Torsten Slok’s Mid-Year Outlook: Three Forces Driving Markets

Q&A with Torsten Slok, Chief Economist at Apollo

As investors enter the second half of 2026, Apollo Chief Economist Torsten Slok breaks down why the AI boom has evolved far beyond a technology story, how an industrial renaissance is reshaping manufacturing and investment opportunities, and how government policy is providing additional support for growth. He also discusses the implications for inflation, interest rates, asset allocation, and credit markets while outlining the key indicators he’s watching for the months ahead.

Q: What is the macro picture telling you right now?

Torsten Slok: What has been surprising more recently is that, despite some of the headwinds coming from the Middle East and oil prices moving higher, we continue to get pretty good data. The employment report for May was strong. Consumer spending has continued to be strong. Business spending has also been very strong.

Overall, the economy continues to be quite resilient and is actually doing quite well.

Q: What are the main drivers for the rest of 2026?

Torsten Slok: There are three very important reasons why things are still holding up quite well. Number one, we are having an AI boom. The data center buildout is helpful for growth, and the energy buildout associated with that is also very helpful. Number two, we have the industrial renaissance, which is another tailwind for growth. And third, we have the One Big Beautiful Bill, which is also supporting growth.

These are quite unique factors. The AI boom is unique, and the One Big Beautiful Bill was, of course, a political decision to support the economy in 2026. Together, these tailwinds are providing quite significant fuel to growth, and that is why we are quite optimistic for the rest of the year.

Q: Why is AI now much bigger than a stock market story?

Torsten Slok: AI is a truly revolutionary technology that is making a significant difference in all our lives, almost no matter who you are or where in the world you are. We are seeing significant improvements from large language models. We are seeing agents and agentic AI take off in a very meaningful way.

As a result, a lot of compute is needed. When compute is needed, more data centers are needed. And when data centers are needed, energy for those data centers is needed.

So there is a very significant tailwind for the economy because so much computing power is needed by businesses and households. That is the key source of growth: a data center buildout and an associated energy buildout unlike anything we have seen before.

Q: What indicators will you be watching to see whether the AI cycle remains intact?

Torsten Slok: A lot of the financing for data centers has already been put in place over the last six to nine months. So in that sense, a lot of the growth in the data center buildout will continue almost no matter what in 2026.

Looking ahead, other indicators become important. Ultimately, it comes down to what demand for compute is going to look like. Will there be bottlenecks in terms of the energy associated with that compute? Can we build enough energy and enough data centers compared to the demand we are seeing?

There are many dimensions around both the supply of compute and data centers, and the demand side. More recently, the debate has also been about token demand. More companies are beginning to organize their token demand differently because tokens have become more expensive.

At the end of the day, it is a discussion around three things: what demand for compute will look like, what the price of compute will look like, and what the supply of compute will look like.

For you and me and everyone listening, we will certainly have large language models for the rest of our lives. This is a truly revolutionary technology.

The question is not whether this ends with a growth rate of 3.4 or 3.7. It is much bigger than that. It is going to have a very significant impact on how we work, how we live, and the broader economy.

Q: What is Jevons Paradox, and why does it matter for AI and employment?

Torsten Slok: When people talk about AI, they normally talk about whether it may replace workers or replace labor. People worry that some task they do can now be done by a computer or by AI, and therefore maybe they are no longer needed.

But the important part of the discussion is not only the individual task that may be replaced. It is also the bigger picture of what AI can do for the economy.

AI has created a significant increase in the number of new businesses being created. It has also created a significant increase in the number of businesses created by solo founders, meaning one person creating a new business. That effect—creating more businesses and more jobs—is more than dominating the negative effect of those who lose their jobs.

Yes, maybe some people in call centers will lose their jobs. But at the same time, we will probably get a lot more call centers created. We saw something similar with radiologists. Ten years ago, it was predicted that MRI and CAT scans would make it easier to do scans, and maybe radiologists would not be needed. Instead, we now have a lot more scans, and as a result, we have a lot more radiologists.

That is the core of Jevons Paradox. When a technology gets cheaper and more effective, it creates much more demand for that technology. In consulting, financial services, and legal services, we will probably see more businesses created because AI makes it easier to do things that previously required more resources.

That is why the latest employment report was very good, and we are still not seeing all these job fears play out. I continue to believe AI will be positive for employment because people are ignoring the fact that this is not only about displacing individual jobs. It is about the much bigger opportunity set that comes with AI technology.

Q: So you are generally bullish on AI’s impact on economies and jobs?

Torsten Slok: I think there will be distributional consequences. Those who have a job that is impacted by AI will of course feel that it “took my job” or replaced some parts of the work or tasks they used to do.

But on the other hand, those entrepreneurs who are now inventing new businesses will see this as a huge opportunity. There will be winners and losers. Some people, though not many in my view, might ultimately lose their jobs because of AI.

