Thoughts on the Market

Can the AI Spending Boom Pay Off?

September 9, 2026

Can the AI Spending Boom Pay Off?

September 9, 2026

Big Tech is pouring more than $1.4 trillion into AI, prompting investors to ask: Is it worth it? Our U.S. Internet analyst Brian Nowak looks at three business models that could earn 25% to 50% returns for Gen-AI-enabled technologies.

Morgan Stanley Thoughts on the Market Podcast

Transcript

Brian Nowak: Welcome to Thoughts on the Market. I'm Brian Nowak, Morgan Stanley's U.S. Internet analyst.

 

Today, can the enormous investment behind Gen AI actually pay off?

 

It's Wednesday, September 9th, at 9am in New York.

 

AI has moved quickly into everyday life. It helps people write software, research purchases, automate work, find information, among myriads of other use cases.

 

But we need an infrastructure build-out of extraordinary scale to support all of this activity and more activity to come.

 

In all, we estimate that the major cloud providers are going to spend more than $1.4 trillion on this AI build-out next year alone. But compute capacity is potentially going to quadruple from 2025 to 2028, reaching roughly 120 gigawatts.

 

But all of the spending has raised a lot of questions for investors. One of the most common questions is: What kind of return on invested capital can these companies earn from all of these trillions of dollars of data center infrastructure investment?

 

Well, our bottom-up work points to encouraging answers to this question.

 

We see paths to roughly 25 to 50 percent return on invested capital, or ROIC, across three emerging AI business models. Now, ROIC is a useful way of measuring whether investments pay off. Think of it as how much after-tax operating profit can be generated relative to the capital required in the first place.

 

The first business model we've analyzed is renting compute power. This is the infrastructure layer of the AI economy. Cloud providers build data centers filled with advanced graphics processing units, or GPUs, and rent that compute capacity to customers. In our base case, a large next-generation data center can generate a return on invested capital of roughly 30 percent simply renting AI compute power.

 

And even if rental prices move around, our scenarios still produce returns ranging from low 20s percent to nearly 40 percent. So, despite the enormous cost of building and capital being deployed for these facilities, we think the economics here are quite attractive.

 

The second business model we've analyzed is  where an AI lab has their own model, and they also own their own infrastructure. They give access to their model through an API to consumers and enterprises who then build upon it, they utilize the model. In some cases, they build applications using that model that can be future sources of productivity or efficiency for the economy.

 

In this scenario, we think the economics can be even stronger. When the model developer owns their own underlying infrastructure, our base case generates a roughly 75 percent incremental operating margin and a return on invested capital of 40 percent plus.

 

These returns on invested capital are impressive, but what determines whether these returns can actually materialize?

 

Well, two things matter a lot. The first is the price the developers are able to charge for tokens, which are the units of information that AI models process. The second factor that matters considerably is how efficient[ly] can this infrastructure process these tokens.

 

This is why continued improvements in chips and software to drive higher token throughput or more tokens per GPU per second are critical to the long-term unit economics across this AI ecosystem.

The third model we've analyzed is when the AI developers rent their compute infrastructure rather than owning it. So, effectively, they are paying someone else for the data centers and the GPUs that they need. While this lowers their returns on invested capital because another provider takes a piece of the unit economics, our base case still produces roughly a 30 percent incremental operating margin and 25 percent post-tax return potential.

 

So, while the AI build-out requires enormous investment, the size of the spending alone doesn't tell the whole story about whether or not there are economic returns to come.

 

What ultimately matters is the revenue and profit that the infrastructure can generate. And as more of the infrastructure shifts from training AI models to serving customers through emerging products and inference, we think we're going to get a much clearer answer to this question investors are asking today.

 

Was all this spending worth it? Our research suggests: Yes.

 

Thanks for listening. If you enjoy the show, please leave a review wherever you listen and share Thoughts on the Market with a friend or colleague today.

 

Hosted By
  • Brian Nowak

Thoughts on the Market

Listen to our financial podcast, featuring perspectives from leaders within Morgan Stanley and their perspectives on the forces shaping markets today.

Up Next

Investors have plenty to digest this month, from economic data to central-bank decisions. Our Glob...

Transcript

Andrew Sheets: Welcome to Thoughts on the Market. I'm Andrew Sheets, Global Head of Fixed Income Research at Morgan Stanley.

 

Today, several catalysts for more volatility later this month.

 

It's Friday, September 4th at 2pm in London.

 

Over more than a century of market history, Septembers have tended to see more volatility than the average month. You can't exactly set your watch by it, but the trend is definitely there. As investors come back from summer and capital market activity restarts in earnest, things historically tend to move.

 

This idea seems especially relevant this year. Despite the headlines, it was a pretty calm summer for markets. Since early June, U.S. stocks, yields, and credit were all modestly higher, and they got there with minimal movement. The realized volatility – that is how much these markets are moving on a daily basis – has been historically low.

 

September offers a number of catalysts that could test that.

 

First and foremost is the Fed. Inflation remains above the central bank's target, and markets are pricing a roughly 50-50 chance of a rate hike at the September 16th meeting. That's more uncertainty this close to a meeting than we've had in a while – and the impact goes far beyond a single decision. Live meetings from the Bank of Japan and the European Central Bank also loom in September.

 

September is also a month that historically sees unusually heavy capital market activity. That makes sense. If you're a corporate and looking to raise money, it's often better to wait until investors are back from the summer before going out looking for those funds.

