SpaceX, OpenAI, and Anthropic, the three major AI giants, are rushing toward an IPO. Which one is the most worth betting on?

Question

null Edited & Compiled by: Deep Tide TechFlow

Host: Josh Kale; Ejaaz Ahamadeen

Podcast Source: Limitless Podcast

Original Title: Money is Running Out for the Biggest IPOs in History

Broadcast Date: June 3, 2026

Key Summary

This episode focuses on SpaceX, OpenAI, and Anthropic nearly simultaneously rushing to IPO, discussing how AI infrastructure development is pushing private capital and tech giants’ balance sheets to the limit. The two hosts believe this is not just a single company's fundraising story, but an unprecedented concentration of capital: companies need more computing power, data centers, electricity, and chips, while public markets, index funds, and pensions are also caught up.

The program also compares SpaceX’s unverified space data center business model, Anthropic’s demonstrated revenue growth, Google’s continued external financing to bet on AI, and OpenAI’s capital needs for training and servicing models. Ultimately, both remain cautious about bubble risks but are generally optimistic: as long as computing supply cannot keep up with demand, AI infrastructure spending is more about building the next-generation technological foundation than a financial bubble in idle.

Highlights and Insights

Massive IPOs in the pipeline: Is AI construction draining capital?

“OpenAI, Anthropic, and SpaceX’s IPOs are expected to raise a total of $300k, surpassing the entire internet bubble’s total financing of $164 billion. And that figure is for three years of total funding, just for these three companies.”

“Why now? Why so urgent? In my view, the answer is simple: AI capital expenditure is becoming more expensive than these companies initially anticipated, and they are choosing to keep adding. Their free cash flow is no longer enough to support current demands.”

“Whether through debt leverage or other financing methods, we’ve already jumped into the abyss. If we don’t go all out now, we’ll have nothing.”

SpaceX rewrites IPO and index rules, pensions buy passively

“Through its IPO, SpaceX is essentially redefining market rules to meet its massive funding needs.”

“Over $30 trillion in passive 401(k) funds, i.e., retirement funds, will be forced to buy SpaceX stock, and at IPO valuation levels. Rough estimates suggest about 24% of the shares issued in SpaceX’s IPO will be absorbed by these passive funds. Such scale of passive buying is unprecedented in history.”

“Some indices only require a 5 to 15-day observation period; as long as the stock price after IPO can stay above a certain level for a few days, the company can be included. In other words, Elon Musk only needs to ensure the stock price remains at a certain level in the short term to easily qualify.”

“It (SpaceX) has not yet truly proven its revenue model. It claims ‘we will send AI data centers into space…’ But this business model remains unverified… it’s more like a ‘trust me’ promise.”

Anthropic’s revenue leap and IPO motivation

“They (Anthropic) reached this figure ($20 billion) in just the first month and a half of 2026. Recently, their annualized recurring revenue (ARR) has hit $401k. This is mainly thanks to the success of Claude Code, Claude Co-Work, and a series of enterprise contracts they signed.”

“Anthropic expects to make about $401k in profit by the end of this month. Compared to trillions in capital expenditure, it’s just a drop in the ocean, but it will be the first large AI lab to achieve this.”

“Their coverage in Fortune 10, meaning 9 out of the top 10 global companies, are using Anthropic, especially Claude Code. Their net dollar retention rate… has increased by 500%. In other words, these companies plan to spend five times more.”

Google’s $300k financing: not an IPO of an IPO

“Google founders Larry and Sergey Brin explicitly stated about a year and a half ago that they prefer to risk losing everything rather than lose the AI race. So they will keep spending until they find enough breakthroughs. They’ve returned to a founder’s mode.”

“Google, as a listed company, raised $80 billion… but about $30 billion of that might be used to cover taxes from employee stock vesting in the coming months. That is, a large part of the financing isn’t really for AI capital expansion… it’s somewhat desperate.”

“They (Google) are not focused enough. They’re doing agents, trying to improve coding models; building better general large models, and infrastructure like TPU; even selling TPU to competitors, yet their own training capacity for Gemini is insufficient… Even with the recent Gemini 3.5 Flash, despite spending so much, they still lag behind cutting-edge models.”

OpenAI and AI infrastructure: money not yet spent, bottlenecks in the physical world

“I (Ejaaz) might have a somewhat controversial view: the money these companies are about to raise and spend will ultimately be a good thing. I believe it’s not a bubble but will build necessary infrastructure.”

