| Taiwan Semiconductor Manufacturing Co. briefly ignited a stock rally today after beating expectations in the third quarter and raising its growth projections for the rest of 2025. TSMC’s results looked like a good sign for the artificial intelligence economy. But what if it’s actually another datapoint that we’re in an AI bubble? Christopher Beam writes today about one sign to watch for. Plus: If you do think we’re on the brink of something big, where should you put your money? If this email was forwarded to you, click here to sign up. Are we in an AI bubble? A lot of people seem to think so. OpenAI’s Sam Altman and Amazon.com CEO Jeff Bezos have said as much. The tech giants are on track to spend an estimated $371 billion on data centers in 2025, and trillions of dollars more by 2030. As I wrote last week, a growing group of analysts and investors say that the numbers simply don’t add up—that tech companies will never be able to generate the revenue necessary to recoup their spending. With all this doomsaying, you’d expect to see some investors eagerly and loudly betting against AI. Yet only a handful of short sellers have published reports on AI-related companies in recent months. Why the timidity? If this is a bubble, where are all the shorts? I called up Edwin Dorsey, who writes a newsletter focused on short selling called The Bear Cave. While in college, Dorsey interned at the short-only hedge fund Sophos Capital Management before starting the newsletter as a way to share his own research. He says he focuses on $1 billion-to-$10 billion public companies in the tech and consumer space, and he doesn’t take positions on the companies he writes about. The following is a transcript of our conversation, edited and condensed for clarity: Christopher Beam: Are you seeing a lot of short bets on AI companies? Edwin Dorsey: Maybe less than you’d expect. Fuzzy Panda Research published a report on Rezolve AI recently, and Kerrisdale Capital published one on CoreWeave in September. One of the reasons you haven’t seen as many short reports on AI-focused businesses is because there’s so much froth everywhere. There’s crypto companies that are frothy, there’s quantum computing companies that are frothy, there’s flying car companies that are frothy. And when there’s so much froth everywhere, AI tends to have, in my view, a little more substance behind it. The low-hanging fruit isn’t the AI companies, even if you are kind of bearish. CB: Is part of the reason that it’s just hard to make money as a short seller when the market is booming? ED: There's a great saying in short selling that applies here. Short selling is about timing, and when do you want to release your short reports? You want to do it “when the jaguar is out of the trees.” So when things are going up and up and up, you don’t want to short. But the moment there’s that first crack, then you’re going to see a lot of shorting activity and more activist short reports. So I think everybody’s sitting on their hands in wait-and-see mode, because the market could go up another 50%, and these things can continue to skyrocket. CB: How have the recent shorts focused on AI companies worked out? ED: I’m looking at the short against Rezolve AI, a small AI-powered shopping company—that one had a little bit of an impact. Kerrisdale’s bet against the cloud computing company CoreWeave hasn’t worked out so far. There was a report by J Capital Research accusing SuperX of “pretending” to pivot to AI. That stock is up 20% in the last month and 500% in the last six months. CB: If someone wanted to short the AI bubble, what would they sell? ED: My personal view is that the best shorts are companies that are going to be harmed by AI even if AI doesn’t progress a lot. An example that I’ve written up in the past is Chegg, the online education company. That had fallen a lot, but it continued to fall, and now it’s like $1 a share after students stopped paying for Chegg and started using ChatGPT. I think there are a ton of businesses like that: call centers and call center software companies and business process outsourcers. Even if you’re not hugely bullish on AI, I think these types of companies are all going to suffer in the future.  Students have turned to ChatGPT for homework help. Photographer: Leon Neal/Getty Images CB: So to use your metaphor, if this is a bubble, how do you know when the jaguar is out of the trees? ED: It’s when the market narrative has broken and a stock is starting to go down. I think novice short sellers or outsiders would say you want to short at the very peak. More seasoned short sellers have the view that if it’s at all-time highs, you just want to let it keep going. When it’s at an all-time high and then down 20% and the market starts to care about reality, then you want to short. CB: Do you think there’s an AI bubble? ED: I am bullish on AI. I use ChatGPT a ton. I think this stuff is groundbreaking. And I think 10 years from now, it’ll still all be very relevant. So I don’t share the view that AI is a bubble. But I think for short sellers that do, they’re going to be looking for these stocks to really break. Because right now you can say a company has no substance, no employees, no revenue and no path to profitability, and it’ll go up 20% the next day. And that’s a tough environment to write short reports in. Related: The Real AI Risk is ‘Meh’ Technology That Takes Jobs and Annoys Us All | 
