One year ago today, on January 27, 2025, a Chinese artificial intelligence startup called DeepSeek released an AI model that performed as well as anything OpenAI or Google had built, except they did it for $5.6 million instead of, oh, several hundred million dollars. The stock market responded the way the stock market always responds to news that threatens the narrative: it had a complete and total nervous breakdown. Nvidia dropped 18% in a single day, losing $589 billion in market value. That's the largest single-day market cap loss in United States stock market history. Congratulations to everyone who was holding NVDA calls that morning. Your options expired and so did a small part of your soul.
The total damage? Nearly $1 trillion wiped from U.S. tech stocks in one trading session. The Nasdaq fell 3%. The S&P 500 dropped 1.5%. Constellation Energy sank 20% because apparently if AI costs less, we need less electricity, and the market priced that in faster than you can say "margin call." GE Vernova dropped 21%. Siemens Energy fell 20%. An entire ecosystem of companies that existed to feed the AI beast watched their stock prices crater because one lab in China proved the beast eats a lot less than everyone thought.
The $5.6 Million Middle Finger Heard Around the World
Let me explain why this was so devastating in terms even a degenerate gambler can appreciate. Imagine you've been told for years that the only way to win at poker is to buy a $10 million poker table. Everyone believes it. Casinos are built around it. An entire industry springs up selling $10 million poker tables. Investors pour billions into poker table companies. Poker table futures are trading at historic highs. Then some kid walks in with a folding table from Costco and wins the World Series of Poker. That's what DeepSeek did. They walked in with a $6 million folding table and beat the entire field.
DeepSeek built their R1 model using just 10,000 Nvidia GPUs. For context, the other leading AI models were using 16,000 or more. DeepSeek showed you could build competitive AI with 2,000 specialized chips. The entire thesis of the AI boom, that you needed astronomical amounts of computing power to build smart AI, got demolished in a single press release. Every company that had spent billions on data centers and GPU clusters woke up to the possibility that they'd dramatically overspent. It was like finding out your $2,000 monthly gym membership delivers the same results as a $30 pair of running shoes.
The Leveraged ETF Graveyard
Spare a thought for the people who were holding leveraged Nvidia ETFs. The Leverage Shares 3x NVIDIA ETP fell 51.18% in a single day. Fifty-one percent. Gone. In one trading session. If you woke up that morning with $100,000 in 3x NVDA, you went to bed with $48,820. That's not investing. That's a Vegas bad beat. That's putting your life savings on a -700 favorite and watching them lose on a last-second safety. Actually, it's worse than that, because at least in sports betting you know you're gambling. These people thought they were "investing in the future of technology."
The analysts, of course, immediately rushed to reassure everyone. Bernstein's Stacy Rasgon called the reaction "overblown." Anthropic CEO Dario Amodei said the panic was "baffling" and "greatly overstated." Translation: please stop selling, our valuations depend on your continued optimism. This is the financial equivalent of the band playing on the Titanic. "Everything is fine, the ship is barely sinking, now who wants to hear Nearer My God to Thee?"
One Year Later: Did We Learn Anything?
Of course not. We never learn anything. Nvidia recovered. The AI hype train kept chugging. Companies kept spending billions on computing infrastructure. The narrative adjusted from "we need infinite chips" to "we need infinite chips but also efficiency matters now" which is basically the same thing with a slight change in marketing language. The stock market had its panic attack, took a Xanax, and went right back to buying tech stocks like nothing happened.
But for one glorious, terrifying day, the market experienced what it feels like to have your entire thesis invalidated by reality. For one day, the most sophisticated investors on Earth felt what every degenerate gambler feels when the heavy favorite loses by 30: the sick realization that you were wrong about everything, that the numbers you based your decisions on were meaningless, and that no amount of analysis protects you from the universe deciding to rearrange the furniture.
Happy anniversary, DeepSeek. You taught Wall Street what sports bettors have known since the beginning of time: the underdog doesn't care about your model, your projections, or your leveraged positions. Sometimes a $6 million lab in China beats a $10 billion American monopoly, and the only appropriate response is to stare at the ceiling and wonder where it all went wrong.
My portfolio still hasn't recovered. But then again, neither has my confidence in betting the favorites. Some wounds never heal. They just become personality traits.