“We are now constrained by the physical world… No matter how much leverage or funding you use, you can’t necessarily spend it all because of regulations, data center physical construction speed, and silicon chip capacity limits. ASML is only one, Nvidia is only one, TSMC is only one. AI’s physical infrastructure is hard to scale. Until I see this bottleneck lifted, I don’t think we’re in a bubble.”

“GPUs from four or five years ago are now rented at even higher prices. They’re worth more… What we see now is the opposite: buyer demand is so strong that we lack enough silicon and compute power to meet it.”

“We are at a very special moment in history. Never before in American or capitalist history has so much capital and value been concentrated on the same idea… The US is re-industrializing in a significant way.”

Massive IPOs in the pipeline: Is AI construction draining capital?

Josh:

The three largest IPOs in history might file within weeks—SpaceX, OpenAI, and Anthropic. On the same day, Google also raised $80 billion externally to support its AI development.

Interestingly, the funding relationships among these companies have become quite complex; they are somewhat “feeding” each other’s balance sheets with their own capital. Over the past few weeks, markets have even begun to modify rules protecting passive investors to allow earlier IPO participation.

Currently, the largest construction wave in capitalist history is underway, so we must ask: Is there enough money? These companies are almost simultaneously choosing to go public, which is clearly no coincidence. The chart we display on screen is astonishing: the total expected financing for OpenAI, Anthropic, and SpaceX’s IPOs will reach $180 billion, surpassing the entire internet bubble’s total of $164 billion. And that total is over three years, just for these three companies.

This scale is incredible. We need to answer key questions: Is this a cycle? Are these companies running out of money? Are they too big for private capital to sustain? There’s much to discuss. Ejaaz, let’s start with SpaceX.

Ejaaz:

SpaceX, OpenAI, and Anthropic are preparing for mega IPOs, but the real story isn’t just individual fundraising; it’s that they might go public in a concentrated window within weeks. Their goal is to complete IPOs by the latest in Q4, with enormous combined funding. This situation is unprecedented in history.

Individually, SpaceX probably filed its S-1 around April 1, signaling its intent to go public. Market rumors suggest it might list this month or by early July at the latest. About ten days ago, rumors indicated OpenAI also secretly filed its S-1 for IPO prep. Just yesterday, Anthropic submitted a confidential S-1. So these three companies are nearly sprinting toward mega IPOs within the same timeframe.

This raises a question: why now? Why so urgent? In my view, the answer is simple: AI capital expenditure is becoming more expensive than these companies initially thought, and they are choosing to keep adding. Their free cash flow is no longer enough to support current demands.

So far, these companies have mainly used private funds—either from investors or their own revenue. Now, they’re turning to the public markets, asking investors: “We need more money to build more data centers, buy more GPUs, train more models to meet growing demand.”

If you ask these companies, they won’t admit they lack demand. In fact, Google, Amazon, Microsoft, and Meta—all reporting profitable quarters despite huge AI capital spending. These four may plan to invest nearly $1 trillion this year, yet it’s still not enough. So they need more capital to sustain this build-out.

What I truly worry about is whether we’ve reached a point of no return. We discussed this before recording: once you get here, there’s no turning back. Whether through debt leverage or other financing, we’ve already jumped into the abyss. If we don’t go all out now, we’ll have nothing.

SpaceX rewrites IPO and index rules, pensions buy passively

Josh:

What surprises me most is that not only are these companies going all out, but institutional investors, retail investors, and even major Wall Street funds are betting on these IPOs, and they’re doing so by changing rules to accommodate them.

A typical recent example is SpaceX’s IPO. To get its stock into indices faster, index providers relaxed the original profitability requirements and shortened the observation period from 90 days to 5 days. Traditionally, a company had to prove ongoing profitability and stable performance for 3 to 6 months before index inclusion. Now, index funds can buy earlier. Even if some investors don’t want to actively invest in SpaceX, their retirement accounts, 401(k)s, and index fund holdings will be passively buying SpaceX stock sooner than before.

Ejaaz:

This has never happened before in history. SpaceX, through its IPO, is effectively redefining market rules to meet its huge funding needs.

Several data points are especially notable. First, over $30 trillion in passive 401(k) funds—retirement funds—will be forced to buy SpaceX stock, and at IPO valuation levels. Rough estimates suggest about 24% of the shares issued in SpaceX’s IPO will be absorbed by these passive funds. Such scale of passive buying is unprecedented.

