Thoughts on the Market

Our CIO and Chief U.S. Equity Strategist Mike Wilson explains the factors behind stock gains across sectors.
Read more insights from Morgan Stanley.

----- Transcript -----

Welcome to Thoughts on the Market. I'm Mike Wilson, Morgan Stanley’s CIO and Chief U.S. Equity Strategist.
Today on the podcast I’ll be discussing why earnings remain the most important variable for equity markets.
It's Monday, May 4th at 2pm in New York.
So, let’s get after it.
The more I think about what’s been driving this market, and the more time I spend with the data, the more I keep coming back to the same conclusion: it’s earnings. Not the headlines, not even the Fed. Earnings are doing the heavy lifting right now.
When I look at this reporting season, what stands out isn’t just resilience, it’s strength that’s broader than most people appreciate. The typical company in the S&P 500 is growing earnings at about 16 percent, and the median earnings surprise is running around 6 percent. That’s the strongest we’ve seen in four years.
What’s really interesting to me is that this strength is no longer confined to just the biggest tech names. Yes, hyper scalers and semiconductors are still playing a leading role, but the story is expanding. We’re seeing earnings revisions move higher across Financials, Industrials, and Consumer Cyclicals, in particular. That kind of breadth tells me this isn’t just a narrow leadership story; it’s something more sustainable.
At the same time, many investors are focused on the geopolitical backdrop, particularly the Iran conflict and what it means for oil, inflation, and supply chains. To be fair, companies are feeling some of that pressure. When you listen to earnings calls, you hear about rising freight costs, tighter supply chains, and higher input prices across industries like chemicals and machinery.
But here’s the nuance: those impacts are uneven. They’re not hitting the entire market in the same way. In fact, at the index level, they’re being offset. Energy has become a positive contributor to earnings growth, and the higher-end consumer remains relatively strong. Even with higher fuel costs, we’re not seeing a meaningful pullback in overall consumption – at least not yet.
That tells me that we’re not dealing with a classic demand shock. We’re dealing with a redistribution of pressure, and companies are adapting. In many cases, they’re passing through higher costs. Revenue surprises are running above historical norms, which suggests pricing power is improving.
Now, of course, earnings aren’t the only piece of the puzzle. Policy still matters, and the shift in rate expectations this year has been meaningful. The Fed has clearly become more concerned about inflation, and the market has repriced expectations to fewer cuts, and maybe even a higher probability of hikes. That repricing is a big reason why valuations corrected so sharply over the past six months.
It’s notable that even with that headwind, equities have managed to stabilize, thanks to earnings. When earnings are growing at an above-trend pace, equities can deliver solid returns regardless of whether the Fed is cutting or not.
That said, I do think that there’s one area of risk that deserves further attention, and that’s liquidity. We’ve seen periods of funding stress over the past six months, and those moments have coincided with pressure on valuations. The Fed and the Treasury have stepped in at times to stabilize these conditions, helping to reduce bond volatility and support equity multiples.
Bottom line, we have already had a meaningful correction in valuations this year with price earnings multiples falling 18 percent from their peak last fall. That adjustment occurred as the market digested the many risks that we have been highlighting. Meanwhile, earnings are not only holding up, they’re accelerating and broadening across sectors. The risks that we’ve all all focused on – geopolitics, oil, supply chains – are real. But they’re being absorbed at the company level. As a result, the price declines were much more modest than the compression in valuations.
Meanwhile, monetary policy is providing some headwinds, but it’s not overwhelming the earnings story. Equity markets move on two things: earnings and liquidity. Right now, earnings are more than offsetting the lingering liquidity concerns. In short, earnings growth is greater than the valuation reset. This is classic bull market behavior and as long as that continues, I think the U.S. equity market will grind higher for the rest of the year with intermittent bouts of volatility.
Thanks for tuning in; I hope you found it informative and useful. Let us know what you think by leaving us a review. And if you find Thoughts on the Market worthwhile, tell a friend or colleague to try it out!

Our Global Chief Economist and Head of Macro Research Seth Carpenter discusses whether the economy can adapt fast enough to turn AI into a productivity boom rather than a labor market shock.
Read more insights from Morgan Stanley.

