The artificial intelligence boom is real — but there’s more to the AI trade than chips alone, according to a credit investor.
“This is a super-duper micro cycle that will outlast many investing careers,” said Scott Goodwin, the cofounder and managing partner of Diameter Capital Partners, a quote he attributed to his partner Jonathan Lewinsohn.
AI represents what Diameter Capital sees as a long-running, disruptive cycle — but buying the most obvious winners isn’t the only way to play it, he said on the “Goldman Sachs Exchanges” podcast published on Friday.
Diameter Capital, which manages approximately $25 billion in assets, has focused on where AI demand could create less obvious bottlenecks — and where those bottlenecks show up in credit markets.
The AI opportunity beyond chips
That view led Diameter to buy the unsecured debt of a midsize telecommunications company in 2023.
Goodwin said the bet was rooted in the idea that as companies move from training AI models to actually using them, demand shifts away from chips alone and toward the networks that carry data.
“It had to leave the data center. How would it leave? It would leave on the commercial fiber, the pipes,” he said.
The telco went on to sign more than $10 billion in contracts with hyperscale cloud providers, and the debt has rebounded to face value, Goodwin said.
Diameter Capital also made “a big bet” on a satellite company tied to the wireless spectrum — a wager that later paid off after the company sold spectrum assets and the debt returned to face value.
Goodwin’s comments come amid growing debate over whether sky-high AI valuations are sustainable and whether investors are overlooking other opportunities tied to the technology.
Risks and rewards
Goodwin warned that parts of the AI-credit boom may be taking on risk that’s hard to price, especially in chip finance.
Some investors, he said, are taking on “residual risk,” or the riskiest slice of chip-financing deals — betting on what the hardware might be worth years from now. Cutting-edge firms refresh their technology often, so chips can quickly become outdated for some customers.
“We call up really smart people in Silicon Valley, we call up really smart people at Big Tech companies and ask them what the residual value is on these chips three, four, five, six, seven years forward,” he said. “None of them have a clue.”
Goodwin said the next phase isn’t just about spending on infrastructure — it’s about competitive disruption rather than capital expenditure.
“Who are the companies, who are the entities that are going to adopt AI and take a step forward versus their peers? And who are going to be the losers?” he asked.
“That is actually a longer cycle than the capex cycle, so that’s really interesting,” he said.
