Episode summary
Why Bubble Talk is Totally Wrong | TCAF 238
Ask The Compound
· · Hosts: Josh Brown, Michael Batnick · Guests: Ankur Crawford
Summary
Ankur Crawford argues the AI buildout is still early, not a bubble, because demand for compute is expanding faster than investors’ linear models can capture. She says the world is “short compute,” that agents and enterprise use cases are already consuming tokens faster than expected, and that the capex plans announced by hyperscalers still point to a supply-demand imbalance through at least 2026, 2027, and into 2028.
She frames Alger’s approach as seeking “unrecognized growth” in both high-revenue growers and mature companies undergoing life-cycle change. In her view, that has made semiconductors, memory, hard drives, and large platform names like Amazon especially interesting, because industry consolidation and rising demand are creating pricing power rather than a glut. She also says private-market exposure matters now because some AI companies are staying private until near-trillion-dollar valuations.
Crawford is more cautious about valuation regime shifts and says she raised 7% cash in February when the market felt off. But she still thinks the AI capex cycle is justified by returns that can be measured in months, not years, and that the market has not yet reached balance.
Key takeaways
- Ankur Crawford says the AI story is being underestimated because investors think linearly while compute demand is compounding exponentially.
- She argues the market is still short compute, with bottlenecks showing up first in chips and then in DRAM and storage.
- Crawford says bubble talk is “really stupid” because current capex and earnings revisions still support a supply shortage through 2026 to 2028.
- She views Amazon as a meaningful AI beneficiary because AWS, Anthropic, and in-house chips like Trainium can lower compute costs.
- She says Alger is investing in private AI companies because some are staying private until they are nearly a trillion dollars in value.
- Crawford raised 7% cash in February because market conditions felt wrong, even though she became more bullish on AI afterward.
Notable quotes
“what I have found is that sometimes it's almost easier if you don't understand something be like, "Oh, it must be a bubble”
“we are in a time of exponential growth and so it's really hard to get your arms around what exponential actually means”
“everyone is short compute”
“I do think that the bubble talk is really stupid”
“If you can sell something for a profit and buy it back 30, 40% cheaper, you you do that any day”
How the episode opens
I listened to my first Taiwan semi earnings call today. — Oh, you did? — It was very strange. I thought I'm sure you listened to a million of them. That the analysts ask questions and then the CFO repeats the question that has been asked for every single question. It's bizarre.
I think um — I don't know what's happened there because he he literally he's like, "All right, so the question was I'm like, "Huh, that's weird." And then he just did it the entire call. So, it used to be that there was like actually an audience and so people would ask in the audience and they were on stage.
Oh, that was that was how it used to be. And um — But you heard the call. Like I heard the question. I don't know. But that that's a possible explanation. — Maybe it's a cultural, you know, thing. — I thought it was odd that their their deck looks like it was made in the 90s, like their investment deck. What is that?
I don't know. — Well, I mean, it's it's only like a trillion dollar company, so you got to give them some time. But it looks like it was made on MS DOS like honestly like I I pulled one chart but it just looks like it hasn't been updated since 1984.
How do how do you listen to conference calls? Do you like you dial in or do you have you discovered quarter yet? — Yeah, I I I've have Quarter but we use something called ERA. — Oh, okay. What's that? — And so it's like it's similar to Quarter. You basically go through and you just click click through into a
Into a call.
From the episode’s captions: roughly the first two minutes, lightly cleaned. Keep watching on YouTube →
Topics
- ai capex
- compute shortage
- ai stock valuation
- semiconductors
- hyperscalers
Companies and tickers mentioned
- NVDA
- AMZN
- Nvidia
- Amazon
- Alger
Watch the episode
Watch “Why Bubble Talk is Totally Wrong | TCAF 238” on YouTube
Summary, takeaways and quote selection by Octagon. Quotes are short excerpts; the full conversation is in the episode.
More from Ask The Compound
Ask The Compound, this episode and its artwork belong to The Compound. Octagon is not affiliated with the show; this page is an independent summary.
