Can the AI Boom Deliver Enough Productivity?

Knowledge at Wharton

Ask about Knowledge at Wharton

Episode Description

ABOUT THE EPISODE The AI industry is spending hundreds of billions on data centers and computing power. The question is whether that investment can produce the extraordinary productivity gains investors are expecting.

Jessica Wachter, Dr. Bruce I. Jacobs Professor in Quantitative Finance at The Wharton School, explains why her research points to a rare productivity boom as the threshold for AI investment to pay off. She discusses the potential for $1 trillion in annual AI investment, the risk if companies fall short, and what the boom could mean for workers, wages, and markets.

ABOUT THE PODCAST Every day, business scholars answer pressing questions in their research - but what do their insights mean for you? In this podcast, Wharton faculty dive into what inspired their studies and how their findings resonate with the world today. Learn how research insights translate into knowledge you can use, with host Dan Loney.

----- The Wharton School of the University of Pennsylvania is committed to sharing its intellectual capital through Knowledge at Wharton, the school’s online business analysis journal.

Launched in 1999, Knowledge at Wharton offers free access to:

- Articles, podcasts, and videos highlighting Wharton faculty research and analysis of current business trends - Interviews with book authors from Wharton School Press - In-depth series of curated content like Wharton Executive Education’s Nano Tools for Leaders - A searchable database of more than 10,000 articles covering all aspects of business - A weekly newsletter that delivers Knowledge at Wharton insights directly to your inbox

Visit the Knowledge at Wharton homepage: https://whr.tn/3qUpSBE Subscribe to the Knowledge at Wharton YouTube channel: https://whr.tn/3DEAgAU