----- Transcript -----

Seth Carpenter: Welcome to Thoughts in the Market. I'm Seth Carpenter, Morgan Stanley's Global Chief Economist and Head of Macro Research.
Today we're going to try to look past the hype and the anxiety around AI and ask what will be the effect on the labor market.
It's Friday, May 1st at 10am in New York.
Now, odds are that you've used AI to draft an email or summarize a document, maybe learn about a new topic, help plan a trip. The new technology is clearly lowering the cost of certain tasks. And I think the research shows that there are plenty and an increasing number of tasks that AI can do better than most humans. But that's not really the question.
What I hear all the time is, ‘Well, if we can get the same amount of output with less labor, then surely millions of people will lose their job.’ I think the same logic also implies that we can just get a lot more output from the economy using all the labor that we have. And the difference between those two views really is at the heart of the debate.
So far, I would say the data allow for some cautious optimism. Despite rapid advances in AI capability and evidence that adoption is spreading, the broad labor market indicators still show remarkably little disruption. Economic growth is holding in there. The unemployment rate is not rising rapidly. If anything, it's ticked down recently. Job openings are not soaring, and separations do not suggest that there's systematic weakness in AI exposed industries.
Now, productivity data are beginning to show perhaps a bit of AI's positive effects, but they don't show the mass displacement that many people fear. According to our research, industries with higher AI exposures have recorded stronger labor productivity gains, driven mainly by faster output growth rather than fewer hours worked. And that distinction for me is critical. So far, the evidence looks like workers are producing more than firms are cutting back on labor.
There's also a physical constraint. AI adoption depends – and will continue to depend – on infrastructure that is still being built. Of the more than $3 trillion in expected data center and related infrastructure CapEx from 2025 through 2028, only about a quarter of that has been deployed so far.
The future remains opaque. No two ways about it. The biggest productivity gains from my perspective are likely still ahead of us, and some job losses are likely unavoidable. Earlier, innovation waves unfolded over decades, and AI is moving much faster, compressing the adjustment period. And that does create the central risk to the labor market; that job destruction happens faster than new job creation happens.
And so, what our research has been doing is to try to look beyond the immediate effects. Yes, some jobs and tasks will likely be disrupted. But higher productivity can also mean higher incomes. Higher wealth. With higher income and higher wealth can also mean higher spending, which, in turn, drives the economy faster.
Inside corporations, new tasks and new roles will likely emerge giving some of the displaced workers somewhere else to go. And even if employment does slow down for a while – and that could put downward pressure on inflation and maybe upward pressure on the unemployment rate – I don't really think policy makers are simply going to sit back on the sidelines. Central banks can respond by trying to stimulate the economy and bring it back towards full employment.
This is something that economists call General Equilibrium. We can't look simply at one side of the equation. We have to think about the system as a whole. And I have to say, if monetary policy runs out of room, fiscal policy makers can get into the game as well. Between automatic stabilizers like unemployment benefits and directed targeted government action, there's another way in which the economy could be pushed back to full employment.
So, the bigger point is this, AI clearly has a chance to create some labor market disruption, but the economy has all sorts of other systems and levers in place that can pull us back to full employment.
And with those buffers in place, any rise in the unemployment rate from AI is probably going to end up being smaller, shorter, and easier to manage – at least for the next couple of years than maybe some of the first pass analysis that I've seen suggests.
AI's labor market impact is not predetermined. The debate will almost certainly come down to speed. How fast is AI adoption relative to the economy's ability to adapt? History suggests that productivity ultimately wins. The economy gets bigger and people stay employed. History also tells us that not everyone benefits equally. And more importantly, not every transition is smooth.
So, what does that mean? Should we be just blithely optimistic? Absolutely not. For now, the early evidence is reassuring, but the story is still being written.
Thanks for listening, and if you enjoy this show, please leave us a review wherever you listen. And share Thoughts on the Market with a friend or a colleague today.

Capital spending usually signals how a company is positioning itself for the future. Our Global Head of Fixed Income Research Andrew Sheets explains why this metric is getting more attention from investors.
Read more insights from Morgan Stanley.

