Carlyle's Thomas on AI's Impact on the Fed Policy, US Economy

Bloomberg

Watch Carlyle's Thomas on AI's Impact on the Fed Policy, US Economy - Bloomberg on Dailymotion

Transcript

00:00You talk about this. It's maybe the wrong side of the equation or not the only side of the equation

00:05to just look at tech stock valuations, but also how it's really affecting how policymakers are

00:10thinking about the economy. Can you explain what that ramification or how that influence really

00:14is percolating through policy? Yeah. When you ask policymakers or people in the administration or

00:20just other advocates of low rates, how is it that you can cut rates by 150, 200 basis points in an

00:27economy where we still have above target inflation? The answer always comes back to AI.

00:33And essentially what they're saying is, sure, the Fed may miss its inflation target for another few

00:38months, but whatever residual price pressures are there are going to soon be overwhelmed by this

00:43massive disinflationary shock from AI. So I think that that's embedded in expectations. And again,

00:50that provides the path to the lower rates. My worry is that this may confuse today and tomorrow.

00:57What I mean by that is what we've learned over the past few years, what's very obvious today,

01:02is that our path to this disinflationary, agentic AI future is paved with trillions of dollars

01:10of investment in what are essentially massive, energy-hungry supercomputers. And that's extremely

01:17resource-intensive. You have the demand for data centers, for infrastructure, for construction,

01:22GPUs, servers, other hardware, and especially, of course, power and generation capacity.

01:29And the last time we had anything like this, in terms of a concentrated CapEx boom of this magnitude,

01:35we had real interest rates that average, real interest rates that average nearly 4% from 1997

01:41to 2000. What people are talking about today is taking rates down further, real rates to effectively

01:48zero. That does not seem at all consistent with an inflation target.

01:53If you do have an inflationary ramification from the AI build-out, what does that imply for

01:59where rates should be, what the neutral rate is for the Federal Reserve?

02:02Yeah. Well, first, it implies that a neutral rate is higher. Any time that you have an increase

02:08in fixed investment relative to savings, we have this shift since the pandemic in investment

02:13savings propensities. The fixed investment boom in manufacturing, it was after the supply chain

02:21crisis, the need for redundancies for additional warehouses. Then, of course, you had a green

02:27energy boom. Now, this has transitioned really almost entirely to AI and the data centers, again,

02:32these supercomputers. So I think that this insistence today that the neutral rate is lower than where

02:39it is today, I don't know that there's that much evidence for that. And I worry that this insistence

02:45that you can move rates down closer to neutral, the neutral rate's unobservable, right? It's just

02:53an estimation. And I think that, again, it seems curious in the context of core PC inflation that's

03:01a 2.9%. So this is a speed question. This is a question of whether we can get the efficiencies

03:05quickly enough to overwhelm some of the investment costs and the potential inflation

03:10to create this equilibrium where there can be this lower inflation. Ken Griffin was talking about how

03:16he's not seeing AI really create so much better in terms of pricing models for bonds. Do you see

03:22the same thing? I mean, is the efficiency not coming as quickly as you think the inflationary boom is

03:26going to come? Well, so again, the boom in inflation, or at least, again, this resource intensity,

03:32that all this demand for construction, for the hardware, for power is today. And again,

03:37we're talking, yes, in the future, I do suspect that there will be a disinflationary shock,

03:42but the timelines matter significantly. Of course, when we're thinking about the return on all this

03:47capital, timelines matter quite a lot. As you see throughout history, whether it was the internet

03:53or electrification or even the railroads, it's not a question of whether the technology is

03:58transformative. It's a question of when the revenues associated with the investment in the

04:02technology actually arrive. And I think recently what you've seen is some of the AI integration

04:08programs in the corporate sector. There has been some, they've fallen short of return on investment

04:14targets. In many cases, I think that that is simply because you have essentially a solution in

04:21search of a problem. You know, you have teams that are, here's the AI tool, figure out what to do with

04:26it. And I think sometimes that just takes time. Sometimes you realize that you have to learn by

04:32doing. And so I think that that is a risk today, that this is just the timelines are being extended.

04:37And that's a very different story than whether this is, whether these gains ultimately don't

04:43materialize. Given this backdrop, if the Fed cuts rates through times, which a lot of people are

04:48expecting, do you expect that long-term rates will rise significantly potentially if we do have an

04:53economy that is not only chugging along, but potentially re-accelerating? I think that right

04:59now the risk for the Fed is to have a repeat of last year. They cut rates by 100 basis points in

05:052024. And you had effectively 100 basis point backup in long-term yields. So I think they want to be

05:13cautious of that. Today, though, I think they also need to look at the foreign exchange value of the

05:18dollar. They need to look at this meteoric rise in the price of gold, other commodities. I think

05:25that a lot of the people talk about a debasement trade. So I think that these concerns about policy

05:30perhaps being too easy, not today, but could be moving in that direction, is actually manifesting

05:36outside of bonds today. It could, of course, there could be a backup in yields. That should be a

05:41concern. But I think that there are some worrying signs outside of the fixed income markets that there's

05:47concern about the direction of policy. How concerned are you about some sort of

05:51rapid-fire reassessment of how quickly some of these tech companies can invest in AI? And I just

05:57wonder what we saw, say, with Oracle a few months back and this sort of tremendous plummeting in

06:03their shares, trading like a penny stock. Do you foresee more of that coming down the pike if right

06:07now AI, for a large extent, is a solution in search of a problem? Well, again, I think that right now

06:14the most fascinating part of all of this, from a market perspective, from at least the way I view

06:20it, is that you have these mega-cap tech companies that have effectively exercised an option to change

06:26strategy. What I mean by that is, really, for the last decade longer, they've been essentially

06:32virtual companies, asset-like businesses. You know, very common to generate $60 billion in cash from

06:37operations, maybe invest only $12 or $15 back in the business. This huge free cash flow to fund

06:44acquisitions, to fund special dividends, to fund share repurchase programs. Now, 70% of the cash

06:52from operations is going into AI. This can continue. I mean, the amount of money that is generated by

07:00these companies means that you have a 15-, 18-month timeline. Oracle went into a negative free cash flow

07:07position. That was partly just this concern that, of course, they have this very large contract with a

07:13company that has only $13 billion in revenue. So, a $13 billion revenue company pledging $300 billion

07:19in spending. Oracle had to increase their CapEx by about 47% annualized rate over two years to build

07:27the infrastructure necessary to make good on those commitments. So, I think there was a reassessment

07:32there. But much, again, of the spending is by companies that have the cash to do it.

07:36Just real quick, are you getting more cautious as you look at the potential for both rates going up

07:40and potentially a reassessment of how quickly AI adoption can take place?

07:45I think that any time you have a market that is as concentrated as it is today, 40% of the market

07:52capitalization of the S&P 500 in 10 stocks, eight of them have their financial future really hitched to

07:59AI. So, you're talking about an AI bet that accounts for about 25% of the total market capitalization in

08:08the U.S. I think that in and of itself is a reason to be cautious, look for other ways to diversify. Of

08:13course, we recommend private markets. But again, I just think in general, having a sense of caution

08:19is very prudent at this juncture.

Category

Transcript

Be the first to comment

Recommended