Your 401K Is Their Exit Strategy (SpaceX, Anthropic, OpenAI)
In a Nutshell
The video argues that SpaceX, OpenAI, and Anthropic are timing massive IPOs to dump overvalued shares into 401(k)s and index funds, enabled by last-minute NASDAQ rule changes that force passive funds to buy these companies at inflated prices despite tiny public floats. These valuations are propped up by circular AI spending—where big tech invests in AI startups that then pay the same tech companies for compute—creating fake earnings that mask negative returns on trillions in AI infrastructure spending. The core warning is that passive investors will absorb the losses when this earnings bubble bursts, just as history shows initial investors lose while later buyers profit from the actual technology at distressed prices.
These notes were generated by AI and may contain inaccuracies.
We're living in a time that might be remembered as one of the biggest bubbles in history. In the next few weeks, retirement accounts and 401ks will be buying shares in some of the biggest IPOs in human history, even if investors do not want to. The rules of the financial system were rewritten to make this automatic. Larry Frink from BlackRock stated that retirement funds and pension funds will be used to build out AI infrastructure, with much of the money coming from the private sector, savings accounts, pension accounts, and insurance companies.
An IPO (initial public offering) is when a private company goes public, allowing investors worldwide to buy shares. SpaceX is one company going public soon with a valuation of $1.75 trillion. This would make SpaceX more valuable than every American defense contractor combined on day one and would represent the biggest IPO in human history, surpassing Saudi Aramco's record from 2019. Saudi Aramco was the most profitable company on the planet when it listed, but SpaceX lost $5 billion last year.
The financial rules meant to protect investors from buying overpriced investments were changed right before these IPOs. On May 1st, NASDAQ introduced the fast entry rule, which cuts the waiting period for a company to be included in an index from 3 months to just 15 trading days. It also removes the float requirement that would have disqualified SpaceX. The rule change forces index funds to artificially inflate how much of the company they must buy.
A handful of insiders got in early at low prices and now need buyers to absorb trillions of dollars worth of stock at peak values so they can exit. Finding enough buyers for the biggest IPOs in history is difficult without changing the rules, meaning 401ks might become the exit liquidity needed. SpaceX is the first company in line, but OpenAI and Anthropic are right behind doing their own IPOs. Three companies with a combined valuation of about $4 trillion would leapfrog every other company in America on day one.
The NASDAQ 100 is the gold standard index for tech companies tracking the top 100 tech stocks. There is over $600 billion worth of investment products that track this index. When a company gets added to the NASDAQ 100, every fund tracking the index is forced to automatically buy that stock with no choice or vote. This automatic buying creates billions in demand. Getting into the NASDAQ 100 is like being given a money printer because the company gains access to hundreds of billions in passive investment flows that must buy their stock.
The NASDAQ fast entry rule changed three things. First, the waiting period was reduced from up to a year until the next yearly review to just 15 trading days. Second, the float requirement was eliminated. A float is the percentage of company shares available for the public to buy and sell. Under old rules, NASDAQ required a minimum 10% public float. SpaceX plans to list with a float of around 4 to 5%, which would have been an automatic disqualification. Third, a hidden multiplier was introduced: for companies with a float less than 20%, NASDAQ now artificially inflates the stock's weight in the index. A 4% float gets treated as 12%, and a 5% float gets treated as 15%, creating a 3x multiplier. Index funds are now legally required to buy more stock than the actual available supply would normally justify.
NASDAQ wrote these rules for a whole class of companies, with SpaceX as the first in line, followed by OpenAI and Anthropic planning their own public listings. Three companies going public could funnel as much as $4 trillion worth of newly issued stock into passive retirement accounts within months of each other. Other index providers are making similar changes: FTSE Russell allows index inclusion after just 5 days, and S&P is doing the same. Every major index provider is racing to update their rules simultaneously. The timing is suspicious because all rule changes were announced weeks and months before the biggest IPOs in human history, with all companies listing within months of each other.
These companies need to sell trillions of dollars worth of stock, and the only buyer big enough to absorb that supply is the passive investment complex with trillions sitting in 401ks and index funds. Passive investors do not choose what they buy; they buy whatever makes up the index. To get guaranteed demand for an IPO, companies try to get into the index rather than pitching investors directly. If the rules say they do not qualify, the rules get changed.
SpaceX has one of the best dreams: making humanity a space-exploring civilization. However, SpaceX consists of three different businesses. The rocket business includes government contracts, NASA missions, and reusable rockets, generating around $4 billion in revenue last year. Starlink, the satellite internet provider, has 10 million subscribers across 150 countries and made $11.4 billion in revenue with 63% profit margins. The third business is XAI, Elon's artificial intelligence company, which burns over a billion dollars a month. SpaceX overall lost $5 billion last year. At a $1.75 trillion valuation, investors are buying all three businesses bundled together.