----- Transcript -----

Andrew Sheets: Welcome to Thoughts on the Market. I'm Andrew Sheets, Global Head of Fixed Income Research at Morgan Stanley.
Today: Why capital expenditure is rapidly becoming one of the most important numbers in earning season across asset classes.
It's Thursday, April 30th at 2pm in London.
This is a high-risk episode in the sense that it may already be obsolete by the time that you hear it. But then again, maybe that's fitting for a discussion of record capital spending on cutting edge technology.
We are in the middle of the busiest part of earning season, and yesterday four of the largest companies in the world reported numbers. These companies – Alphabet, Amazon, Microsoft, and Meta – have a combined market cap of nearly $12 trillion.
Yet, while the focus of earning season is traditionally about earnings, another line item is rapidly rising in importance. Capital spending on AI infrastructure – the chips, power cooling, and connections that are required to build and run AI models is soaring. And the companies that reported yesterday are at the leading edge of this trend.
The first thing about all this spending is simply the scale. For this year alone, Morgan Stanley estimates that it will amount to over $600 billion across the largest U.S. hyperscalers. To put that in perspective, that means just a handful of U.S. tech companies are now set to spend almost as much on capital and equipment this year as every non-technology company in the S&P 500 did in 2025. And as big as that spending is, it's been accelerating.
That over 600 billion spending number that we forecast for 2026? Well, a year ago we thought it would be roughly half that, and that estimate was well above consensus at the time. U.S. companies have repeatedly guided their spending higher as they seek to capture the AI opportunity. And we think that continues.
By 2028, my Morgan Stanley colleagues estimate that this U.S. hyperscaler capital spending could hit an annual rate of $1 trillion. In other words, as big as these numbers may seem, much of the spending story still lies ahead.
All of that investment, both recently and in the future, has big implications. First, one company's spending is another company's revenue, and many of the stock markets recent winners have been directly tied to this historic buildout.
As of this recording, U.S. semiconductor stocks have risen over 30 percent this month alone.
Second, while these large U.S. tech companies have enormous financial resources, this spending is at a scale that still requires significant borrowing. Our credit strategy teams expect record bond issuance this year, with U.S. tech borrowing a big part of that.
And so far, it's playing out. The first quarter was the busiest quarter for U.S. investment grade bond issuance on record. Which brings us back to these recent earnings – and a dilemma that seems negatively skewed for credit relative to equities.
If these companies continue to sound confident about their capital spending plans or even raise expectations further, that could support AI suppliers and the broader equity market. But it would mean even more borrowing needs to be absorbed by the corporate bond market, a credit negative. The results we got yesterday certainly hint at a continuation of this trend.
On the other hand, if capital spending is guided down, that could undermine a key pillar of recent market strength and broader risk appetite, which could drag credit wider by association. In the near term, the risk reward seems better in other parts of fixed income, such as mortgage-backed securities.
The implications of yesterday's results may also extend to the Federal Reserve. As we discussed last week, Kevin Warsh, nominee to be the next Fed Chair, believes that large levels of investment can boost productivity, lowering inflation, and thus justifying lower interest rates.
And so, what these large spenders do, how confident they feel about the future, and what all of this spending can ultimately deliver – well, the implications of that may extend even into the monetary policy story.
Thank you as always, for your time. If you find Thoughts of the Market useful, let us know by leaving a review wherever you listen. And also tell a friend or colleague about us today.

Still six months out, the U.S. midterm elections are likely to influence government initiatives to deal with higher energy costs. Our Head of Public Policy Research Ariana Salvatore and Global Chief Economist Seth Carpenter discuss how the Congress and the Fed might react.
Read more insights from Morgan Stanley.

