This week we talk about initial public offerings, Anthropic, and investment flywheels.
We also discuss AI, financial entanglements, and backstops.
Recommended Book: Superconvergence by Jamie Metzl
Transcript
An initial public offering, or IPO, is what happens when a private company goes public and starts selling shares of itself, occasionally to just institutional investors like banks and sovereign wealth funds, but usually also to retail investors, which means normal people who buy stocks as part of their investment strategy.
Often private companies go this route, go public, because it’s one of the primary ways of gleaning new, oftentimes large inflows of money, and that money can then be used for investments in assets for the company, but it also allows employees who have shares in the company as part of their compensation to cash out, to get paid possibly a huge bonus for all their efforts, and it’s often a means by which executives garner huge paydays for themselves, because they can now sell their accumulated shares, or borrow against them, or because they have something in their contract that says they get x amount of bonus money or new shares if they take the company public, or achieve a certain valuation goal—and going public is a good way to do that.
This is also one of the primary ways investors in a company, whether that’s a bunch of smaller seed investors or big-name venture capitalists, to get their money back; the 10 or 100x-ing of their investment, getting ten or 100-times the money they put into the company, generally happens through an IPO, because it can balloon the valuation of that company, and it gives them a more conventional and reliable way of getting money back for their shares: they can just sell those shares on the open market.
So an IPO allows a private company to make shares of itself available to others, on scale. And the ‘initial’ part of initial public offering points at the early days of the process, during which the baseline price of a share of stock is established.
A fairly arcane and complex process has emerged around this, and it’s an entire industry at this point, with some institutions specializing in taking companies public, helping them get as high an initial price on that stock as possible. They also help them leap all sorts of regulatory hurdles set by the Securities and Exchange Commission, if they’re going public on a US exchange, at least, other bodies handle such things in other countries, and these going-public entities, called underwriters, which are usually investment banks, also typically have their own stake in the matter, earning compensation through a fee called a ‘gross spread,’ which is the difference between a discounted rate on the stock and what the stock is sold for on the open market on that first day it’s available.
What I’d like to talk about today is a wave of very closely watched unusual, impending IPOs that are coming later this year, and one of them in particular that looks to be even more unusual than the rest.
—
SpaceX, OpenAI, and Anthropic are three of the largest companies in human history; on paper, at least.
And that’s an important caveat. Market valuation for private companies is generally determined by how much investors are willing to spend on a percentage ownership of the company. So if you start a lemonade stand and I offer to buy 1/10th of that lemonade stand from you for $100, that implies, using this logic, that your lemonade stand has a valuation of $1000; 10 times that $100 that I offered to pay you.
Such valuations are also informed by independent analyses from outside experts and institutions. SpaceX, for instance, pre-IPO, is estimated to be worth somewhere between $780 billion and nearly $2 trillion, depending on who you listen to, based on their assets, their potential future earnings, and any advantages they might have in the markets in which they operate.
AI company Anthropic is estimated to be worth something like $965 billion, based on a May 2026 series H funding round, through which it raised $65 billion; based on that funding round, the calculations were done, and just shy of a trillion dollars is what the math says the company is worth, though some outside analyses say it’s worth a bit less than that, while others suggest it’s maybe closer to $1.4 trillion.
OpenAI, a direct competitor of Anthropic, is valued at about $100 billion less than Anthropic based on its most recent $122 billion funding round, but again, analyses put the company’s actual value, what people and investors would pay for it on the open market, all over the place.
Each of these companies have different variables acting upon them heading into a period in which it’s expected that all three will IPO.
OpenAI kicked off the current AI race, for instance, but it’s burning money at an incredible rate, and has yet to make a profit, losing billions per year, and will probably continue to lose billions each year for a while into the future.
Anthropic, on the other hand, offers a similar product as OpenAI, but is projected to post its first quarterly operating profit of just over half a billion dollars in Q2 2026, making it one of the first frontier-model-making AI companies to make a profit, as most of these companies are investing so heavily in research and infrastructure like data centers that they’re still in heavy cash-burn mode.
SpaceX is distinct from these other two also high-flying, cash-burning tech companies in part because of its colorful and controversial owner, Elon Musk, and in part because it’s a rocket launch company that also sells internet services beamed down to earth from satellites, and until recently, most of its reliable income has come from that single offering, selling internet access. But it also recently had X, formerly called Twitter, a social network, and an AI company meant to compete directly with OpenAI and Anthropic, called xAI, folded into it.
So it’s now a multifaceted company with several edgy, but somewhat mature and difficult to compete with offerings, most of which make no money, but all of which in theory at least kinda sorta orient around AI and other sci-fi goods and services.
The surge in interest and investment in AI over the past several years led to a pivot for most of Musk’s companies, and that led to the merging of the smaller xAI and X into SpaceX, which was the only really profitable company of that trio of companies, and that merging, until just recently, made SpaceX unprofitable, as well.
Because of the unprofitability and relative unpopularity of xAI’s offerings, like the controversy-ridden Grok chatbot, SpaceX has recently taken to leasing out its data centers to competitors, like Anthropic and Google, each of which are paying around a billion dollars a month to use some of SpaceX’s data center capacity, which xAI hasn’t needed, because of the unpopularity of Grok, for their own AI services. That, in turn, has suddenly made SpaceX a little bit profitable, which is important for reasons I’ll get into momentarily.
This portion of the US-based AI industry is kind of a tangle in many ways, all of these companies competing, but also intersecting and overlapping, often investing in each other and in the infrastructure that underpins them, while also being invested in by those same infrastructural entities. And these three companies’ IPOs are being seen as something of a weathervane, their success or failure, and the degree to which they succeed or fail hinting at the direction of this industry, and whether or not this is a financial bubble that will soon, or eventually, pop.
