Episode summary
AI’s Next Big Leap
Thoughts on the Market (Morgan Stanley)
· · Hosts: Tom Wigg, Stephen Byrd
Summary
The host argues that investors are still thinking about AI too linearly, while the guest says the pace of model improvement is becoming nonlinear and could soon look much different from the last two years. He cites a rough scaling law: 10x more training compute has historically produced about 2x better model capabilities, and says multiple frontier models should show much stronger performance across a larger share of the economy through the spring and summer.
The guest also says the economics are favorable for both hyperscalers and adopters. He says token economics can generate excellent returns for data-center owners, while enterprise users can already realize large savings at very low token costs. In his view, compute remains scarce, which supports pricing power for those with the power and infrastructure.
On market timing, the guest says this is likely a 2026 event rather than a later-cycle story. He expects fast adopters to show tangible productivity gains first, with broader enterprise uptake lagging behind. He also argues that as AI shifts from query-based use to agentic workflows across knowledge work, token usage should rise sharply, reinforcing demand for compute and the rest of the AI infrastructure chain.
Key takeaways
- The host says the market needs to stop assuming AI improvement will be incremental and start planning for exponential progress.
- The guest says a 10x increase in training compute has historically produced about 2x better model capabilities.
- The guest expects the strongest AI revenue and productivity signals to show up in 2026, led by fast adopters rather than average enterprises.
- The guest says compute is scarce enough that hyperscalers with power and infrastructure should retain pricing power.
- The guest argues that agentic AI could drive about a 10x increase in token usage versus query-based usage.
Notable quotes
“The market has been thinking about AI in linear terms, but we need to reframe that assumption of only incremental improvement and think about exponential improvement”
“approximately the law is if you increase the training compute by 10x the capabilities of the models go up by 2x”
“we are expecting through the spring and summer to see multiple models that are able to perform a much greater percentage of the economy”
“I think this is a 2026 event”
“when you go from a query based usage of LLMs to an agentic use for any occupation, you see about a 10x increase in token usage”
How the episode opens
Welcome to Thoughts on the Market. I'm Tom Wig, head of specialty sales in the Americas at Morgan Stanley and a sector specialist in technology, media, and telecom. We wake up every day to new AI product releases, so it's easy to lose sight of the unprecedented nonlinear improvement in AI capabilities, but things are about to get weird.
It's Tuesday, April 28th at 8 a.m. in New York. The market has been thinking about AI in linear terms, but we need to reframe that assumption of only incremental improvement and think about exponential improvement.
That was my takeaway from a conversation with Steven Bird, global head of thematic and sustainability research at Morgan Stanley. In our conversation, we zeroed in on Steven's bullcase for broader AI model improvements. First, I want to talk about one obsession that you've been writing about for the last several months is this idea that we're going to see nonlinear improvements in the frontier models coming out this spring.
Yes. — There's been, you know, some big headlines around new models, benchmarks coming out publicly. Is this, you know, your your bull case playing out on these models and what are the implications?
Absolutely, Tom. So we have to your point we are obsessed and I no I'm not shy about that with the nonlinear rate of AI improvement. It is the most important impact to so many stocks that I can think of in the sense that it can impact all industries all business models.
So what we've been saying for some time is if you look back over the last couple of years at the relationship between the amount of compute used to train these LLMs and the capabilities we have a very clear scaling law and approximately the law is if you increase the training compute by 10x the capabilities of the models go up by 2x now as I' you've talked about this a lot just meditate on that for a moment I think things are about to get weird in the sense that on the positive side we're going to see all kinds of underappreciated capabilities across many industries.
From the episode’s captions: roughly the first two minutes, lightly cleaned. Keep watching on YouTube →
Topics
- ai breakthroughs
- ai development
- machine learning acceleration
- future of ai
- technology drivers
Watch the episode
Watch “AI’s Next Big Leap” 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.