----- Transcript -----

Ariana Salvatore: Welcome to Thoughts on the Market. I'm Ariana Salvatore, Head of Public Policy Research for Morgan Stanley.
Seth Carpenter: And I'm Seth Carpenter, the firm's Global Chief Economist and Head of Macro Research.
Ariana Salvatore: Today we're discussing the run up to the midterm elections and what it could mean for the macro outlook and policy response.
It's Wednesday, April 29th at 10am in New York.
Last week, Mike Zezas and I talked through the midterm elections and their potential consequences for the economy and markets. This week we figured it might be helpful to talk about the setup into November, especially as we're both increasingly being asked about the macro outlook and potential for targeted stimulus to offset the oil shock.
So, Seth, let's start there. we know cost of living is a key issue in elections, and we've seen a pretty meaningful oil shock feed through markets. How are you thinking about that in the context of the broader economy?
Seth Carpenter: Our U.S. economics team has estimated that the higher gas prices that we have now and likely to have for the rest of the year are going to be more than enough to offset any boost to consumer spending from the higher tax refunds this year. So, I think that's the first point.
If you're expecting a boost to come through that channel, you probably want to unwind that. And In fact, overall, what we've done is lowered our forecast for U.S. growth by about three or four tenths of percentage point worth of growth this year because of the higher energy prices. So, it's a drag on spending, I think, no matter how you cut it.
Ariana Salvatore: And that's not happening in isolation, right?
Seth Carpenter: No, that's exactly right. That's exactly right. We've also got at least somewhat restrictive monetary policy layered on top. So, financial conditions are already a little bit tight and the oil price shock sort of amplifies that tightening by weighing on spending. That's going to be really important.
I think an extra complication then is what does it do to inflation? For now, we don't think it's going to be that big of a deal. History says at least looking at the data that when energy prices go up, when oil prices go up, gasoline prices go up. It does boost headline inflation for sure, but the pass through to core inflation is pretty limited, and the effects tend to go away on their own without too much time.
So, I think the real hit here is going to be from the higher costs acting like a drag on consumer spending.
Ariana Salvatore: Right. And importantly, it's a very visible shock. Gasoline prices feed directly into how consumers and voters perceive the economy, which brings us into the political overlay as we approach the midterms…
Seth Carpenter: Yeah, I think that's exactly right. And whenever we economists are thinking about inflation and prices and consumers, we think about exactly that – what we call salience, just how visible are these prices. And gasoline prices tend to be some of those prices that stick out in people's minds.
So, if people are seeing it. And people are reacting to it, give me some idea of what the Congress can realistically do between now and the midterm elections.
Ariana Salvatore: Well, I would say in theory there's a range of options. Direct stimulus, targeted transfers. We tend to frame affordability policies across five vectors: energy, healthcare, housing, consumer credit and trade policy. But in practice, the constraints are pretty binding right now and as we've been saying, tariff policy is really the only lever the president can pull easily to have a real impact on voters.
Seth Carpenter: All right. So, you said constraints and constraints for the Congress. Can you walk us through what those constraints are?
Ariana Salvatore: Sure. So, the first and most obvious is deficits. We're already running large fiscal deficits in the U.S., and I would say there's limited political appetite to expand them meaningfully from here in the near term, especially heading into an election.
The second is procedure. If you want to pass something sizable, you're either looking at reconciliation, which requires political alignment in a number of procedural hurdles. Or bipartisan cooperation to get around the filibuster. Both seem difficult to us in this environment.
Seth Carpenter: So my experience in Washington for a couple decades of working on policy is that when things are difficult, they tend to take more time. So how does the timing component of all of this matter, and how does it fit into the way that you're thinking about it?
Ariana Salvatore: Timing is the third constraint. The legislative calendar in particular. What we see is as you get closer to midterms – really any election – the window for passing major legislation narrows pretty quickly. That's because lawmakers shift their focus toward campaigning, and the agenda itself just becomes more limited.
And then to finish off the constraints, the fourth I would say is implementation. Even if something were to pass, there's a lag between legislation and the actual economic impact. Getting funds out the door, whether it's checks or programmatic spending, tends to take time.
Seth Carpenter: Yeah, even well targeted policy might not hit the economy in time to have the desired effect before the election.
Would you agree with that?
Ariana Salvatore: Yeah, but for argument's sake, let's say we're wrong on that and Congress does manage to pass something. Maybe not a broad-based stimulus package, but let's say some form of targeted relief.
From a macro perspective, what do you think would matter most? Is it the size of the package, how quickly it gets implemented, or which consumers are targeted?
Seth Carpenter: Yeah, I'm going to have to say a little bit of all of the above. I mean, economic analysis really tends to show that tax cuts tend to simulate less than increased spending and transfers matter. But it matters to whom those transfers happen.
So, I do think if we're aiming at the lower end of the income distribution, probably has a higher propensity to spend; and so, you're more likely to see more of those dollars getting spent and faster – if that's where it's going. The size of the package has to matter as well, because more money out probably means more money getting spent. But I will add, there are two caveats this time around that we probably need to take into consideration.
