Apollo's Torsten Slok Talks Big Tech Risk

Summary

Podcast Summary: Bloomberg Talks – Apollo's Torsten Slok Talks Big Tech Risk

Date: July 7, 2026
Guest: Torsten Slok, Chief Economist at Apollo
Host: Bloomberg

---

Episode Overview

This episode features an insightful conversation with Torsten Slok of Apollo, who dives deep into the risks and realities of current AI investments, the impact on corporate profit margins outside the Magnificent Seven (the largest US tech firms), and the broader macroeconomic consequences. The discussion critically examines the gap between high earnings expectations for AI and the actual, observable financial returns, with a particular focus on implications for investors, job markets, and financial stability.

---

Key Discussion Points and Insights

1. Profit Margins and AI Outside Big Tech

• Slok observes that while the Magnificent Seven tech companies show clear profit margin gains thanks to AI, this effect has not materialized for the wider S&P 493 (all other S&P 500 companies).
  • “What really is critical is that now we need to see profit margins go up outside the Magnificent Seven.” – Torsten Slok (01:03)
• This imbalance could pose a valuation risk: expectations for AI-fueled growth may be overestimated outside of major tech.
  • “The mismatch between are we going to see profit margins, earnings growth go up outside this, the Magnificent Seven, is that going to come slower? Is it going some faster? Is absolutely critical for a conversation about what should the value be of the Magnificent Seven.” – Torsten Slok (01:47)

2. AI Return on Investment (ROI) and Company Valuations

• There is widespread investment and demand for AI compute capacity, but the timing of actual returns remains uncertain.
• The possibility that it may take years before non-tech sectors see tangible benefits throws current stock valuations into question.
• The conversation also highlights that some firms are finding it may be more lucrative to sell excess compute capacity rather than using it internally (01:38).

3. Parsing AI's Impact on Jobs in Specific Sectors (Banking, Consulting, Legal)

• Slok points out the lack of consensus on how to quantify “AI exposure” within different jobs and sectors.
  • “The challenge is that there is not an agreement about what does AI exposure mean... for the financials, that means for legal services, that means for consultants, that we're having some challenges figuring out how do we even quantify what is the impact of AI?” – Torsten Slok (02:59)
• He outlines two main types of studies: one focusing on actual measured use of AI, and one using more theoretical models to estimate potential impact. Neither paints a clear picture yet on future employment risk or opportunity.

4. Shifting AI Narratives: Job Loss vs. Job Enhancement

• Recent corporate messaging, including from foundational AI vendors, is retreating from apocalyptic job-loss scenarios. Instead, the narrative is now about AI helping employees do their existing jobs better (04:22).
• A recent RAMP study found companies that spent more on AI also reported higher job growth, challenging the idea that AI only automates away jobs.
  • “They did indeed find that more spending on AI, it actually resulted in more job growth.” – Torsten Slok (04:37)

5. Macro and Market Risks Tied to AI Concentration

• Stock market and credit market performance are now highly concentrated around the “AI trade.”
  • “In equities, the concentration is very high. In AI, you also look at IG issuance, high yield, issuance, even venture capital, the concentration in AI is very, very strong.” – Torsten Slok (05:28)
• Actual economic growth is being lifted by both AI-related capital investment and the wealth effect from higher stock valuations.
  • “If you add up what that contributes to GDP this year is roughly about 0.7% out of a 2% GDP growth this year. If you add about 0.3 coming from the wealth effect because of high stock prices, that's also very important.” – Torsten Slok (05:28)
• Hyperscalers issuing large volumes of investment-grade debt are crowding out traditional demand for US Treasuries, with potential systemic implications.

