Is The AI Bubble Here? No. Plus Dan Niles, OpenEvidence And Vercel

Hey, hey, hey.

Tack and Markets are moving fast. I'm here to keep you ahead of it. This is Dear Draosa Live, where Wall Street meets Silicon Valley in real time.

Welcome to Dear DBosa Live, where the hype meets the P&L. It is Thursday, July 17th, 12:00 p.m. here in San Francisco, 3:00 on Wall Street. And the AI trade, it is flashing a fresh round of signals. Bullish, frothy, maybe both. Start with Taiwan semi raising its 2025 revenue outlook this morning. Now this isn't just a chip story. This is a demand story. The kind of demand that says big

tech is not pulling back. It is accelerating. Infrastructure spend is still ripping. And Zuckerberg says he's going to be spending hundreds of billions of dollars to build the highest compute for researcher ratio in the world. You're going to want to remember that because that's going to come up in the AI talent wars. Now this isn't just cautious deployment. This is total war. He wants the meta to outgun everyone in raw AI firepower. Not just OpenAI, not just Google, but anyone who dares play in the foundational model race. Google, meanwhile, inked a $3 billion US hydro

power deal in the largest clean energy agreement of its kind this week. For what else? AI compute. Nvidia, meanwhile, still gaining earnings power with China quietly back in play. Anthropic is reportedly flirting with a $100 billion valuation. And all of this brings us to the question everyone is bound to ask again. Or maybe you're already asking it. Is this the AI bubble? Well, no. It suggests a new AI

baseline, I'd argue, not a bubble, even if the numbers look stretched at first glance. Let me explain. Underneath the headlines, there's real revenue traction. In fact, this chart from Bespoke, it shows that the NASDAQ's post chat GBT rally is actually lighting the pace of the dotcom boom. It tracks the NASDAQ's performance in the first 645 days after Netscape's IPO. Remember that company? It was the darling of the internet era, the early internet era. That was in 1994. And then it compares that to the post chat GBT run. Now the takeaway is that this rally is actually

behind where the.com boom was at the same point even though the fundamentals today are stronger because back then we were trading on page views. Today it's tokens, chips, compute, inference. So is this a repeat of 2000? Well, not yet. But this trade is now built on assumptions that AI demand will keep compounding, that capital will keep flowing, that models will keep scaling and monetizing. If any of that breaks the whole thing, it could stumble. So, let's get into it because even if this isn't 2000, it does not mean the road

ahead is smooth. Dan Niles has been tracking tech cycles for decades through bubbles, crashes, everything in between. He called it risks in the AI trade before and was right. Now, as the NASDAQ hovers near record highs and spending ramps to historic levels, we're asking the question again, Dan Dan Niles, does this AI run still have legs? Well, I mean, Dearra, to your point, if you take a step back and you look at the