First, with the increase in tax refunds that we've seen this year, survey suggests that households are using that money to pay down outstanding debt more than they would historically. And so, we might be in a situation because of the past couple of years of affordability issues where households are going to try to get ahead of things and pay down some of that debt. And as a result, maybe there's a more muted effect on spending.
And second, we are living in a world right now where inflation is well above the Fed's target. So, if the extra stimulus leads to extra spending at a time when prices are already high, well, there's a chance we might give an extra boost to inflation and then the Fed would have to reconsider what it's doing on monetary policy.
But you said Congress is probably constrained. So, let's shift then and ask, is there something that the president could do unilaterally with executive authority? And in particular, sometimes I get this question from clients, even if there's not clear, well-defined legal authority. We've seen something like that before with the tariff policy under the IEEPA authority. It was imposed and then later it was pulled back when it was judged by courts not to be the right authority.
So, why wouldn't we think – the argument goes; why wouldn't we think that some sort of large scale maybe rebates or direct payments, could get deployed quickly, even if the, let's say, legal authority is a little bit murky?
Ariana Salvatore: Yes, it's an interesting question, but I think there are a few important distinctions that make something like the administration sending out checks, for example, very different from tariff policy. First, fiscal transfers are much more clearly tied to congressional authority, legally speaking.
Spending power, as you know, resides in Congress, and that's a pretty firm constitutional boundary. And importantly, even something like tax refunds, which can look like direct payments aren't discretionary. They're preauthorized in the tax code, and Treasury is just returning overpayments under a standing appropriation. So, there isn't really a comparable mechanism the administration could use to send out broad-based checks, for example, without new legislation.
Now, trade authorities by contrast, have historically allowed for more executive flexibility, even if contested, like we saw with the IEEPA tariffs. Direct fiscal outlays are different. You generally need explicit appropriation. And then second, there's the operational side to all of this. Even if you were to set aside the legal questions, there isn't a standing mechanism for distributing very large sums of money quickly without legislative backing.
Seth Carpenter: Fair enough. And if we stay in this totally hypothetical world, what would you imagine would be the timing of any legal challenges if they did happen?
Ariana Salvatore: In a scenario like this, you'd likely see challenges fairly quickly and courts could intervene early in the process, potentially before funds are even fully dispersed. So, Seth, the idea that you could deploy something on a massive scale and only deal with the legal consequences much later is all the more uncertain.
But Seth, let's stay with the upside risk scenario for a moment. If Congress did pass something targeted instead, where would you expect policymakers to focus? Can we talk through maybe energy rebates, child tax credits, SNAP or nutrition support… Or do you think something else aimed at the most rate sensitive or cost of living sensitive households might make more sense?
Seth Carpenter: Yeah, I think you've laid out there a pretty rational strategy for trying to make things targeted for the people who are going to be feeling this affordability crunch the most. And so, the SNAP benefits, like you said, are nutrition support. That's lower income households, families with children, people who really are living paycheck to paycheck and noticing these higher prices.
Energy subsidies or some sort of tax rebate – again, trying to target where the pain is most acute; the higher electricity prices, the higher gasoline prices that people are noticing, that people are feeling. I think all of that seems very plausible.
I just want to flag though, that there is this possible hidden effect, which is the more these policies mask the higher cost, the economic pain from the higher energy prices – the more it allows people to keep spending despite the higher prices. And that spending with higher prices, well, that could easily lead to a tick up in inflation.
That could lead to a change in the Fed's reaction function. And if it was strong enough, if growth picked up enough and inflation picked up from here, you could easily see the Fed hiking rates instead of cutting.
Ariana Salvatore: So, in other words, even if the policy surprise is maybe good news for consumers in the near term, markets would still need to think through whether it extends the inflation problem or changes the expected rate path.
Seth Carpenter: I think that is exactly right. I think this is very much a case where good news could be good news, but there are going to be lots of details.
So maybe if we take a step back, we've got a constrained Congress, maybe limited scope for unilateral action and a macro backdrop because of inflation that's probably already under some pressure.
Ariana Salvatore: Which means the key drivers heading into the midterms later this year are likely to remain the ones that are already in place: energy prices, monetary policy, and underlying growth dynamics rather than potential new fiscal stimulus.
Seth Carpenter: And so that means for markets, focus needs to stay on the fundamentals.
Ariana Salvatore: Exactly. Elections can shape the policy path at the margin, but the macro cycle is doing most of the heavy lifting here. And we think that's the case following the midterms as well. If you'd like more detail there, please go ahead and listen to our podcast from last week on this topic.
Seth, thanks for taking the time to talk.
Seth Carpenter: Ariana, thank you for inviting me. And for the listeners, thank you for listening. If you enjoy Thoughts on the Market, please share it with a friend or colleague today. And leave a review wherever you listen to podcasts.

