Woofun AI reports that a fundamental paradigm shift is reshaping global equity markets, where a $1 billion fund has successfully driven trillions in market capitalization for companies operating at severe losses. This phenomenon, characterized by low float, grand narratives, and inflated valuations, marks a decisive break from the financial orthodoxy that governed markets for two decades following the dot-com crash. The era of discounting cash flows and adhering to price-to-earnings ratios has been supplanted by a new logic where scarcity of tradable shares and the power of storytelling dictate asset prices, creating a speculative environment reminiscent of the early internet boom but with distinct structural differences.

The historical context of this shift is rooted in the lessons of the year 2000, when Pets.com spent $1.2 million on a Super Bowl commercial featuring a sock-puppet dog despite generating less than $6 million in annual revenue and posting losses exceeding $60 million. Nine months after that ad aired, the company liquidated, cementing the lesson that valuation without revenue support is a bubble. For the subsequent twenty years, investment textbooks were written around this failure, deifying Warren Buffett and establishing DCF, PE, PEG, and Free Cash Flow Discounting as the undisputed mainstream valuation methods.

However, the technology race unfolding between 2025 and 2026 reveals a stark contradiction: the most hyped companies in the AI sector are losing money at an accelerating pace, yet their stock prices are soaring to unprecedented heights.

CoreWeave, an NVIDIA-backed AI computing infrastructure firm, exemplifies this divergence. The company reported revenue of $16 million in 2022, which exploded to $5.1 billion in 2025, representing a growth rate of over 300 times in just three years. Despite this astonishing revenue expansion, the net loss widened from $31 million to $1.2 billion. In the first quarter of 2026 alone, CoreWeave generated $2.1 billion in revenue while incurring a net loss of $740 million, resulting in a debt-to-equity ratio of 10.7. By traditional banking credit standards, such a balance sheet would be deemed insolvent or highly distressed. Yet, following its public listing, the stock price surged by 190% at one point, defying all conventional risk assessment frameworks. A similar trajectory is observed with Nebius, a company that pivoted from being part of Russia's Yandex to providing AI cloud services. In the first quarter of 2026, Nebius posted revenue of $399 million, a 684% year-on-year increase, yet it still reported an adjusted net loss of $100 million. Over the past 12 months, its stock price has risen by over 510%, demonstrating that the market is pricing in future dominance rather than current profitability.

Shifting focus to the Chinese market, the pattern of high valuation amidst deep losses is equally pronounced. Zhipu reported full-year revenue of 724 million RMB in 2025, approximately $100 million, but suffered a net loss of 3.182 billion RMB, which is 4.4 times its total revenue. This implies that for every $1 earned, the company spends far more than $1 on computing power and R&D. Another AI stock listed in the same batch, MiniMax, saw its Hong Kong shares surge by 109% on the first day, with subsequent increases exceeding 700%. With full-year revenue of $79.038 million, roughly 560 million RMB, MiniMax's revenue is even lower than Zhipu's, yet its market reception was explosive. Similarly, the Hong Kong GPU company Biren Technology, the A-share domestic GPU firm Mu Xi Technology, and the Sci-Tech Innovation Board's Moore Thread saw first-day increases of 120%, 693%, and 425%, respectively. These astonishingly surging new stocks are all in a state of severe losses or lack of profitability, rendering traditional PE ratio calculations impossible as the denominator is negative.

When analysts attempt to apply Price-to-Sales (PS) ratios, the figures become even more surreal. Zhipu trades at a PS ratio of over 1,200 times, while SpaceX is valued around 95 times. Using Discounted Cash Flow (DCF) analysis, a slight change in the discount rate or terminal growth rate could cause the valuation conclusion to plummet from $100 billion to $10 billion, rendering the model highly sensitive to the point of losing its guiding significance. Aswath Damodaran, the author of the definitive DCF textbook, valued SpaceX at $1.2 trillion, which was 30% lower than its IPO price, acknowledging that even a slight adjustment in parameters when dealing with this generation of IPOs can lead to dramatic fluctuations in results. Some observers argue that this mirrors the early internet era when PE ratios were ignored and Amazon took twenty years to turn a profit. While true, there is a critical distinction: investors in the internet era looked at user growth, GMV, and page views, using quantifiable intermediate metrics to anchor valuations. Today's AI companies have metrics like ARR, but even ARR cannot explain a P/S ratio of 1200x. The hype has detached from financial fundamentals, pricing in all expectations for the next three to five years into the present.

