"History doesn't repeat… but it rhymes." — Mark Twain
◉ THE PRESENT
OpenAI dropped a confidential S-1 with the SEC on Monday, putting an $852 billion price tag on a company that lost money on nearly every query it answered last quarter. Anthropic filed its own paperwork the week before. SpaceX's coming offering is already oversubscribed, with institutions lining up for $10 billion or more in shares. The biggest private names of the AI age are all heading for the public-market door at once — and they're doing it in the same week the chip trade cracked and the Nasdaq logged its worst day since April of last year.
OpenAI S-1 filed at $852B | $20B revenue vs. $14B projected 2026 loss | SpaceX book oversubscribed $10B+ | Nasdaq −4% Friday
When the best companies in a boom all decide it's time to sell stock to the public at the same moment, it's worth asking who's on the other side of the trade. We've watched this exact line form before. It was a Thursday in December, 1999.
◉ THE ECHO — DECEMBER 9, 1999
The stock opened at $299. Nobody had ever seen anything like it.
Larry Augustin ran a small outfit in Sunnyvale that sold computers loaded with free Linux software. VA Linux Systems. The night before, his bankers had priced the IPO at $30 a share — a number they'd already raised twice from an original range of eleven to thirteen dollars, because the phones would not stop ringing. They thought thirty was aggressive. They had no idea.
When trading opened the next morning on the Nasdaq, the first print came across at $299. Not thirty. Two hundred and ninety-nine. The order imbalance was so large the stock couldn't even open at the bell. By the close it sat at $239.25 — up six hundred and ninety-seven percent in a single day. It was the largest first-day gain in the history of American capital markets, and it still is. The company had sold about five million shares. By that afternoon it was worth more than ten billion dollars on paper. It had almost no revenue and no profit at all.
Nobody on the floor blinked. By the end of 1999 more than four hundred companies had gone public, and the internet ones jumped an average of two hundred and sixty-six percent on their first day. A company called Foundry Networks closed its debut at $156, up five hundred and twenty-five percent from a $25 offer. theglobe.com had held the old record — six hundred and six percent — and it had stood barely a year. Records were falling on a schedule. The Nasdaq rose eighty-six percent that year alone.
The logic, if you could call it that, was simple and intoxicating. These were the companies building the future. The internet was real. Growth would come, profits would come later — much later, but they would come — and anyone who waited for proof would miss the whole thing. So the bankers kept filing, the funds kept buying, and the founders kept ringing the bell. Get public while the window's open. Everybody knew the window doesn't stay open forever.
Swap the Linux box for a language model and the free software for a free chatbot, and you have June of 2026.
◉ THE RHYME — WHAT'S IDENTICAL
In both eras the biggest names in the boom rushed to sell stock to strangers at the exact moment the boom looked most certain. That isn't a coincidence. That's a top doing what tops do.
◉ THE DIVERGENCE — WHAT'S DIFFERENT THIS TIME
The rhyme is loud, but the differences are where your edge lives. Four of them matter.
Real revenue, real scale. VA Linux had almost nothing behind it. OpenAI books around twenty billion dollars in annualized revenue and serves nine hundred million people a week. The dot-coms sold a dream; OpenAI sells a product millions of people actually open every day. The losses are real, but so is the business. That is a different animal than a company with a website and a logo.
Private for far longer. The 1999 frenzy threw barely-formed startups onto the public market within months of founding and handed the risk to retail investors. OpenAI, Anthropic, and SpaceX stayed private for a decade or more, soaking up hundreds of billions from deep-pocketed funds first. The public gets a later, more mature company — though it also means insiders may have already banked the easy multiples.
The Fed is leaning the other way. In late 1999 Greenspan was uneasy but still feeding the machine, and the real tightening didn't bite until 2000. Today the Fed is already on hold and the market is pricing a roughly seventy percent chance of a hike in December. Money got cheaper into the last IPO wave. It is not getting cheaper into this one.
The cracks showed up first. In 1999 the IPOs and the index peaked together and the warning came afterward. This round, the chip trade already broke — the Nasdaq just had its worst day in over a year — before the marquee deals even reached the floor. The market is filing these IPOs into visible weakness, not into blue sky.
◉ THE RECKONING — WHAT HAPPENS NEXT
Here is how December 1999 ended. The party ran three more months. The Nasdaq climbed straight through the new year, and on Friday, March 10, 2000, it closed at 5,048.62 — a peak it would not see again for fifteen years. Then it turned. Not a crash at first, just a slow leak that became a flood. By the end of 2002 the Nasdaq was down seventy-eight percent from that March high. Trillions in paper wealth simply evaporated.
VA Linux told the whole story in miniature. The stock that opened at $299 was trading near $40 by August of 2000. On December 8, 2000 — one year and a day after its triumphant debut — it closed at $8.49. A company worth ten billion dollars on its first afternoon was a single-digit stock inside of twelve months.
So what did the smart money do? The smart money had already done it. The bankers took their fees, the early funds had their exits, and the founders rang the bell. The people who got hurt were the ones who bought the story at the top — who looked at a 697 percent first day and decided the only mistake was not buying more. The signal was never in any single company. It was in the flood itself. When every marquee name decides at once that now is the time to sell stock to the public, the market is telling you something plain about the supply of that stock, and it isn't bullish.
That maps cleanly onto a week when OpenAI, Anthropic, and SpaceX all filed inside of nine days, into a tape that had just cracked.
The lesson of 1999 isn't that the technology was fake — the internet changed everything, exactly as promised. It's that being right about the technology and right about the price are two different things, and the gap between them is widest at the moment everyone is rushing for the public-market door. Watch the calendar of filings, not the size of the valuations. A flood of marquee IPOs into a wobbling tape has happened before. You already know how that one ended.
◉ TOMORROW’S WATCH
Wednesday's CPI print will set the tone, but the real tell is whether the SpaceX book stays oversubscribed if the chip selling resumes. In March 2000 it wasn't a single headline that broke the spell — it was the quiet morning when buyers simply stopped showing up for the next deal.