Tom Wigg and Stephen Byrd discuss the accelerating pace of AI breakthroughs, the forces driving them and why the next phase of development may look very different from anything we’ve seen so far.
Read more insights from Morgan Stanley.

----- Transcript -----

Tom Wigg: Welcome to Thoughts on the Market. I’m Tom Wigg, 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 non-linear improvement in AI capabilities. But things are about to get weird.
 It’s Tuesday, April 28th at 8am 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 Stephen Byrd, Global Head of Thematic and Sustainability Research at Morgan Stanley. In our conversation, we zeroed in on Stephen’s bull case for broader AI model improvements.
 Tom Wigg: 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.
Stephen Byrd: Yes.
Tom Wigg: There’s been, you know, some big headlines around new models, benchmarks coming out publicly. Is this, you know, your bull case playing out on these models? And what are the implications?
Stephen Byrd: Yes! Absolutely, Tom. So we have, to your point, we are obsessed. And I know 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 you and I’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. So this disruption discussion, I think, is going to spread, but it’s also going to require investors to, kind of, be more thoughtful about what they do with that concept. Meaning you can’t sell everything. In the sense that AI will disrupt some businesses.
I actually think this is healthy in some ways because now it forces investors to really look at each business model and assess which is going to get disrupted, which can get supported and enabled by AI, which are immune. Because there are some business models that actually are immune.
But essentially from here, Tom, I’d say we are expecting through the spring and summer to see multiple models that are able to perform a much greater percentage of the economy at better levels of accuracy at incredibly low cost. Which I know you and I have talked a lot about the cost of actually doing this work from the LLMs.
This is massive. This is going to impact so many industries. I think this is all to the good for the AI infrastructure plays because it shows the importance of getting more intelligence out into the world.
Tom Wigg: So, you mentioned the constraints we’re seeing across compute, memory and power. It seems like most of the CEOs of the labs and hyperscalers are talking about this. Investors are bullish in terms of the ownership in, you know, memory, optical, semi-cap, et cetera. But the question I’m getting more recently is around what’s the ROI on all this spending. And does the market action in these hyperscalers, which have been pretty bearish year-to-date, force a cut on CapEx? So, maybe if you can marry that with what you’re picking up on the ground in terms of compute spend and whether the frenzy still continues, you know, versus the ROI? And, like, what could happen?
Stephen Byrd: Yeah. The short answer – I’m going to go through detail – is I think the bullishness is going to get more bullish over the coming months. And let me walk you through a couple of the mathematics and then just what I’m seeing on the ground to your point, Tom.
So the mathematics. We have a token economics model that looks from the perspective of a hyperscaler or an LLM developer in terms of – if they sell their token at a certain price and you fully load the cost of a data center and all associated costs, financing, you name it – in what are the returns? And the bottom line is the returns are excellent.
The other element we spend a lot of work on, and you and I talk a lot about, is the demand for compute. In this world where the LLMs are increasing in capability and the token usage goes way up with agentic AI, video world models, all that stuff, we think that there is a massive shortage of compute. So, if you’re lucky enough to be a hyperscaler with the compute, with the power, we think that they will have a lot of pricing power on the tokens.
Let me explain why we see price power on the tokens. Now I’m going to flip to the perspective of an adopter. Let me give you just rough mathematics. There was a study last year from one of the big labs showing that on average, an enterprise user using an LLM might be able to replace work that would take about one and a half hours from a human. That would save about $55 of cost. A million tokens, depends on whether you’re looking at input or output – but let’s just call it $5 for a million tokens.
The average usage case today for a fairly complex agentic task in an enterprise setting is in the tens of thousands of tokens. Okay? So let’s just do that math again. $55 of savings. A million tokens cost $5, and a typical agentic usage is far less than the million tokens today, though that will accelerate. The economics are a home run for adopters.