The debt-based AI boom with trillions in valuations and record profits is partly an illusion built on an accounting trick where money passes from one company to the next and back, artificially increasing apparent profits. This strategy works only as long as markets stay strong. There is also a geopolitical context with the Middle East situation, and the longer they wait, the more likely financial markets could break due to supply chain realities. Companies need to move fast, pump up values as high as possible, and get as many people to buy.
An example of how the system works: if someone lends $100 and the borrower uses it to buy a Pokémon card that goes up $1,000 in value, the lender can write down $900 in paper profit on their tax return even though nothing was sold. That paper profit then justifies lending another $5,000 to buy more cards. The money is mostly going in a circle.
Financial Times analysts examined capital expenditure by Microsoft, Google, Amazon, Meta, and Oracle for building AI infrastructure through 2030 and compared it to expected revenue from those investments. Under the most generous assumptions with zero costs for salaries, electricity, or overhead, Microsoft's implied return on AI investment was 9.2%, Google's was negative 15.7%, Meta's was negative 28.8%, Oracle's was negative 35.6%, and Amazon's was 7.2%. These are best-case numbers, and all companies still lose money.
These same companies issued over $150 billion in corporate bonds this year alone to fund AI spending, more than double what they were doing two years ago. Microsoft, Meta, and Google have seen their free cash flow margins collapse toward zero when accounting for AI capital expenditure. JP Morgan analysts project that by 2027, several of these companies will have negative free cash flow, spending more than they bring in, despite being the most profitable businesses in human history just a few years ago.
OpenAI committed $280 billion to Microsoft and $138 billion to Amazon. Anthropic committed $30 billion to Microsoft and $100 billion to Amazon. OpenAI and Anthropic spending commitments represent roughly half of Microsoft's entire revenue backlog, 54% of Oracle's, and 51% of Amazon's. Big tech invests in AI startups, the startups use that money to rent computing power back from big tech, big tech books the investment gains as profit, and then uses that profit to justify spending even more.
Michael Bur's company analyzed this circular system. The circle only keeps spinning as long as everyone agrees about the valuations. The moment these companies go public, valuations stop being what a couple investors decided in private funding rounds and become what the open market decides. If those numbers come in lower than what Google and Amazon have been booking as profit, those profits must be revised downward, and the whole loop goes in reverse. This is why the timing is critical: companies need IPOs to validate the paper before it runs out or gets exposed.
History shows a pattern where the biggest, most hyped IPOs occur at market peaks. Companies like Xerox, Ford, McDonald's, Apple, Goldman Sachs, and Blackstone went public when investor excitement was highest, and in almost every case the market peaked right after. The IPO is rarely about the company needing money; it is almost always about the seller needing a buyer. The best time to find a buyer is when everyone wants what is being sold. By the time SpaceX, OpenAI, and Anthropic complete their IPOs, they will have raised as much money as all the 300.com IPOs from 2000 combined, adjusted for inflation.
Many say the AI boom differs from the dot-com bubble because PE ratios are not insane and valuations look reasonable. However, BCA Research suggests the AI bubble is not a valuation bubble but an earnings bubble. In a normal bubble, stock prices rise while earnings stay flat, causing PE ratios to spike. In an earnings bubble, the earnings themselves are inflated, keeping PE ratios low and reasonable. The paper gains from Anthropic investments inflate Google and Amazon earnings, making their PE ratios look healthier than they actually are.
History shows earnings bubbles with homebuilders before 2008, where PE ratios looked reasonable until they did not, and with banks before the financial crisis, where earnings appeared healthy until assets were shown to be worth a fraction of book value. The theory suggests this is happening now with semiconductors. Global semiconductor sales have gone completely parabolic, a straight vertical line up. Every historical instance of semiconductor sales looking like this was followed by a brutal earnings collapse. In almost every case, stock prices peaked before earnings did.
Nvidia peaked in December 2001 and went down 83% before earnings caught up. In November 2021, Nvidia peaked and went down 53% before earnings caught up. The same pattern repeats with Micron, Intel, and the S&P 500 tech sector across every semiconductor cycle. The stock always knows before the earnings do.
The S&P 500 has hit four consecutive record highs on negative market breadth, meaning the index keeps going up but more stocks are going down than up. Gains are driven by a handful of the biggest companies while everything else deteriorates. This has never happened before in market history. AI-related stocks make up almost 49% of the S&P 500's market cap, with 41 stocks out of 500 representing about half the entire index. If anything goes wrong with AI, it will hurt retirement accounts because almost 50% of retirement accounts in these indexes is essentially a bet on AI.