6. Strategies for Managing the AI Concentration Risk

• Classic portfolio theory dictates that to de-risk from a dominant factor, one must diversify—but it’s not clear what the best alternative is.
  • “If there's one factor you're trying to avoid is to try to pick another factor. But the question is, what is that other factor?... Value investing is indeed one place to hide. But it has to be value investing that's indeed protecting you against the downside risks that come along if AI does not deliver in the end.” – Torsten Slok (06:50)

---

Notable Quotes & Memorable Moments

• “We all know that [AI] is going to make a big difference, but the speed with which this difference comes along... becomes absolutely critical when you again think about that company valuations today are the net present value of the cash flows that these companies get in the future.” – Torsten Slok (03:49)
• “The challenge is page one in your finance textbook.” – Torsten Slok, on diversification and factor risk (06:50)
• “The main lesson in finance is factor investing. And now we’re staring at one factor that’s driving all financial markets.” – Torsten Slok (05:28)

---

Key Timestamps

• 01:03 – Profit margin dynamics in the S&P 493 vs. Magnificent Seven
• 01:38–02:40 – Challenges in measuring AI ROI and impact on valuations
• 02:59–04:22 – Quantifying AI’s impact on employment and the fuzzy concept of “AI exposure”
• 04:22–05:09 – Shift in AI narrative: job loss vs. enhancement, and the RAMP study results
• 05:28–06:48 – Macro and market risks stemming from AI concentration and debt issuance
• 06:50–07:19 – Managing concentration risk, value investing as alternative

---

Conclusion

Torsten Slok’s commentary provides a nuanced, candid examination of the current AI boom: while optimism remains high, the financial and economic returns outside of the biggest tech companies are far from guaranteed, and the risks of sector and market concentration are mounting. For investors and policymakers alike, the coming quarters will demand careful scrutiny of AI’s real-world impact, not just its promise.

Transcript

[A]

So there's a lot of noise about AI, but time's too tight for more promises. So let's talk about results. At IBM, we work with our employees to integrate technology right into the systems they need. Now a global workforce of 300,000 can use AI to fill their HR questions, resolving 94% of common questions, not noise proof of how we can help companies get smarter by putting AI where it actually pays off, deep in the work that moves the business. Let's create smarter business. IBM,

[B]

Bloomberg Audio Studios Podcasts, Radio news

[C]

so here's the latest this morning. Apollo noting a lack of profit margin gains due to AI outside of the tech sector. Thorsten Slok of Apollo writing, there's a mismatch between current earnings expectations and the actual time firms need to generate ROI on AI investments. And it could have significant implications for many AI company valuations. Torsten joins us now for more. Torsten, good morning. Good to see you. Let's build on that quote. What are you tracking right now? What are you saying?

[D]

Well, what's really, really important in this discussion is that of course, profit margins have been phenomenal in the Magnificent Seven. But what really is critical is that now we need to see profit margins go up outside the Magnificent Seven. In other words, what's going on with S&P493 becomes very, very critical because at this point, profit margins S and P493 have just not gone up. So therefore, one very important conclusion and one very important place to look for signs of AI is beginning to have an impact is to look at what's going on in earnings growth, profit margins and overall the health of the S&P493 as a result of the technological improvements we're seeing at the moment.

[C]

Mets are also making a call potentially as well that the ROI on selling excess capacity might be higher than using it internally. Does that reinforce some of this message for you?

[D]

Well, the issue of course, is that there is a huge, of course, build out of capacity and compute and there will literally be unlimited demand for compute. The question is just at what price and who will the bias be and where is that capacity coming from? And the big picture still remains that AI is a very revolutionary technology. Everyone agrees on that. But the key question now is how long time is it going to take before this shows up, especially in profit margins outside the Magnificent Seven? Because if the S&P 493, let's say that it takes several years before profit margins begin to go up, the question is whether the implicit earnings assumptions in the Magnificent Seven are too high or too fast relative to what's actually going to happen. So that mismatch between are we going to see profit margins, earnings growth go up outside this, the Magnificent Seven, is that going to come slower? Is it going some faster? Is absolutely critical for a conversation about what should the value be of the Magnificent Seven.

[B]

Today, in exactly one week's time, as John was reminding us, the big banks begin reporting earnings and we have the kickoff of the earnings season. There's going to be a lot of discussion on AI and what it means for jobs in the banking sector. How do you parse through what the companies say to really understand what it means in terms of whether they're going to cut jobs or not?

