI and Anthropic might trade at in their upcoming IPOs. OpenAI will likely target a $1 trillion valuation, and with roughly $13 billion in FY2025 revenue confirmed by CFO Sarah Friar (Investing.com), its P/S multiple will likely sit at around 76x. Anthropic will also target $1 trillion; estimates for 2025 are around $9 billion, giving it a multiple of 111x, however, given Anthropic’s explosion in the enterprise agentic AI market over the last 6 months, it might make more sense to use its projected annualized run-rate of $30 billion instead (Bloomberg). This would generate a multiple of 33x.

Whichever way you slice it, these multiples show just how much investors are prepared to spend to get their foot in the door, and it’s considerably more than the premium already placed on the more established tech trades. For some, this overpricing has triggered traumatic memories of the early 2000s stock market crash, where several dot-com flameouts were trading at similarly stretched multiples. In March 2000, at the peak of the bubble, Yahoo! hit 100x, and Cisco was the world’s most valuable firm with a $500 billion market cap and a multiple of 26x (Yahoo! Finance).

A pure comparison of P/S dot-com multiples between now and 2000 would indicate the existence of a bubble; dot-com flameouts traded at multiples of around 30x, which was extreme at the time, and so, the 100x ratio of AI startups today means that we are squarely in bubble territory. But the multiples alone do not tell the entire story, and you need to dig through the froth to get a feel for the business foundations underneath.

These business fundamentals are the core difference between the dot-com bust and now. At the turn of the millennium, there were thousands of companies across the entire market trading at wild multiples. The froth was systemic. Currently the Nasdaq 100’s estimated 12-month forward P/E ratio (earnings instead of sales) is 26.3x (WSJ). In March 2000, it hit 60x (Intuition Labs). Stretched multiples such as SpaceX are concerning, however they are not replicated across entire indices as they were in 2000. Instead, markets today are grounded by mega-cap tech stocks with established revenue drivers and profit generators which were not replicated in the dot-com bubble. Nvidia’s market cap of $4.7 trillion is huge, larger in nominal terms than any dot-com company, but with FY2025 revenue of $215.9 billion with a gross profit margin of 71% (net margin ~53%) it can back this up. Similarly, other Magnificent 7 stocks, have established, profitable business cores to turn to in case their recent AI capex outflows don’t generate returns. Alphabet and Google’s ads businesses, or Amazon’s cloud computing, are insulators against a wider AI bubble today.

This also applies at the frothier end as well. SpaceX, Anthropic and OpenAI are, or will, trade at extremely stretched multiples for unprofitable businesses. But what distinguishes them from the dot-com darlings is that there are proven business cores at their foundation, built on genuinely innovative technology. For SpaceX, this is Starlink, which generated $11.4bn in FY2025 revenue (61% of total revenue), an operating profit of $4.4bn (BitMEX), and has, for now, cornered the low-earth-orbit broadband market. OpenAI, despite $13bn of revenue in 2025, doesn’t expect to turn a profit until 2030. However, in ChatGPT, it has built a category-defining product that has dominated the consumer market and hit 1 billion monthly active users (Reuters).

Similarly, the enterprise market has undergone an enormous process of AI penetration, largely led by Anthropic’s agentic Claude CoWork and Claude Code products. 87% of large enterprises report to have implemented AI and enterprise AI spending hit $37 billion in 2025. Underneath all the froth then, are genuinely innovative products with demonstrable business value and unprecedented levels of adoption. The internet technology that powered the dot-com bubble was never commercialized as effectively and the startup cohort of 2000 never converted the technology into products with the same business and consumer application as AI today. Many of them, such as Pets.com, were mostly spending on marketing fluff with no product core at the centre. It was largely smoke & mirrors.

High multiples are always a bet; but a bet on a high multiple with real, revenue-driving products and projected future expansion is a different bet to a bet on a vague story with little to no revenue. It doesn’t make the SpaceX or Anthropic multiple ‘right’ today, but it does cast questions over the dot-com style bubble narrative.

The bubble proponents are not entirely without cause though, and the SpaceX IPO contains the tremors of a potential oncoming correction. After its 29% pullback, the volatility of the SpaceX stock has been touted as an example of AI-driven instability. However, this volatility is a symptom of deeper concerns I have around the float mechanics themselves. This was an extremely thin float; SpaceX floated about 4% of their shares in the IPO. This bakes volatility into the stock as a small number of enthusiastic buyers can purchase a relatively small number of shares, generating an outsized pop - which is exactly what happened. Similarly, modest selling can then create large price drops as there isn’t sufficient resting liquidity to absorb the demand change.

In his book, Irrational Exuberance, Robert Shiller analyses the ‘anchors’ behind stock markets, one of which is psychological. Our psychological anchors can slip extremely quickly around news events like stock price changes, as herd mentality creates mass panic. This is why the ‘breaking of a psychological anchor can be so unpredictable: people discover things about themselves, about their own emotions and inclinations, only after price changes occur.’ Razor-thin floats like SpaceX’s intensify a generalised atmosphere of unpredictability by making large swings more likely, and thereby slipping our collective ‘psychological anchors.’

What’s even more concerning than the volatility that the thin float bakes in, is that institutional guardrails and rules have been bent specifically to accommodate such a low percentage of the total stock. This highlights just how far not just retail investors, but also wider anchoring institutions, have been sucked into AI-driven irrational exuberance. Normally, there are float requirements that mandate a minimum of around 10% before a company can enter an index. This was specifically designed to mitigate the volatility fluctuations produced by thin floats. Both Nasdaq and FTSE Russell replaced this guardrail in favour of a raw dollar lower limit to accommodate SpaceX’s early entry into the indices (and likely OpenAI and Anthropic as well). Not only that, but they also removed the seasoning window, a mandatory 12-month waiting period between going public and entering an index, which gives time for prices to stabilise and data to accumulate on the stock’s performance.

Here, Shiller’s ‘anchors’ on exuberance are clearly loosening. He writes that with stocks, ‘people are weighing a story,’ the story that we tell ourselves about our own wealth and the quantity of it in relation to others. Fear of missing out on a stock market bonanza deeply impacts this ‘story,’ and worryingly, this appears to be loosening psychological anchors at not only an individual retail investor level, but also at an institutional one. These institutions are designed to guard against the ‘pervasive human tendency toward overconfidence in one’s beliefs’, yet it appears that frothiness surrounding AI stocks right now is such that these institutions that are supposed to anchor the moment to the rules are now anchoring the rules to the moment.

The willingness of market infrastructure to bend towards tech darlings is exactly the kind of institutional exuberance that signals an oncoming reckoning of some sort. Extremely stretched multiples and hype-driven IPOs are threads that run from the dot-com bubble of the 2000s to today. However, there are differentiators; critically, the business fundamentals beneath the froth mean that this reckoning will likely take the form of a tech-isolated multiple correction, rather than a total market meltdown in the vein of 2000. Mega-cap tech stocks with established profit moats and the deep enterprise and consumer market penetration of AI products give the AI bull run a grounding that was simply not replicated in the dot-com internet boom. So whilst market ‘anchors’ are undoubtedly loosening, we are not quite completely adrift yet.