Dan Niles on the importance of OpenClaw

The one thing you have going on right now that you didn't have during the internet build out, on January 30th you had the formalization of this thing called OpenClaw [a free, open-source, self-hosted autonomous AI personal assistant/agent], which really kicked off the whole agentic [AI] move by corporations.  So Dan Nathan, instead of prior to that saying, "I'm going to ask ChatGPT a question, it's going to give me an answer" might go to ChatGPT and instead of asking, "Well, how did internet stocks perform in the late '90s?" you might go and ask ChatGPT, "Hey, go to Bloomberg, pull down data for all of these 20 internet companies, then go to the SEC website and go look at the 10-Ks and Qs, see if there were any writedowns, and then go and see what the news stories were at The Wall Street Journal, and then create a spreadsheet where I'm putting all of this data."

When you do that, that takes 10 to 100 times more tokens [basic units of text that the LLM reads, processes and generates; 100 tokens ≈ 75 words] and you can see that in the tokens being generated.  And so you have this step function change in compute demand caused by that that started on January 30th.  So I think you're going to see very strong demand for at least another year until you sort of anniversary that, and I think stocks continue to go higher because of that.  

Do I think we're in a bubble?  100%.  Do I think we're going to get a 30-50% drawdown at some point next year?  Yes.  But, do I think you can make a lot of money between now and then?  I think the answer to that is also, yes.

 ~ Dan Niles, "We're 100% in a Bubble and Dan Niles is Still Buying," RiskReversal Media, 6:00 mark, May 29, 2026

 

Dan Niles on agentic AI and the need for compute

With agentic [AI] you're doing multiple actions, multiple calls to APIs, opening up applications like an Excel or a Bloomberg or a CNBC website or SEC website, etc. and doing all of these things.  So you need an orchestration engine.  You need something that goes ahead and organizes all of that.  That's what CPUs do very well.  If you go back to training, you're just doing the same thing over and over again where you are training these AIs.  That's what a GPU does incredibly well, and that's why you saw an Nvidia absolutely explode higher because CPUs weren't really valuable in training.  You move to chat-based AI, again, that's pretty decent for a GPU, but now when you move to agentic, that's great for a CPU.  So you went originally from about 8 GPUs for 1 CPU.  When you move to agentic, that ratio gets closer to 1-to-1.

[...]

Agentic [AI] requires 10 to 100 times more tokens than chat-based AI.

~ Dan Niles, "Dan Niles: Be Nimble - 30-50% AI Crash By 2027," The Master Investor Podcast with Wilfred Frost, 11:40 mark, May 12, 2026