So tomorrow, the biggest IPO of 2026.

So tomorrow, the biggest IPO of 2026. Prices on the NASDAQ. And shockingly, it's an AI company. Specifically, it's one that makes chips the size of a dinner plate. 20 times the size of normal chips. Naturally, the IPO is also 20 times oversubscribed. Because everything's bigger in AI. It's like the new Texas, but instead of cowboys, we have biohackers and Elon Musk. Tomorrow, a company called Cerebras goes public on the NASDAQ. They make AI chips the size of a dinner plate. Which is not a metaphor. Their chip is literally the size of a plate. For context, most AI chips are the size of a postage stamp. Cerebrous takes the entire disc that normally gets caught into hundreds of individual chips, and turns the whole thing into one giant mega processor. Just like their chips, their ipos are naturally large. They raise the price range twice in one week. Kinda like if so many people wanna come to your concert that you have to get a new stadium. At the top end, the company would be worth $48 billion. And during the Musk trial last week, Greg Brockman testified that Musk was open to merging Cerebras with open AI. Resulting company would have been an absolute behemoth of AI. It would have broken the timeline. But that didn't happen. Instead, it's going public tomorrow. Every investor on Wall Street is clamoring to get in. It's the hottest club. So to take you back a bit, Cerebrous tried this once before. They filed for an IPO back in 2024. They pulled it after the government opened a national security review of their biggest customer. Happens. D42, a company based in the UAE that accounted for over 80% of their revenue. That's not really a normal customer. That is a contingency and not the kind you want on the eve of a listing. The government wanted to know why an AI chip company was financially dependent on a company in Abu Dhabi. Honestly, fair question. But they survived. The government cleared them and Cerebristed what any company would be intend on doing. After nearly dying from having one massive customer. They went and got two more massive customers. Maybe the biggest ones on earth. Open AI signed a deal worth over $20 billion for 750 megawatts of Cerebras computing power dedicated to running the model that writes code. That good dinner plate compute. Amazon started putting Cerebrus strips in its data centers, selling access through AWS. The company that almost collapsed because it depended on one client now has opening eye and Amazon unlock. So weirdly impressive. Come back to earth from a government inquiry, but I respect it now. Cerebrals by the numbers. Kind of explains all the hype. They saw $510 million in revenue last year, up 76%. They had a net income of $238 million. For context, that's a that's 47% profit margin. For context, most companies going public right now are losing money. Take open AI looking towards A potential IPO and losing $14 billion a year. Core Weave is another AI company. They went public in March. Still aren't profitable. Cerebral is profitable and growing. That's why investors put in orders for 20 times the available shares. In this economy, they're an outlier. Back to what you may have been wondering for a minute. Why does anyone need a chip the size of an inner plate? Is it a novelty? Is it just a party trick? Think of it this way. In keeping with dinner plates. Training a model is like building a restaurant. Running it is like keeping the kitchen open. The building of the restaurant costs a lot, but you only ever pay for it once. Meanwhile, the kitchen runs everyday, forever. Running the model is running the kitchen. Keeping it stocked with chefs that show up to work, don't get stuck in freezers. That's the hard part. So when millions of people ask Claude or Chatty BT questions all day long, that part never stops. And it's becoming the bigger cost centre and bigger source of demand over time. And videos. Chips were built for the training part for the most part. Cerebra says their giant mega chip does the running part faster and cheaper. Open AI is using them specifically for Codex, the coding tool. Because when millions of developers are asking AI to write code in real time, speed is the game, and video knows this. They spent $18 billion on R&D Last year. They bought a company called GROC for $20 billion in December specifically to get better at this exact thing. Nvidia also has a decade of software tools that every developer already knows how to use. So Cerebus is betting that a better chip can beat a bigger head start in video. Tomorrow, the market will tell us whether that bet is worth $48 billion. Of course, all roads lead back to the opening eye Musk trial. During the trial last week, Greg Brockman testified that in the early days, Open AI discussed merging with Cerebras, and Musk was open to it. Brockman said that Cerebras represented the compute we thought we were going to need. Open AI and Cerebras almost became the same company. But instead, Musk left open AI, started X AI, built his own data centre, now selling compute to anthropic Cerebrous state independent opening eyes using their chips anyway. So the chips fell where they did. Everyone ended up where they were going to end up. Everyone also just took the most expensive, most dramatic route imaginable. But that's show business and AI business. Tomorrow, a chip the size of a dinner plate goes public at $48 billion. The most expensive dish at the most expensive restaurant on Wall Street. Because everything is bigger and AI.