The Wooden Nickel is a collection of roughly a handful of recent topics that have caught our attention. Here you’ll find current, open-ended thoughts. We wish to use this piece as a way to think out loud in public rather than make formal proclamations or projections.
1. What if the AI Bubble Just Started Last Month?
Talk of an AI-driven market bubble has gone mainstream. Daily market coverage now reaches for late 1990s analogies; here’s a headline from MarketWatch from just last week.
The cursory parallels are obvious to anyone paying attention: a hot new technology that catalyzes dreams of a productivity boom, huge capital expenditures that precede utilization and debated ROI analyses, market concentration, hazy financing patterns like related party transactions and off-balance sheet mechanisms, and more.
But widely anticipated inflection points rarely arrive on cue. The recession that wasn’t in late 2022/2023 is a recent and memorable example. Instead, growth decelerated but avoided contraction as forecasters progressively marked down recession odds. The lesson isn’t that risk vanished; it’s that consensus obsessions often miss the turn.
With regard to the AI build-out, there are some important structural differences. For one, valuations are not as stretched (Figure 1). Elevated? Yes, without question. But that has been the case for quite some time, and while a prudent predictor of long-term returns (think 10 years +), the ability of multiples to forecast one-year returns is nonexistent statistically.
Figure 1: Multiple Comparison between 2000 and Today, Source: Evercore ISI
Second, the ’99-2000 bust came at the end of an easing cycle, whereas expectations for monetary policy today put us at the start of a cutting cycle (whether the full cutting cycle comes to fruition is another matter). Third, the core players in the market building AI and financing the expansion are the most profitable companies in the market and some of the most profitable firms in history. Consider the contrast with the dot-com boom. At the peak of the dot-com investment cycle, companies were spending all of their operating cash flow (and more!) into building out backhaul fiber to bring connectivity everywhere; some estimates even had them spending two dollars of capex for every dollar of operating cash flow. Today, the Mag7 is well below such levels.
Figure 2: Investments Are Being Financed with Cash, Source: Apollo
The absence of debt is the most consequential difference. Classic bubbles tend to be credit stories as much as equity stories. Fiber builders leveraged themselves like crazy; Level 3 Communications didn’t even generate positive operating cash flow, and yet still spent billions on capital.
Figure 3: Level 3 Communications Leveraged Themselves To Build
It is the credit picture that seems to be pivoting as Oracle has thrown down the gauntlet. Their most recent quarter has shown that Oracle is pushing all of its chips in and is willing to lever itself to do it, all in the name of OpenAI and the proliferation of AI. The company has enthusiastically and quickly burned all its free cash flow to finance data center construction and the necessary systems and hardware to furnish the largest capital investment cycle in human history. Shareholders won’t see any free cash flow for years; if anything, the duration and magnitude of the capital expenditure cycle coming from the company are being understated.
Figure 4: Oracle Throws Down the Gauntlet
The odds of Oracle being the lone player to go to such extremes seems low. I don’t think it’s coincidental that after this report, a massive reorganization of AI research and talent, Zuckerberg stated just a couple of weeks ago that Meta “misspending a couple of hundred billion in the U.S. would be unfortunate, but the risk is higher on the other side.” Or that Google’s founders have said internally they are “willing to go bankrupt rather than lose this race.” Or that Microsoft has resurfaced in the infrastructure market.
If you believe you are building a digital god, then the NPV is always above zero, regardless of the discount rate.
2. What if the Recession is Here?
A common maxim among economists and eulogized by financial market participants is that the labor market and its many measures is a lagging indicator of the economy’s health; that at the time when it starts to roll over, the economy has already begun to contract. Anecdotally, at least, I think a case can be made that at the current juncture it is more of a coincidental indicator than we’ve been trained to believe.
The Bureau of Labor Statistics publishes its jobs report and details jobs added across 13 economic sectors (mining, construction, finance, business services, etc.). If one were to build a diffusion index of the quantity of sectors growing versus those contracting, you would get some pretty interesting results and with quite a bit of historic relevancy.
More specifically, when you look at a 6-month average of the number of expanding industries, things aren’t going well when we cross below 60%. Drilling down to the GFC, the metric crossed below 60% in December 2007. When did the NBER declare the start of the recession? December 2007. Economies don’t grow if businesses are looking to shed labor as an input. And we know how reliant the economy is on consumption; that’s hard to resume if people are getting laid off.
With the September jobs report, this measure of the labor market crossed below the 60% level. Under the surface, the weak labor market is getting confirmed in broader measures of corporate profits. While most look to public firms and their earnings reports, the National Income Product Accounts are a much broader look at the economy, incorporating small businesses. The most recent update (removing adjustments to align the data series with accounting in the S&P500) shows corporate profits declined in both the first and second quarters of this year.
Ned Davis Research expands upon the finding: “Since 1950, there have been 19 instances of NIPA profits declining for two consecutive quarters. But in only three of these cases S&P earnings growth was 10% or higher, similar to today: Q3 2000, Q2 2007, and Q1 2018. In the first two cases, S&P earnings converged toward the falling NIPA profits within two quarters, ushering in recessions. In the 2018 case, however, S&P earnings growth remained in the double digits through Q2 2019 and turned negative only with the onset of the Covid recession in Q1 2020. The point is that while the historical record warns of a high risk of an S&P earnings slowdown, the timing of one is still uncertain.”
The one exception to the diffusion index above was World War II, when both the whole economy pivoted to generating output for the war effort, and a concurrent massive fiscal stimulus occurred. Maybe the AI buildout can do the work of the fiscal stimulus this time around?
3. Recommended Reads and Listens