But what comes along instead is a significant boom. It creates a whole new opportunity for you and me and everyone else to say, now I can do a business plan using a large language model. I can use agents to do some work for me.

That effect will be much more positive. Some people worry more about job displacement, but we should be focusing on all the opportunities that come from business creation as a result of the AI tools that are now available.

Q: Shifting to another driver you mentioned, what do you mean by an industrial renaissance?

Torsten Slok: Over the last 20 years, we first had globalization, especially after China entered the WTO in 2000. A lot of manufacturing capacity was moved out of the U.S. to China and the rest of the world.

Over the last five years or so, we have seen a significant reversal of those globalization trends. In the U.S., that began with the CHIPS Act under President Biden in 2022, when there was a strong political willingness to bring home production of semiconductors and manufacturing capacity. That created a dramatic boom in manufacturing capacity, specifically in semiconductors.

That was the first prime example of industrial policy: bringing back manufacturing production in a sector we had not had domestically for a long time.

Now we also have home-shoring coming from pharmaceuticals, defense, and other strategic sectors that politicians, both Republicans and Democrats, believe are important to bring back to the U.S. Financing that manufacturing capacity is very important for financial markets. These are long-duration assets that can be matched with long-duration liabilities.

So the industrial renaissance is essentially a reversal of globalization, with sectors like defense, prescription drugs, semiconductors, and other strategically important areas being rebuilt domestically.

This is not only happening in the U.S. It is also happening in Europe. Politicians are saying that if Europe cannot rely on energy from Russia or the Middle East, it needs to produce more energy, including renewable energy, at home. It is all about creating more production capacity, and that requires financing.

Q: You've talked about AI as a picks and shovels opportunity, referencing how the gold rush wasn't about the gold, but about the picks and the shovels. In that context, how do AI and the industrial renaissance connect?

Torsten Slok: Building the infrastructure, both in hardware and services, is an important part of the AI buildout. That is the “picks and shovels” comparison to the gold rush.

There is a need to build computing power. In plain English, once we have that computing power, we can debate who is using it and at what price. But there can be no doubt that we are all using OpenAI models, Anthropic models, and other tools on our phones, and those models require compute.

There can also be no doubt that demand from consumers and companies will continue to grow. As a result, there will be huge demand for data centers and compute for the next five, 10, and 20 years.

The infrastructure and services that provide the basic highway for AI are likely to be among the more interesting areas from an investing perspective.

Q: Beyond AI, where else is the industrial renaissance showing up?

Torsten Slok: The prime areas are sectors where politicians have made strategic decisions about what needs to be produced domestically. Defense is one. Energy is another, especially renewable energy, which is becoming more important, including in Europe.

We are also seeing more focus on prescription drugs and pharmaceuticals. Broadly speaking, these are areas where people are politically beginning to say, maybe we should produce this at home rather than abroad.

From an investment perspective, those are the sectors that generally have the most tailwind at the moment.

Q: Do investors fully appreciate the scale of the industrial renaissance?

Torsten Slok: The challenge is that the industrial renaissance is a longer-term theme. It is not something you measure this quarter relative to last quarter.

A lot of it rests on what politicians will do going forward because, in the broadest sense, this is truly industrial policy, which we had not had for a long time. For many decades, globalization said countries should not have industrial policy. We should just get goods where they are cheapest.

Now it has become more acceptable for politicians to say we should not simply get things where they are cheapest. We should get them where we are sure we can actually get them, especially if there is a conflict with other countries.

Countries are now willing to produce things domestically even if it is not the most economically efficient idea. They are willing to do that because if they cannot get defense, prescription drugs, AI, or compute when they need them, they will have a serious problem.

That is why politicians are willing to pay a higher price for some manufactured goods or services, as long as they are produced domestically and available when needed, particularly if geopolitical conflicts arise.

Q: How is government policy supporting U.S. growth right now?

Torsten Slok: The One Big Beautiful Bill had two very important provisions.

Number one, it lowered taxes for households. Tax refunds last year were around $3,000. Because taxes were lowered retroactively, households filing taxes this year, including those filing over the summer with extensions, will experience an average tax refund of around $4,000.

That is more than $1,000 per household as a result of the One Big Beautiful Bill. With more than 130 million households in the U.S., that means more than $100 billion, potentially around $130 billion, in extra consumption over the next two or three quarters. That will be very helpful, and we are already seeing consumer spending remain relatively strong.

The second provision is also helpful for the economy: companies can now do 100% immediate expensing. If you and I were to build a factory across the street, we could write that down in our taxes 100% in 2026. Normally, if we build a company or factory, we would write it down over five to 10 years. Now the bill allows us to do it immediately.

That incentivizes companies to build factories and invest, which is good for GDP growth. It also fits with the other two forces because it helps the data center buildout and the industrial renaissance.