Read the full transcriptHide the full transcript
  1. We've seen large companies go bankrupt in the United States, one of our strengths.
  2. So I absolutely think there's a risk of bankruptcy, but people take risks like this.
  3. Welcome to The Ripple Effect, the podcast that takes you on a journey through the minds
  4. of Wharton faculty. I'm your host, Dan Loney, and in each episode, we'll be diving deep into the inspiration
  5. behind the groundbreaking research that Wharton professors have conducted and exploring how their findings resonate with
  6. the world today. One of the bigger questions in and around the AI boom is just how much productivity
  7. will be impacted. But maybe a bigger question that comes from our guest today is just how much does
  8. productivity have to increase in order for the AI boom to be successful and not a
  9. bust? Pleasure to be joined by Jessica Wachter, who is a professor of finance here at the
  10. Wharton School. Jessica, great to have you with us today. Thanks for a couple of moments.
  11. Thanks for having me. I'm very happy to be here. Thank you. And I guess we can start out with
  12. that question of the topic of productivity and just how important a variant it is to
  13. the success of AI as we move through the next several years. Right. So these companies need to see a return
  14. on their investment, right? The investment is funded by equity dollars or by debt dollars.
  15. Either way, there needs to be a return. Otherwise, that impacts the value of their company.
  16. So that return has to come in the form of higher earnings. And that's another way of saying that is
  17. higher productivity. So what my co-author and I do in this paper is we look at the
  18. dollars of investment and we assume that the companies are investing up to the point where
  19. there is zero economic rent. So they're adding capital up to that point. And that allows us to back out a
  20. implied productivity number. And we find that that productivity number is very high.
  21. It's quite substantial. It's a multiple 2.7. So it's a what we call a rare productivity boom.
  22. And it's rare in the sense that you might have one, you know, small company make
  23. a discovery here and there. But in this case, these are some of our largest companies in our economy all at
  24. once being subject to this boom. So that's why we refer to it as a boom in the AI sector.
  25. And over what time frame are you looking at when you're talking about reaching these levels
  26. of productivity? So the near future. Now that doesn't say that disaster will occur
  27. if we don't get there in the next few years. But the assumption of the model is that
  28. it's basically by 2028. Does this then to a degree explain why we're seeing such incredible high levels of investment
  29. by companies on infrastructure, on server farms, on all of these different components of AI is
  30. that there is somewhat of a time window here that they want to make sure that they reach?
  31. To me, that is the most compelling explanation is that there's a short time window where
  32. if you take advantage of it, you can grab these productivity gains. As I said, it's kind of amazing just
  33. to see how much that is being invested by these companies. We're already seeing tens of billions of dollars
  34. by some. Are we talking about at some point we're going to be reaching hundreds of billions of
  35. dollars of investment by these companies? Well, I'm going to pull up the numbers
  36. in my paper. So, you know, we look at this in two different stages. So we look at essentially what's occurred in
  37. the sense of what happened in 2025, what's projected, what's forecasted for 2026 and estimates for
  38. 2027. So the current estimate for 2027 is about a trillion dollars, roughly. So that's pretty substantial.
  39. For 2026, it's rounding 800 billion. How much of a concern then does there have to be around potential bankruptcy because of
  40. all of this high level of investment? I mean, it's absolutely a concern because if
  41. companies turn out to be wrong, the risks of bankruptcy increase. And that's simply because to maintain these levels
  42. of capital costs money. There's interest payments, for example. By the way, if these were wasted money,
  43. then the stock price is going to fall. And meeting other meeting their ongoing commitments is
  44. going to be difficult. We've seen large companies go bankrupt in the United States, one of our strengths.
  45. So I absolutely think there's a risk of bankruptcy. But people take risks like this.
  46. What are the what are the growth rates that they need to see then over the next how many ever quarters to reach the
  47. the levels that they need to? And and what levels do the companies really need to see in order to make this
  48. viable longer term? Yeah, so that 2.7 in growth and productivity that so that that translates into a
  49. growth in top line earnings. Now, one of the things we write about in the paper is whether that how much
  50. that get translated into increased wages for people who work there versus how much big quota
  51. shareholders is something that is not within our model is not something that you can answer.
  52. You need a somewhat richer model. But we basically need to see that kind of growth with like a multiple of 2
  53. .7 in the near future. Yes. One of the questions I've had, and I don't know if it's directly connected with this
  54. research, but one of the things that is talked about when you're trying to reach a
  55. higher level of productivity is how this is going to impact the firm and their level
  56. of investment around employees. Is there a component that of part of this research?
  57. And again, I'm not sure if it's directly linked to what you have done so far that the employment rate will be impacted by
  58. this as well as we move forward here because AI is expected to be much more productive than the human than the human being.
  59. So, you know, clearly there's a large debate about this. But within our model, this is actually only
  60. good for workers, that it just raises people's wages. That's that's the effect in our model.
  61. It doesn't displace people. You can view the productivity either as something that augments labor or as additional workers in
  62. the form of AI. Those are mathematically equivalent. But either way, it actually it raises wages
  63. or it doesn't raise wages, but it does not lead to unemployment. Right. To a degree, it kind of secures the
  64. employment in place, then. It does, actually. Yes. OK, so for at least right now, does
  65. the investment match the necessary level of productivity? Are our companies on the right path right
  66. now? Well, so that's that's impossible for me to say. So we are backing out what productivity has
  67. to be for this investment to be rational or profit maximizing on the part of the
  68. companies. We so something has to be given there. And so we're taking as a data point
  69. the investment and we're backing out the productivity. So so I don't know. I figure they know better than I do.
  70. That's my answer. From your research, then, that you've done so far, is there a natural next step that
  71. you would like to take this research? And I guess to a degree, it may be even open ended because this is such
  72. a a process that's in development right now. I mean, I'm just watching this really eagerly,
  73. like I'm, you know, I'm on the edge of my seat for the next set of quarterly earnings.
  74. So this will be something that will pop up on your radar every three months, depending
  75. on how the company's numbers are looking as we move forward. I'm watching for it.
  76. That's exactly right, not not just not just for the earnings, but also for how the
  77. further expenditure is projected. So I'll say these earnings, the information that
  78. we have is from the previous earnings cycle. And so far, sell side analysts have not
  79. made their their estimates further, like they've not lowered their estimates. So everything that I see, this is now
  80. for the last three months seems pretty stable. Just have to see going forward. And the companies that you're really focused on
  81. are the ones that we're hearing that are making these big investments. They're setting up these infrastructure components.
  82. They're they're doing all of these pieces. These are the ones that are that are of the greatest focus right now.
  83. Yeah. So I think that there's various parts of this fascinating ecosystem. The ones that we focus on are the
  84. ones that are actually making the physical investments as opposed to their customers who are the
  85. ones who provide the compute. That's because we don't want to double count. So we just want to kind of count
  86. each dollar in a data center once. Does then it potentially multiply or slide over
  87. to those customers at some point because they obviously are trying to reap the the benefits
  88. of the investment that these companies are making here right now? Yeah. So if we think about a dollar of
  89. revenue, you know, we all hear about the revenue numbers, for example, of Anthropic, not to
  90. pick on anyone, anybody in particular. Right. So that's their revenue then goes to pay
  91. these guys. Yeah. And so that's that's kind of the the flow of the money. This is, as you said, an ongoing process.
  92. So congratulations on your work and thank you very much for your time. And we will circle back with you because
  93. this is obviously, as we said, a story that continues to develop. Thank you, Jessica.
  94. It's been a pleasure. Thank you. Thank you, Jessica Wachter, professor of finance here
  95. at the Wharton School. Thank you for listening to The Ripple Effect. We hope you found this episode informative and
  96. engaging. Don't forget to subscribe and leave us a review so that we can continue to bring
  97. you the best insight from the Wharton School.

Brief Summary

Jessica Wachter discusses AI's impact on productivity, investment risks, and employment in this episode of The Ripple Effect.

Key points

  • The Ripple Effect Podcast. Join host Dan Loney as he explores groundbreaking research from Wharton faculty.
  • AI's Productivity Impact. A discussion on how much productivity needs to increase for the AI boom to succeed.
  • Investment vs. Bankruptcy Risk. Exploring the risks of bankruptcy due to high levels of investment in AI.

Tension

45/ 100Some tension

Related episodes