The US personal savings rate just hit 2.6%, the lowest in 4 years, meaning the average American has almost nothing left in reserve. Real wages are going down, and the bottom half of the consumer economy is already in distress. The top 10% of earners are the only reason the economy is not in a recession, holding up half of the economy. Corporate earnings are going up, but worker incomes are flat. The gap between those two lines represents AI. Companies are automating their way to higher profits while workers make less and less. At the limit, firms automate their way to boundless productivity and zero demand. Passive investment money that SpaceX, OpenAI, and Anthropic count on only exists as long as American workers have paychecks to invest. The moment AI displaces those paychecks, the money supply shrinks.
The Strait of Hormuz has been closed for 3 months due to the Iran war. Exxon Mobil's senior vice president stated at a New York conference that oil inventories will hit critically low levels within the next 2 to 3 weeks. When that happens, oil prices could reach $150 to $160 a barrel. Countries that import oil will suddenly need many more dollars to pay, and when countries need dollars fast, they sell
The main reason countries sell assets during conflict is because the biggest dollar-denominated assets most countries hold are US Treasury bonds. When they sell US treasuries, interest rates go up because the US needs more buyers. The way to attract more buyers is to offer higher interest rates. When rates go up, borrowing gets more expensive for governments, companies, and consumers.
A chart from FFTT shows that since the beginning of the Iran war, wherever oil goes, Treasury yields follow almost perfectly. If ExxonMobil's prediction of $150 oil materializing in the next few weeks proves correct, Treasury yields are likely to reach levels that start breaking things. Early signs are already appearing.
Emerging market countries sold US treasuries in March at the fastest rate since at least 2023. Twenty-seven countries have approached the World Bank looking for emergency crisis funding. Global bond yields are breaking out of multi-year consolidations across the US, UK, Germany, Japan, and Canada.
The new Federal Reserve chairman Kevin Walsh was hired by President Trump to lower interest rates. However, the market is currently predicting that interest rates will go up. For the first time in four years, the two-year Treasury yield is above the federal funds rate, which is the market's way of signaling that the Fed should be increasing interest rates, not lowering them.
If Kevin Walsh lowers interest rates into an inflation spike when oil is rising, the dollar falls, inflation worsens, and bond yields rise even higher as more countries and investors lose confidence and sell more US bonds. If he increases interest rates instead, the stock market drops because corporate borrowing costs rise. The entire AI spending boom that has been holding up 93% of US GDP growth would face a much higher cost of capital.
Either path leads to higher yields, and higher yields are the one factor that could pop the paper profits, circular spending, and the $150 billion in corporate bonds issued to fund AI spending.
All of these pressures converge right at the three biggest IPOs in history, which is the macro-level reason these rule changes might be happening now—the window may be closing.
Technology itself works. Starlink is genuinely one of the most amazing developments in the last decade, and AI is real and will change the world over the next century. However, there is a significant difference between a technology winning and investors winning.
In the late 1800s, the railroad boom transformed America by connecting the country and expanding commerce at unprecedented scale. Railroads revolutionized agriculture, manufacturing, and supply chains. The technology was real and railroads still exist today. Yet almost every investor who funded the railroad boom got wiped out. Companies went bankrupt and stocks collapsed. A new group of investors later bought those same railroads out of bankruptcy at pennies on the dollar and made fortunes.
The same pattern occurred with the fiber optic cable boom of the 1990s. Hundreds of billions of dollars were spent laying cable across the country and under the ocean. The companies that laid it went bankrupt in the dotcom crash. A decade later, that same fiber optic infrastructure became the backbone of the internet. Technology won, but the initial investors lost. The people who bought those assets afterward built the modern world.
This historical pattern is likely to repeat with AI. The initial investors buying at $1.75 trillion valuations, along with passive funds forced in by rule changes, will absorb potential losses. The technology will continue evolving, and a new group of investors will acquire it at much lower prices to build the next era.
The first action is to understand what index funds actually own, because in the coming weeks they will own SpaceX, followed by OpenAI and Anthropic at valuations that history indicates are extremely unfavorable entry points for passive investors. This does not mean selling everything, but knowing what is owned and why.
The second step is determining where money should be positioned to protect against multiple possible outcomes, because none of this may matter. The IPOs could succeed, Kevin Walsh might navigate the next Fed meeting successfully, Hermeus might reopen next week, and the AI boom might generate returns justifying all spending. History might not repeat.
The job of investing is not to be right about one specific outcome. The job is to allocate money across multiple outcomes based on their probabilities, since no one knows exactly what will happen.
Keep Andrei Jikh in your library
Save the videos and channels worth coming back to, and find them again in one place.