So, we’re in a situation where compute is very scarce. I see pricing power all over the place for those who have the compute and have the power.
Tom Wigg: So, when you put it like that, Stephen, it seems so inevitable and obvious. But I wonder why the hyperscalers are trading the way they are? And when do they see the revenue inflection you’re talking about? Is this like a stay tuned kinda 2026 event? Is this something we have to wait for for 2027-2028?
Like, how do you think this flows through to the extent that the market will get more comfortable that all this free cash flow pressure is worth it on the other side?
Stephen Byrd: Yeah. This is, in short, I think this is a 2026 event. But let me dive into that because what you just asked is so important for so many stocks.
So, let’s talk through this. The capabilities of the models are advancing so fast that the average corporate user is not yet keeping up. There is this gap. But that will happen quickly, and we’re seeing signs from these labs of revenue at the lab level that is accelerating. So that’s a good sign.
What we’re seeing, though, among fast adopters is those adopters who really understand the capabilities are quickly realizing just how economically beneficial there is. An example, one of my best friends founded a software company many years ago. Last month was – that was the last month in which his programmers wrote code. They’re done with writing code.
The efficiency benefits for his business are absolutely massive. But he feels like he’s just scratching the surface, and he’s about as technically capable as anyone I know. He has two PhDs in the subject matter. He’s very, very good.
So long way to say that we’re living in almost two worlds where the fast adopters will show what’s possible. The average utilization for enterprises will still take some time. But I do think that the market will react to what they see from the fast adopters in the sense of – the tangible economic benefits are so big.
Now, on the ground, what I’m seeing on the infrastructure side, my friends in power tell me that a couple months ago is when they saw the sense of urgency from the AI community go up a couple of notches for them to get the infrastructure they need. So they saw this explosion in compute coming. In the last two months, the weekly usage of tokens according to OpenRadar is up a couple hundred percent in a couple months.
So, I do think we’re seeing this. So, this is; it’s happening quickly. What I would say is the market will have these signposts in every industry of early adopters showing this benefit. I think that’s enough for us to start to get bullish. We also… I just think when you look at the demand for compute, the compute numbers need to go up. And with that, you know, everything in the AI value chain, infrastructure value chain, the volumes need to go up.
Tom Wigg: One bear case that I wanted to interrogate was – there’s one view that, yes, there’s a token explosion right now. But it’s because the first use case is coding. Which is inherently, you know, very developer-friendly and token-intensive relative to other knowledge work.
Can you talk about, you know, whether you subscribe to that? Or whether the token intensity will be as high or lower as this expands to other areas of knowledge work in the next several years?
Stephen Byrd: Yeah, it’s a great question. The short version is that, yes, it’s true that software usage is more token intensive. However, what we’re going to be seeing – we’re starting to see it – is in almost every knowledge-based job, we’re going to move to agentic AI. And when we do that, you tend to see an explosion in compute.
Let me walk you through the numbers. There are a couple studies that show essentially 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 per use of those models. And you can see why.
I’ve anecdotes of some of my friends who are newer to this – who set their agents loose overnight to do non-coding work. And in the morning they get some pretty amazing results. But they also used a lot more tokens than they’d expected … (laughs)
Tom Wigg: And a five grand credit card bill?
Stephen Byrd: Exactly. It’s like maybe next time you put a few parameters around that. But long way to say, it’s agentic across every workflow that I can think of that will still result in an explosion in token demand.
Tom Wigg: It’s definitely a good idea to put some parameters around your agentic workflow.
My thanks to Stephen for that conversation. And thank you for listening. Let us know what you think of the show by leaving us a review where you listen. And if you find Thoughts on the Market worthwhile, tell a friend or a colleague about us today.