Second, under traditional rules, to be included in major indices like NASDAQ 100 or Fortune 500, a company usually needs to demonstrate large, sustained revenue and stable quarterly performance over 3 to 6 months. Now, rules have been significantly relaxed. Some indices only require a 5 to 15-day observation period; as long as the stock price remains above a certain level for a few days post-IPO, the company can be included. In other words, Elon Musk only needs to ensure the stock price stays at a certain level in the short term to qualify easily.

Josh:

This change is indeed concerning. For decades, major indices followed two core rules to protect ordinary investors, especially those participating through retirement or regular investment accounts. The first is that companies must be profitable for four consecutive quarters; the second is a minimum free float of 5% to 10%.

These rules were established after the 1999–2000 dot-com bubble burst, when many indices included high-growth but unprofitable companies, leading 401(k) holders and retail investors to passively hold these stocks, suffering huge losses when the bubble burst. These rules were introduced to protect ordinary investors.

Now, it seems history is repeating. Rules are being reversed. Companies no longer need four quarters of GAAP profitability—just a good 15-day performance window. This threshold has been significantly lowered. It’s concerning. While this relaxed rule might support SpaceX’s stock price with strong passive demand, it could harm ordinary investors holding SpaceX shares if things don’t go as planned.

Ejaaz:

Josh, I think many criticisms of SpaceX’s IPO are valid: it hasn’t truly proven its revenue model. It claims “we will send AI data centers into space… via rockets,” but this business remains unverified, not even at the proof-of-concept stage. Yes, it’s launched rockets, but we haven’t seen GPUs in space actually training cutting-edge AI models. So, in a sense, it’s more like a “trust me” promise.

However, one company is different. It has proven its revenue-generating ability and is growing astonishingly fast: Anthropic. Just yesterday, Anthropic filed an S-1 registration statement with the SEC, indicating plans for an IPO in the coming months.

Anthropic’s revenue leap and IPO motivation

Ejaaz:

Anthropic’s story differs from SpaceX’s in several ways. Their CFO Krishna Rao previously said the company had no immediate IPO plans, would take it slow. At that time, they had already reached about $9 billion in annualized recurring revenue, expecting to hit around $20 billion in 2026. Yet, they achieved that figure within the first month and a half of 2026. Recently, their ARR hit $45 billion. This is mainly thanks to the success of Claude Code, Claude Co-Work, and a series of enterprise contracts. They’re also involved in many joint ventures, financing from Blackstone, and pushing various initiatives.

Compared to SpaceX, Anthropic is already making a lot of money. So, it’s reasonable for them to want to go further. They’re aggressively acquiring compute capacity, mainly competing with OpenAI: training frontier models, serving those models, and providing access to as many users as possible. That’s why I think they’re pushing for an IPO.

It’s worth noting that no detailed disclosures have been made yet, right? It’s more like a required statement. OpenAI probably doesn’t need to do this, but Anthropic chose to disclose for transparency. Josh, what’s your take?

Josh:

Interesting—this looks like a “confidential disclosure of a confidential submission,” quite ironic. But I think it caught many by surprise. Almost everyone expected OpenAI to IPO earlier than Anthropic. When this news broke, the odds shifted dramatically.

From The Information, the data might already be outdated, because I’ve heard rumors that Anthropic’s growth is even faster than previously expected. It’s showing an incredible growth trajectory, driven by real enterprise value creation and very strong model capabilities.

Think of Mythos. They announced this model just two months ago, meaning training was completed even earlier. These models are extremely powerful. I feel Anthropic is very confident about going public.

This also raises another question: how much money can the market absorb from these financings?

We know SpaceX will likely go first. Rumors suggest around June 12, within the next two weeks. That could absorb $10k in capital. Although their target is $75 billion, I believe it will go higher. How much can Anthropic raise from the market? If OpenAI goes public afterward, how much funding remains for OpenAI? We’re asking for huge amounts of capital from the public markets. Private markets may already be drained or not—either way, we’ll soon see how quickly public market reserves are consumed, as each company’s funding needs are staggering.

Ejaaz:

My view is: there will be a class of investors buying these stocks because their core logic is simple—bullish on AI. Then, a large number of retail investors will say, “I use Claude every day, it’s very helpful, so I’ll buy this company’s stock,” or they use ChatGPT daily and think similarly.