The bill was intended to boost the economy in 2026, not only on the consumer side but also on the CapEx side. The Congressional Budget Office estimates these forces together will lift GDP growth this year alone by 0.9 percentage points. If AI spending lifts GDP by 1%, the industrial renaissance adds 0.3%, and the One Big Beautiful Bill adds 0.9%, that gets us well above 2%, which is the long-run growth rate for the U.S. economy.

That is why we are beginning to worry less about stagflation, which might normally be the response to conflict in the Middle East, and more about the risk of overheating, meaning inflation could begin to move up as we look into the rest of the year.

Q: What does all of this mean for inflation and interest rates?

Torsten Slok: The outlook for inflation is driven by three forces. The economy is strong. We still have upward pressure on inflation delayed from tariffs. And we have upward pressure on inflation coming from energy prices.

That is why markets are debating whether the Fed should hike or not hike, and what it will do later this year. Remember, the Fed’s target is 2% inflation. Today, inflation is not 2%. It is closer to 4%.

The 12 voting members of the FOMC have continued to signal that they are worried inflation is not back in the bottle the way they would like. That is why the Fed has continued to communicate that it cannot quite cut interest rates, because inflation is still a problem.

If inflation is higher for longer, then interest rates will be higher for longer. That means yield levels in credit will also be higher for longer.

That creates both questions and opportunities. If we have a higher interest-rate environment, what sectors of credit markets should investors be looking at? Software, in particular, is vulnerable because the sector has a lot of debt and very low coverage ratios, meaning high vulnerability to higher interest rates.

It also raises the question of how much interest rates could go up. If they go up a lot, that becomes an issue for the economy. We are worried about upside pressure on interest rates, particularly because inflation is staying higher for longer. As a result, interest rates will also have to stay higher for longer.

Q: What are the implications for asset allocation?

Torsten Slok: The first conclusion is that if the Fed says it wants to keep interest rates high, or even raise them in another scenario, investors should view that as the Fed telling us to put more money into fixed income. Put more money into things that cut coupons and give you cash flow.

The Fed is making it more attractive to put money into fixed income with the simple view that it does not want investors to take as much risk. So when interest rates are high and at risk of moving higher, it is very important to focus on fixed income and coupon income. That means products such as credit and rates that provide returns in a higher-yield environment, rather than taking more risk in assets such as equities or venture capital.

The second conclusion is that when interest rates are higher for longer, it becomes important to look under the hood in credit and ask which sectors are more vulnerable. Software stands out because it has high leverage and low coverage ratios. A coverage ratio measures earnings divided by debt-servicing costs. If debt-servicing costs are high, the coverage ratio is low.

So software is vulnerable not only because of AI disruption, but also because interest rates are staying higher for longer.

The third investment implication is in equities, including private equity. If interest rates are higher for longer, it becomes important to invest in companies with earnings. I know that sounds a little funny, but if debt-servicing costs are higher for longer, companies need earnings to service those costs.

That generally means large-cap companies and companies with more diversified business lines. In both private and public equity, investors need to focus on companies that can pay the higher debt-servicing costs. If you are invested in companies without earnings, there will be a problem when debt-servicing costs go up.

So in summary, when the economic outlook is good and interest rates are going up, there are three interesting things to do as an investor. First, make sure you are in fixed income because the Fed wants you to be in fixed income to get a higher yield. Second, make sure you are in sectors of fixed income and credit that can survive when yield levels are higher. Third, do the same thing in equities: focus on sectors and companies that are able to survive when debt-servicing costs are higher.

Q: What would need to change for your outlook to materially change?

Torsten Slok: If we think about the three engines of growth, the AI boom will for sure still be here in six months. It is indisputable that we need more data centers, more compute, and more energy associated with that. So the AI boom will most likely still be a very strong engine of growth.

The industrial renaissance is, as we discussed, a weaker engine of growth, but I still think it is very clearly a theme for the next many years. Politicians want to home-shore strategic manufacturing sectors.

The One Big Beautiful Bill, on the other hand, will begin to run out of steam because it was designed to boost growth around the middle of 2026. That engine of growth could become weaker.

So when we sit down again in six months, the question will be about the duration of these engines of growth. Can they continue to support growth going into 2027? My best guess at this stage is that we will still have strong growth when we talk about the outlook for 2027.

Q: What risks might investors be underestimating?

Torsten Slok: The homework for all of us in investing is to spend as much time as we can understanding AI and how it is impacting us across many different dimensions.

It is not only about the data center buildout. It is also about whether valuations are right at the moment. Are they too high? Are they too low? Will the picks-and-shovels theme continue to do well, which in my view is the most likely scenario? Are there other impacts from AI where we may see bubble features? Are there areas where we may have seen underinvestment?

The opportunity set around AI is enormous. That is why the homework around finding those opportunities becomes absolutely critical.

Note: The responses above have been edited for clarity and concision and do not represent a verbatim transcript of the podcast.

Podcast recorded on June 8, 2026.