The core assumption of traditional pricing methods—that future cash flows can be extrapolated from historical financial data—is failing against this new asset class. Model weights, algorithm capabilities, developer ecosystems, and computing power scheduling are the true core assets of AI companies, yet none of these can be written into a balance sheet. The programming capabilities of GLM-5.2 led Vercel's CEO to say 'almost shocked,' a sentiment that will not be reflected in the income statement of Meta. CoreWeave sits on a backlog of $100 billion in orders, but this does not alter the fact of its net loss for the quarter. Nvidia's GPUs are called the oil of the AI era, and the pricing of oil has never been based solely on quarterly production but also on reserves, demand curves, and geopolitics. The revenue curve of AI companies is not linear; it depends on leaps in model capabilities, network effects, and sudden policy shifts. After the release of GLM-5.2, Meta's narrative status could change overnight; Llama going open source rapidly amplifies Meta's AI influence; U.S. chip restrictions on China transform Wallen and Muxi from marginal companies into 'domestic alternatives.' These variables are difficult to incorporate into any financial model in advance.

The market's tolerance for narrative dominance is rising because, over the past few years, those who believed in narratives have indeed made money. Those who ignored financial statements and bought Nvidia at the beginning of 2023 saw their investment increase tenfold. Those who ignored financials and bought Meta at the beginning of 2026 saw a 24x return. When an 'incorrect' approach consistently delivers 'correct' results, the market adjusts its methodology rather than correcting the outcomes. A Nasdaq study looking back at data from 1980 to 2020 revealed that in the 1980s, the average free float of US stock IPOs was about 30% of total shares outstanding. By 2020, this number had dropped to around 20%. J.P. Morgan provided a more macro figure in a report from June 2026, stating that newly issued shares in IPOs, combined with early investor shares allowed to be sold after lock-up expiry, only represent approximately 1% of the total market capitalization. This shrinking free float has been an almost thirty-year trend.

Nasdaq also found a clear inverse relationship between free float and first-day returns. In years with a smaller free float, the first-day returns were larger. In our analysis of US stock IPOs from 2024 to 2026, we observed the same pattern. Defining low float as 'current float/total shares outstanding less than 30%,' in samples where first-week performance could be calculated, low float IPOs saw a first-day increase of 67.4%, a third-day increase of 65.2%, and a fifth-day increase of 63.6%. The corresponding percentages for non-low float IPOs were only 47.9%, 48.9%, and 49.6%. With fewer chips available for purchase, the same buying pressure has a greater impact, leading to stronger price elasticity. The logic is simple: the same $1 billion buy order is a ripple in a $200 billion free float but a tsunami in a $30 billion free float. As the free float shrinks from 20% to 3%, it is not a linear change but a qualitative shift in price elasticity.

Newly listed companies are increasingly inclined towards low float because this maximizes various interests. First, founders benefit from secure control. Elon Musk of SpaceX holds about 85% of the voting rights through Class B shares, and the 4.3% free float to the public market means external investors have almost no governance influence. He can simultaneously serve as CEO, CTO, and Chairman, merge xAI into SpaceX without shareholder approval, and have complete control over the company's strategic direction. The smaller the free float, the weaker the voice of external shareholders, and the greater the freedom of the founder. Scarcity also directly drives up market capitalization figures. A company's market cap is not determined by all outstanding shares but by the price of the last trade multiplied by the total shares outstanding. If only 3% of the chips are in play and these chips are bid up to an exorbitant price, the entire company's market cap will be calculated based on this price. The 97% of shares held by founders and early shareholders that are not actively traded follow suit in paper value expansion. This inflated market cap can be used for fundraising, M&A currency, and talent attraction. When SpaceX went public with a $1.77 trillion valuation, this number would show up in all job postings and be on the table for all partnership negotiations.

This phenomenon is not limited to small-cap stocks. Figma (FIG), a collaborative design software platform, has only 2.36% of its outstanding shares in circulation, rising by 250% on the first day, 168.48% on the third day, and 173.7% in the first week. Circle (CRCL), the stablecoin and blockchain financial infrastructure company behind USDC, has 13.68% of its outstanding shares in circulation, seeing an increase of 168.48% on the first day, 271.77% on the third day, and 278.06% in the first week. Bullish (BLSH), a digital asset trading platform, has 19.78% of its outstanding shares in circulation, experiencing an 83.78% increase on the first day, 87.95% on the third day, and 60.84% in the first week. Cerebras (CBRS), an AI computing power infrastructure company, has 13.66% of its outstanding shares in circulation, rising by 68.15% on the first day, 60.35% on the third day, and 57.13% in the first week. Investment banks also benefit, as the IPO 'first-day return' is a key metric to measure underwriting success. Goldman Sachs designed a 4.3% float for SpaceX, resulting in a 19% first-day increase, hailed as a great IPO. If the float were 20%, with the same scale of buy orders distributed among five times more chips, the increase might only be 4%, leading to completely different media headlines. The incentive structure of investment banks inherently favors low floats.

Cornerstone investors in the Hong Kong Stock Exchange system also benefit from this dynamic. The system is a trade-off: 'I'll lock up the chips for you, and you guarantee me an allocation.' Cornerstone investors receive a fixed IPO share in exchange for a 6-month lock-up period.