Ultimately, these two groups will lead to the same result: buying these stocks. From the companies’ perspective, their IPO funding purpose is clear. As Krishna Rao, Anthropic’s CFO, said, and Sarah Friar of OpenAI, and Elon Musk also mentioned: we need more compute. More compute means better AI, better AI means better products, better products serve more customers, and ultimately generate more revenue.

Specifically for Anthropic, rumors about their AGI model Mythos are very real. There’s also breaking news about Project Glasswing, a staged, sandboxed release of Claude Mythos, covering 150 new organizations globally. They recently mentioned in a statement that they will release to the public in the coming weeks. All this happening now seems very coincidental, or perhaps deliberately arranged.

Another difference for Anthropic is that they expect to make about $550 million in profit by the end of this month. Compared to trillions in capital expenditure, it’s just a drop in the ocean, but it will be the first large AI lab to reach that milestone. Their growth rate is truly astonishing. Among all these IPOs, I might be most optimistic about Anthropic, but each company has its own path.

Google’s $80 billion financing: not an IPO of an IPO

Josh:

Previously, we speculated whether big tech firms like Google would start spending beyond their income capacity—that is, financing through debt.

Now, we see signs that the market is gradually moving into that territory. Google’s balance sheet income no longer covers their needs, so they’re seeking external capital. This isn’t an IPO, since Google is already listed, but they still need more money. So what did they do? They raised $401k to support AI development—an enormous sum.

I don’t recall the exact total capital expenditure they committed to, but I estimate this amount is close to 30–40% of their annual planned capital spending. Notably, Berkshire Hathaway, Warren Buffett’s firm, committed $10 billion. It’s a big deal: $401k from underwriters, $40 billion from a market offering starting Q3 this year, and $500B from Berkshire’s private placement.

We previously analyzed Google’s balance sheet, seeing how much they earn and spend. At that time, they were positive. Now, it looks like they’re heading toward losses—or are just preemptively buffering to keep a sufficient safety margin?

Ejaaz:

I think they’re all-in, and their books might turn red eventually. Larry and Sergey Brin explicitly said about a year and a half ago that they prefer risking everything in the AI race rather than losing it. They will keep spending until they find enough breakthroughs. That’s the founder’s mode—they’ve returned to it. Sergey Brin is back at Google to steer the company back into this state.

This is my favorite IPO story of the week, but it’s not even an IPO. Google, as a listed company, raised $80 billion. The question is: what is this $40k really for? The headline is obviously: to build more AI infrastructure, more TPUs, more compute, etc.

But many overlook a story: about $30 billion of that might be used to pay taxes from employee stock vesting over the next few months. That is, a large part of this financing isn’t really for AI expansion.

Setting that aside, I don’t think Google is a villain. They’ve been as transparent as possible about how much they’ve spent on AI and what they plan to do. They are indeed trying hard. But it reminds me of our discussion late last year about OpenAI’s state: we said OpenAI was somewhat distracted, doing many random AI products, missing the core focus on coding AI, then declared Code Red and refocused.

I feel Google is now drifting into a similar over-broad state. They lack focus. They’re working on agents, trying to improve coding models; building better general models, infrastructure like TPU; even selling TPUs to competitors, yet their own training capacity for Gemini is insufficient… and Gemini 3.5 Flash, despite all this spending, still lags behind the frontier models. Now they need to raise even more to train better models. It seems they need to truly lock in targets and focus.

From a financing structure perspective, $80 billion is enormous—almost like a secondary IPO for their own support. I’m not entirely convinced that using $30 billion to cover taxes is the best approach. It feels somewhat desperate. But I remain optimistic because in past large corporate financings, companies like Berkshire Hathaway that committed $10 billion often performed well afterward. I hope Google will be the same. But this story is definitely very interesting.

Josh:

Long live Berkshire, right? They’ve been very disciplined and accurate in their judgments. We hope this time they can continue that. Also, note that Google holds large stakes in many upcoming IPOs. It’s a major private shareholder in SpaceX and Anthropic, so as their stock prices rise, Google benefits significantly.

But these numbers are getting bigger and a bit frightening. We seem to have become numb to hundreds of billions of dollars. Google’s $180–$190 billion capital expenditure this year, a few years ago, would have been unimaginable. So when they say they’re going all-in, it’s at a scale we’ve never seen before.

I think this is also one of the themes of this episode: we are in a very special moment in history. Never before in American or capitalist history has so much capital and value been concentrated on the same idea… The US is re-industrializing in a significant way.