There are hints that those at the top of these companies are attempting to hedge their bets, in case their IPOs don’t do what they need them to do, or don’t do what they need them to do at the right magnitude.
Sam Altman, OpenAI’s also fairly controversy-ridden CEO, has been very close with US President Trump, and has reportedly been holding meetings about the possibility of the US government taking a significant stake in OpenAI, and maybe other AI companies as well. The idea here is that US funds, so taxpayer dollars, would be invested in these companies, and that would tie the companies more closely to the US government, which could be beneficial if these companies then increase in value, making the US government a profit on that investment. This would be beneficial for the companies, in turn, because they would basically be backstopped by the US government; the US would be more likely to help them stay solvent to avoid losing that invested capital, with its regulations and laws related to AI, but it would also make these companies too big and too important to fail, giving them a lot of leeway in how they behave and compete, or fail to, from that point forward. And if they do still fail, the US taxpayer would be paying for a significant portion of that loss while those in charge, investors and the higher-ups of these companies, would walk away with a bunch of money.
SpaceX is taking another approach to IPO bet-hedging, by asking top US stock indices, like the Nasdaq 100 and S&P 500, which track top stocks, ‘top’ designated by value, but also other metrics, usually related to stability and profitability, to ignore some of those other metrics and allow SpaceX entrance into their indices more rapidly than would typically be allowed.
These indices are meant, in part, to help protect investors from volatility. High-flying startups might surge at the beginning, immediately after their IPO, but then fizzle out when it becomes clear their fundamentals aren’t good, and they’re not actually a solid investment, long-term.
What SpaceX wants is to be allowed into this club of valuable, long-term profitable and stable companies, because it is big and flashy and might have the largest IPO in history. And if these indices don’t want to be left out of all that, the argument goes, they should allow SpaceX into their club, regardless of those long-time rules of admittance.
Nasdaq, which runs the exchange where SpaceX will be listed, agreed to a rules change in May of 2026 that will allow large private companies, like SpaceX, that go public on their exchange, fast entry onto the Nasdaq 100 list.
This change of rules was made exclusively for SpaceX, and it could have a significant impact on the company’s IPO, because many index funds and exchange-traded funds, ETFs, track the Nasdaq 100, which means they balance their portfolio based on what’s in the Nasdaq 100, keeping things relatively or absolutely proportionate to that fund.
That means because of this change, a lot of everyday, passive investors, who have their retirement funds and pension plans and even their personal portfolios in index funds and ETFs that track the Nasdaq 100 will automatically end up holding some or a lot of SpaceX stock, despite it being an untested, new, currently unprofitable company. Some of these funds are automatically managed and will just buy SpaceX because that’s what they’re programmed to do, and others are managed by humans, but because they’ve promised their customers to keep their funds aligned with the market, more money going into SpaceX means they’ll be inclined to join the club and buy a bunch of SpaceX, as well. And because of how this works, the more funds buying SpaceX stock, the more funds will be required or inclined to buy; it’s a sort of stock flywheel.
That exposes all these investors to more volatility of the kind they maybe hoped to avoid by tracking this index, which isn’t supposed to be volatile. But SpaceX’s Musk was able to demand this change because, again, this is looking to be the biggest IPO in history, the company valued at $1.77 trillion dollars after the IPO. As a result, he can demand these sorts of things, and typically be listened to.
Some other stock market indices have also said they would allow quick entrance to their lists for SpaceX and possibly OpenAI and Anthropic, as well.
The S&P 500, however, after assessing the possibility of quick entry, has rejected the idea, saying it won’t bend its rules, no matter how big these three IPOs are looking to be. That means folks with money in S&P 500-tracking funds will be protected from that initial volatility.
That said those recent deals SpaceX made with Anthropic and Google nudged them into profitability, and if they can maintain that profitability for a year, post-IPO, then they’ll be able to enter the S&P 500. And because Google’s parent company Alphabet is a significant investor in SpaceX, they’ve already made money, on paper, on the deal they made with SpaceX for that datacenter capacity, paying out less than they’re making back in valuation.
So that tangle of relationships is likely to continue to enrich those in charge of these companies, and those who hold a bunch of shares of their stock, but it’s also likely to get more of these massive, but volatile companies into ostensibly less-volatile indices, faster, which could have repercussions for the one-third of private US wealth that is currently invested in the stock market.
Show Notes
https://www.investopedia.com/terms/i/ipo.asp
https://en.wikipedia.org/wiki/Initial_public_offering
https://www.bloomberg.com/news/articles/2026-06-05/spacex-s-75-billion-ipo-draws-more-orders-than-shares-available
https://www.marketwatch.com/story/elon-musk-needs-the-cultish-support-of-everyday-investors-to-pull-off-the-massive-spacex-ipo-08e7ea49
https://uk.finance.yahoo.com/news/spacexs-ipo-dream-runs-into-wall-streets-oldest-test-chart-of-the-day-114542191.html
https://www.cnbc.com/2026/06/05/tech-download-anthropic-ipo-ai-valuations.html
https://www.nytimes.com/2026/06/05/technology/spacex-indexes-401k.html
https://nypost.com/2026/06/04/business/one-third-of-americans-wealth-is-now-tied-to-the-stock-market-a-record-high/
https://arstechnica.com/tech-policy/2026/06/sp-500-blocks-fast-spacex-entry-wont-waive-rule-for-unprofitable-ai-firms/
https://arstechnica.com/tech-policy/2026/06/we-pissed-off-a-lot-of-people-giant-data-center-plan-cut-50-amid-protests/
https://www.notus.org/technology/trump-ai-stake-openai
https://techcrunch.com/2026/06/05/google-will-pay-spacex-920m-per-month-for-compute/
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit letsknowthings.substack.com/subscribe