[D]

What's really challenging about this is to talk about exposure because there's a lot of different studies already that look at what is the exposure, meaning AI exposure in different occupations. These two, these buckets, these two studies fall into different buckets of one saying what is the actual exposure in terms of actual usage? So this is trying to measure, say, what are people using Claude for what tasks and what request did they get? And therefore measuring and quantifying the actual usage of AI. Another bucket is studies that look at what. Well, let's theoretically assume what is the economist exposure to AI and then try to match that with occupations and figure out what is employment in those sectors. And those studies, of course, are what you call more theoretical. So the challenge is that there is not an agreement about what does AI exposure mean. So that's raising all these questions around, well, what does it even mean when we say that a certain part of the economy is exposed to AI? Because that's just not, at this point in the studies that look at this, a really good way to quantify what exposure means. And that means also for the financials, that means for legal services, that means for consultants that we're having some challenges figuring out how do we even quantify what is the impact of AI? We all know that it's going to make a big difference, but the speed with which this difference comes along, how many people is impacting today, next month, next year? It becomes absolutely critical when you again think about that company valuations today are the net present value of the cash flows that these companies get in the future.

[B]

We're also hearing a changing story, a changing narrative from companies themselves. There's a Wall Street Journal story about how the narrative has shifted, at least from open AI and anthropic, from these doomsday scenarios to a future where workers actually keep their job jobs, but just do it better thanks to what does that tell you?

[D]

Yeah, and RAMP had a really interesting study over the last week that they put out where they basically looked at what has been the cost and the spending on AI among different companies. And what they did that they looked at, well, what was the job growth in those companies that had more spending on AI? And they did indeed find that more spending on AI, it actually resulted in more job growth. So one way of looking at this is one dimension of saying what's the actual spending relative to the more theoretical matching of occupations with what's been going on with job growth in those sectors that also have been spending on toast.

[C]

And to your point for this all together, this is a potential risk, the valuations, which is an obvious market risk. Is it a macro risk as well? In central Portugal, I believe you were there, big conversation about this. We've got a story in the market right now that feels like one trade and increasingly a story in the economy that feels like we're firing on one engine. How much is it one and the other?

[D]

Absolutely. It's actually three different things. First of all, it's of course everywhere in markets because the concentration to the story is so strong. Think about it. For the last 15 years, the main lesson in finance is factor investing. And now we're staring at one factor that's driving all financial markets. In equities, the concentration is very high. In AI, you also look at IG issuance, high yield, issuance, even venture capital, the concentration in AI is very, very strong. So in markets, let's disagree, the air exposure or the factor plays a very, very critical role at the moment when it comes to the economy. Also, if you look at actual spending on data centers and energy, if you add up what that contributes to GDP this year is roughly about 0.7% out of a 2% GDP growth this year. If you add about 0.3 coming from the wealth effect because of high stock prices, that's also very important. And finally, let's also not forget that the hyperscalers are issuing so much IG corporate debt that this is crowding out demand for US Treasuries. Because if you are bond manager and investment grade credit, you could either buy sovereigns, US Treasuries, you could buy financials. That's what you've been doing for a long time. But now you can also buy 700 billion in hyperscalers. So it's not only that it has an impact on markets overall and has an impact on gdp, but it actually also has an impact on demand for Treasury. So yes, the conversation and the panel I was on in Central was exactly around this risk that AI is indeed now a more and more prominent factor. Basically not only the economy, but also in financial markets.

[C]

What's the consensus on how to manage that risk?

[D]

Well, the challenge is page one in your finance textbook. If there's one factor you're trying to avoid is to try to pick another factor. But the question is, what is that other factor? Momentum is a growth. Is it value? Value has some opportunities because it's not growth. So that's why ways of looking at parts of the private markets, public markets, that is Value investing is indeed one place to hide. But it has to be value investing that's indeed protecting you against the downside risks that come along if AI does not deliver in the end.

[C]

Torsten Slok of Apollo Torsten, fantastic note. Great rate in the last week or so. I appreciate it. Tossing stock of Apollo there on AI and tech.

[B]

Get the news you need in just 15 minutes.

[C]

Start your day with Bloomberg Daybreak, the

[D]

podcast with a global view on the stories that matter.

[C]

I'm Nathan Hager.

[B]

And I'm Karen Moscow. Join us each morning for curated stories on current events, politics, business and foreign

[D]

relations, plus one conversation on the day's biggest developments, all in just 15 minutes.

[B]

Subscribe to Bloomberg Daybreak for a precise, thoughtful take on the stories that matter.

[C]

Listen to Bloomberg Daybreak each morning on Apple, Spotify, or anywhere you listen to.