Episode summary
AI’s Next Stress Test
Thoughts on the Market (Morgan Stanley)
· · 12 min · Guests: Thomas Wigg, Stephen Byrd, Ariana Salvatore
Summary
AI capex beneficiaries have pulled back as investors worry about three things at once: open-source models taking share, enterprises pushing back on token spending, and political resistance to new data center builds. Stephen Byrd argues that token economics still look favorable for enterprises because typical use cases can generate far more value than the cost of the tokens required, so he is not worried that “token maxing” will derail demand.
Byrd also says the market will likely support both proprietary frontier models and cheaper open-source models. In his view, more mundane workloads will shift to lower-cost options, while higher-end models will retain value in cases where accuracy matters and mistakes are expensive. He thinks the broader setup still resembles Jevon’s paradox: as models improve and prices fall, usage rises and total compute demand keeps outpacing supply.
Ariana Salvatore says the data center backlash is becoming a real bipartisan issue, but she does not expect a federal ban while the Trump administration is in power. She expects more conditional buildouts instead, with projects increasingly forced to include community benefits, grid modernization, or other concessions. Byrd adds that the result could be more off-grid projects using natural gas turbines, fuel cells, and energy storage, especially where local permitting becomes harder.
Key takeaways
- Byrd says enterprise AI use cases can deliver about $55 of value for only a few dollars of token cost, so token spending still looks economically justified.
- Byrd expects both frontier models and open-source models to flourish, with cheaper models serving mundane tasks and premium models winning where errors are costly.
- Byrd argues AI capex should still follow Jevon’s paradox, with lower model prices driving more usage and stronger demand for compute.
- Salvatore says a federal ban on data center buildout is unlikely under the Trump administration, despite rising local and state opposition.
- Byrd says political friction will likely push developers toward conditional or off-grid data centers, including natural gas turbines, fuel cells, and more energy storage.
Notable quotes
“for a few dollars of token costs, an enterprise can generate benefit of $55”
“It does look that way very much and the chevon's paradox dynamic is what we still see today”
“Our view is that as long as the Trump administration is in power, something like a federal ban is unlikely to come to fruition”
“What I'm increasingly seeing is that the developers are going to go off-rid and they just don't want to show any impact to the community”
How the episode opens
Welcome to Thoughts on the Market. I'm Tom Wig, Morgan Stanley's head of America's specialty sales. — I'm Steven Bird, Morgan Stanley's global head of thematic and sustainability research. — And I'm Ariana Salvatore, Morgan Stanley's head of public policy research.
Today, the rally in AI capex beneficiaries has taken a breather in recent weeks on concerns of competition from open source models, backlash to token maxing, and growing political opposition at data center builds. It's Tuesday, July 7th at 10 a.m. in New York.
Let's start with you, Stephen. There's a lot of discussion recently around a backlash at token maxing, essentially enterprises trying to curtail their high spending on AI tokens from the frontier labs and in many cases shifting to cheaper open-source China models. Can you first offer some perspective here on the value of tokens for enterprises?
I know you have a popular token factory model that walks through the economics of agents. Yeah, Tom, we uh we do have this model that really walks through token economics both from the adopter side as well as the hyperscaler side. So, let's do the adopter side.
So, there's a study out that shows a whole range of enterprise use cases of AI. And the average single use case that they identify would save a company about $55 or provide that that much benefit. And while we don't know exactly how many tokens it will require, we can make some educated guesses as to a typical token usage to achieve that $55 outcome.
And we know that a typical American model, though this varies a lot, you can think of as a cost per million tokens being in the range of $5 per million. Some will be lower, some will be higher. So for a few dollars of token costs, an enterprise can generate benefit of $55.
From the episode’s captions: roughly the first two minutes, lightly cleaned. Keep watching on YouTube →
Topics
- ai stock durability
- investment cycle
- competition in ai
- public policy research
- thematic research
Companies and tickers mentioned
- MS
- AAPL
- Morgan Stanley
- Apple
- Spotify
Watch the episode
Watch “AI’s Next Stress Test” on YouTube
Summary, takeaways and quote selection by Octagon. Quotes are short excerpts; the full conversation is in the episode.
More from Thoughts on the Market (Morgan Stanley)
Thoughts on the Market (Morgan Stanley), this episode and its artwork belong to Morgan Stanley. Octagon is not affiliated with the show; this page is an independent summary.
