The Real Eisman Playbook · 3 min · Oct 19, 2026

Real AI demand does not guarantee profitable infrastructure

Danny Moses · Host, On The Tape

Danny Moses compares AI infrastructure spending with the fiber-optic buildout: the internet was real, but deployment and financing outpaced workable economics.

In this episode

16:50 AI financing echoes the warehouse lines behind subprime4 min
22:38 Private equity's fundraising model is colliding with exits6 min
5:04 Small Fed and Treasury moves cannot offset the debt backdrop3 min
8:30 Equity wealth now poses more economic risk than housing2 min
11:05Real AI demand does not guarantee profitable infrastructure3 min

Up next · How I InvestUnderwrite the team and milestones, not a distant market forecast6:05

“The math didn't work. You could have wrapped that fiber around the globe 10 times.”

Danny Mosesat 13:59 · The Real Eisman Playbook

Takeaways

The internet story was real, but fiber capacity was overbuilt.
Vendor financing helped fund equipment purchases during the boom.
AI's long-term demand does not settle which companies earn returns.

About this segment

Danny Moses compares AI infrastructure spending with the fiber-optic buildout: the internet was real, but deployment and financing outpaced workable economics. He says investors must distinguish lasting demand from who earns returns.

Danny Moses is a former Big Short partner and host of On The Tape.

Spotted the housing bubble before the crisis and helped execute the trade chronicled in The Big Short.

Steve Eisman, Host

From “Big Short Reunion w/ Danny Moses: Rising Rates, AI & Private Equity | The Real Eisman Playbook Ep 80”, Oct 19, 2026. Watch the full episode

Transcript

Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.

11:09
So, a couple of comparisons, and then every cycle is very unique. So, this secular trade is massive. It's a secular change, right, that I thought would, by now, have become more cyclical. What do I mean? NVIDIA's gross margin staying up in the high 70s. If you told me two years ago that we would still be in the high 70s, I'm like, there's no chance. There's no chance. At some point, competition comes.
11:29
Right. The memory chips. I know what's going to happen. I don't know when, but Micron, great, good for them. There's a reason the stock's traded eight to nine times earnings, because people don't care what earnings multiple trades at today. It's like, what does it look like? Is this a commodity? Is this going to be a commoditized product? And the answer is yes, at some point, because all you've got to do is build a fab and start making these chips at some point.
11:49
What really bothers me, and the comparisons that I see, it's a little bit of 1999-2000 and a little bit of what we lived through, GFC. What are the comparisons? 99-2000, the fiber optic craze. You and I were working together at the time. As everybody knows, Danny and I were at Oppenheimer. I was the financial services analyst. Danny was one of the more senior salespeople. And Danny used to take me on the road to go see clients.
12:16
I could tell that story. That would be a seven-hour podcast. Taking you to Alabama to an embassy suite with no hot water and you coming downstairs. That was quite something. But let me tell a story now that you bring it up. So, I started there in 1997. 1996, you were one of the first analysts I met. Merith Whitney was working for you. Vincent Daniel was working for you. And your first call on subprime auto and home equity lending, I'm a salesperson sitting there covering accounts. I cover hedge funds, state pension funds. And you got up on this podium and said, the following eight stocks are going to zero. And I turned to the guy I was kind of learning from, Barry Koretnik.
12:52
And I said, Barry, how do I make that call exactly? He goes, you don't make that call. I said, I don't make that call. He goes, you can't call a portfolio manager or analysts and say, the stock you own is going to zero. You get him over here. You talk a little bit, whatever. So, I grabbed Meredith. I grabbed Vinny. I grabbed you. I'm like, what is this Moody's trust securitization data? And why should ... Anyway, I wanted to tell that story because that looked ... Anyway, so, 99, 2000, Henry Blodgett, Jim Jungjahan.
13:17
We're living in this very similar ... Yeah, Henry Blodgett was an op and on him. Yes. And Jim Jungjahan, who was one of the ... Jungle Jim. Jungle Jim. There's a reason his name was Jungle Jim. covering JDS Uniphase and these things, right? And you sit there as a salesperson, and you sit there as an analyst covering financials, which you can't really meme, book value this, ROE this, it's hard to get past the stuff.
13:37
They can dream the dream. What was the dream? The dream was that fiber optics were changing the world, that the internet was growing so fast. They were right. They were right. But if you actually stop to do the math, which is what I'm going to get to in a second, on the amount of capex that was being spent, the amount of fiber that was being laid, the amount of fiber that was being leased, amount of equipment that was being leased by Lucent and Nortel, how Lucent and Nortel were basically, you know, providing loans.
14:01
to people to buy their equipment to do this stuff. The math didn't work. You could have wrapped that fiber around the globe 10 times. So stop and do the math back to where all this capex is on AI. And again, it's a real story. I'm not saying that it's not, but so was the internet. And there was winners and there were losers into that trade and trying to identify the secular and the cyclicality of various names within the AI play.
14:25
For example, I love Alphabet, Amazon,