 

But this September could be unusually active, given a growing IPO pipeline and continued funding needs from AI-related construction. And so, it's fair to say that even adjusting for September's usually heavy pace, there's an unusually wide range of outcomes around where capital market activity could land this month.

 

Investors are also coming back from the summer with major uncertainty still hanging over global energy markets. Morgan Stanley's commodity team still sees global energy flows as severely restricted and recently raised their forecast for oil prices, seeing them reach about $100 a barrel in the fourth quarter of this year.

 

The price of what's in that barrel is becoming even more extreme, with the price of diesel fuel in Europe up 140 percent since January 1st. And so, as inventories continue to draw down and questions around the duration of this conflict persist, both factors could drive more market movements.

 

The good news is that while Septembers have historically been more volatile months, they're not necessarily a bellwether. And that could apply again. By month-end, we should have a much better idea of the Fed's path, the scale of capital market activity, and the state of energy supply.

 

But until then, the level of expected volatility across many markets, particularly interest rate and foreign exchange markets, remains unusually low. Given this backdrop, we think those levels of expected volatility can rise.

 

Thank you, as always, for your time. If you find Thoughts on the Market useful, let us know by leaving a review wherever you listen. And also tell a friend or colleague about us today.

Morgan Stanley Thoughts on the Market Podcast
Our Global Commodities Strategist Martijn Rats explains how tightening supply and shrinking buffer...

Transcript

Welcome to Thoughts on the Market. I’m Martijn Rats, Morgan Stanley’s Global Commodities Strategist.

 

Today: why the oil market is tightening, and why we now see Brent reaching $100 per barrel later this year.

 

It’s Thursday, September 3rd, at 3pm in London.

 

It has been an extraordinary summer for oil. Brent — the global benchmark price for crude oil and the reference point for most of the world's oil trade — traded above $110 per barrel in mid-May, fell to $71 by early June, climbed back above $100 three weeks later, and then dropped again to $79. More recently, prices have moved higher again.

 

But the question now is whether that is another temporary swing. Or whether there is a sign that the underlying market has changed.

 

We think it has changed. Supply is tightening, inventories are falling, and some of the buffers that helped absorb earlier disruptions are fading.

 

The clearest evidence is in inventories. Crude oil sitting on the water fell from nearly 1.3 billion barrels in mid-July to 1.1 billion barrels recently. That was a decline of about 190 million barrels. During one four-week stretch, oil-on-water fell at the unusually high rate of 5.3 million barrels a day, the fastest four-week decline since this data series began about eight years ago.  

 

Usually, when there is such a large amount of crude oil that is brought on land, it drives up onshore oil inventories. However, not on this occasion. On a global basis, onshore crude oil inventories have fallen by another 38 million barrels over the same period. That means that those offshore barrels arriving were being used straight away rather than put into land-based storage.

 

The biggest supply issue is still the Middle East. Crude flows from the Strait of Hormuz briefly recovered to around 15 million barrels a day after the June Memorandum of Understanding. That was close to pre-conflict levels. More recently, they have been running around about 7 million. Now, Red Sea exports have also fallen sharply, from roughly 4 - 4.5 million barrels a day in March and April to around about 1.5 million barrels a day at the moment.  Therefore, total regional exports are still up from the lows in March and April, but they are sharply down from that late June peak.

 

Another source of support is fading: strategic petroleum reserves. These are government-held oil stocks that can be released during a disruption. Globally, those releases added around 2.5 million barrels a day to supply in March and April. But that has fallen sharply, and we do not anticipate material further releases from global SPRsafter September.

 

Then China is important, too. Its seaborne crude imports are normally around 10 to 11 million barrels a day, but briefly fell as low as 5 million barrels a day leaving more oil available elsewhere.  Now, China's buying activity still appears low, but at a minimum it has stabilized, and there are tentative signs of an increase. If Chinese imports have stopped falling and possibly go into reverse, they can no longer free up additional barrels for buyers elsewhere, making the global oil market tighter.

 

So why hasn’t crude become even more constrained?  It's because of refineries. Global refinery outages are running 5 - 6 million barrels a day above normal. Although supply of crude oil is constrained, this means that demand for crude is also reduced.

 

Now, the result of that is that the tightness in the system has instead shown up in refined products rather than in crude. And diesel is the clearest example of this, and the one most likely to be felt throughout the economy, since diesel prices feed straight through into trucking, freight, farming costs, and many other areas.

 

The front-month diesel benchmark in the U.S. was recently around $195 a barrel, versus Brent at $95. The difference between the value of a refined product and the crude used to make it is called a crack spread. For diesel, that crack spread reached around $100 per barrel, an all-time high. Over time, that gives refiners a very strong incentive to bring back capacity where they can. If they do, crude demand should rise, whilst inventories are already falling and Middle East supply so far remains constrained.

 

We now expect a full recovery in Middle East supply to take well into 2027. On that path, oil inventories should keep falling throughout the fourth quarter of this year as well as the first quarter of next year. We now forecast Brent to average $100 per barrel in the fourth quarter.

 

Now, for much of this year, the oil market had several shock absorbers, strategic reserves, abundant barrels at sea, and unusually weak Chinese imports all helped. Those cushions are thinner now. That leaves less room for another disruption, just as the road back to normal supply is getting longer.


Thanks for listening. If you enjoy the show, please leave us a review wherever you listen and share Thoughts on the Market with a friend or colleague today.

 

Morgan Stanley Thoughts on the Market Podcast

More Insights