However, this cost often turns into a reward because the cornerstone locks up most of the float, making the remaining tradable chips scarce and easy to drive up. Six months later, when the lock-up expires, if the stock price has increased several times, the cornerstone investors see returns far exceeding a typical IPO. Smart Map's 11 cornerstone investors, such as Hillhouse Capital, Taikang Life Insurance, and GF Fund, took away 70% of the already limited number of outstanding shares, resulting in less than 4% of shares available for trading. These shares are all subject to a 6-month lock-up, creating a scarcity premium for themselves.

The systematic turning point is evident in NASDAQ abolishing the 10% minimum public float requirement. This rule had been in place for decades to ensure liquidity and protect public investors. The S&P 500 requires a certain level of public float, MSCI requires 15%, and the Russell series requires 5%. If NASDAQ can waive the 10% threshold for SpaceX, what obstacles would the next company face wanting to IPO with only a 3% float? If the largest U.S. exchange believes a low float is acceptable, will other exchanges follow suit? The cornerstone system of the Hong Kong Stock Exchange already allows for the majority of IPO shares to be locked up. If NASDAQ also relaxes its rules, could we see a global competition where exchanges compete to be the most favorable to low floats to attract the best IPO candidates?

In the 1990s, as the options market matured, the zero-cost collar became a standard practice for the wealthy. Holding a stock, one would buy a put option to protect the downside while simultaneously selling a call option to earn back the cost. These actions offset each other, allowing one to lock in a price range without spending money. In the late 1990s, Michael Dell used variable prepaid forwards to cash out a portion of his Dell shares without triggering taxes. Previously, this strategy was used by a few ultra-rich individuals, but now, after SpaceX's IPO, wealth management firms openly offer this strategy to thousands of employees. Wealth managers like Bernstein and Mercer are now providing guides to SpaceX employees on how to use collars, a level of widespread adoption not seen before. The Bernstein report contains sobering data: looking back at all U.S. stock IPOs over the past decade that raised over $50 million, six months after the lock-up period expiration, the median return was a decline of about 10%. One in ten IPOs dropped at least 62% in the six months following the lock-up. The conclusion is straightforward: if you are a SpaceX employee holding locked-up shares, statistically speaking, by the time you can sell, the price is likely to be lower than it is now. Therefore, you should use derivatives to lock in gains before the lock-up period ends.

Michael Burry, who famously shorted the U.S. subprime mortgage market in 2008, openly stated after SpaceX went public that he had looked into buying put options to short the stock but found the prices prohibitively high. In the end, he neither went long nor short. Even the 'Big Short' found the cost of shorting too high, indicating that too many people were attempting the same trade, driving option prices to absurd levels.

In addition to these strategies, whether it's collars, pair trading, or lock-up arbitrage, they all share a common prerequisite: the company is already public.

However, if you have spent a few years in the crypto market, you will find that these changes are not at all surprising. In 2024, Binance Research, from the largest exchange in the crypto market, released a report titled 'Low Float & High FDV: How Did We Get Here?' The report listed a group of tokens that had just been launched, with the lowest circulating supply being only 6% of the total supply and the highest not exceeding 20%.

In 2024, the newly issued tokens had the lowest market cap to fully diluted valuation ratio in the past three years. These projects raised their valuations to the sky by releasing only a small amount of chips at launch. These projects at launch initially appeared to have caught up with the market caps of Layer-1 and DeFi blue chips that had been in operation for several years. Then, once the lock-up period ended, the price plummeted. This is the lesson that the crypto market taught everyone in two years: a low float can drive up the price, but it is not sustainable. Because eventually, the lock-up will be released, and unlocking is the supply. If there is no corresponding demand to absorb it, the price will drop. Binance Research did the math: between 2024 and 2030, an estimated $155 billion worth of tokens are set to unlock gradually. To maintain the current prices of these tokens, the market would need an additional $80 billion in buyer liquidity. According to data compiled by Memento Research, there were 118 major Token Generation Events (TGEs) in 2025, out of which 100, or 84.7%, are currently trading below their fully diluted valuation at listing. The median drop is 71%.

In its 2025 year-end summary, CryptoRank put it bluntly: 'The token issuance model of low circulating supply and high FDV continues to hinder the season of altcoins, as most of the upside potential has been taken by private sale and early investors, leaving very limited opportunities for the open market.' The low circulating supply tokens in the crypto market and low float IPOs in traditional finance have almost identical price formation mechanisms in the early stages of listing: scarce chips, abundant narratives, overwhelming buying pressure far exceeding the tradable supply, pushing prices far above the fundamentals. So, will the price drops of crypto altcoins be replicated in the stock market? Not necessarily. After all, the stock market has index passive funds providing continuous buying pressure, deeper institutional involvement, and a more diverse set of funding sources. These are structural supports that the crypto market lacks. This allows low float IPOs in the stock market to rise for months. Of course, the stock market still cannot avoid the inevitable lock-up expiration day, where the true test of value versus narrative will finally occur.