The Market Cycle Puzzle

Preface: Explaining our market timing models

 

 

 

Two weekI highlighted the relative breakouts of gold prices compared to the S&P 500 and the 60/40 portfolio and argued that the breakouts represented a transition of paper to hard asset leadership. However, the last time the gold/paper asset cycle turned, it coincided with a bottoming in other market leadership factors, namely value/growth, small cap/large cap and international/U.S. stocks. This time, the turn in factor leadership isn’t evident.

What are the investment implications of the continuing divergence? Changes in market leadership often occur when a market transitions from bull to bear. Does this mean that the bull is still alive and how should investors position their portfolio allocation?
 

 

The Debasement Trade

The most likely explanation is to distinguish between increasing nervousness over the outlook for U.S. sovereign paper compared to the enthusiasm for dominant U.S. companies. The rise in the gold price is reflective of the so-called “debasement trade”. It just doesn’t reflect nervousness over the U.S. federal deficit, but deficits all around the world. Robin Brooks observed that bond yields and forward yield curves are steepening in the major developed market economies, not just the U.S., which is a signal of increasing fiscal dominance where central banks are forced to accommodate the rising fiscal deficits of their governments.
 

 

 

U.S. Corporate Dominance

By contrast, investor enthusiasm for dominant AI-related companies which are all based in the U.S. continue. If AI stocks are in a bubble, it’s only just starting. The forward P/E gap between the cap-weighted and equal-weighted S&P 500 really only started to diverge with the introduction of ChatGPT three years ago

 

Investors have seen this movie before. U.S. stocks dominated global stock performance after the GFC, and for the correct reasons. The dominant technology stocks were all based in the U.S., and the Apples, Googles and Amazons of the world.
 

Even though the capital expenditure rates of the hyperscalers look odd scale insane, valuations may not be totally out of line. Goldman Sachs argues that while the capex to sales rate (dark blue line) has risen to Dot-Com Bubble level highs, capex to free cash ratios (light blue line) are still reasonable.

 

 

The Value vs. Growth Dilemma

The dominance of U.S. megacap growth stocks brings up a valuation dilemma for equity investors. Sure, value stocks are trading at a large historical discount to growth stocks, but the earnings growth of the growth stock universe has outpaced value for the last four years.
 

 

A similar dilemma arises when investors consider the forward P/E ratios of major global regions. The U.S. equity forward P/E ratio is elevated on an absolute basis, but the earnings multiples of other regions are elevated relative to their own history as well.
 

 

 

How the Bubble Could Pop

Where does that leave us?

It’s unclear whether AI stocks are in an investment bubble, and investors won’t know until after the fact. My assessment is that they are in a bubble, but the bubble inflation is in early stages.

Here is a scenario of how the bubble may pop, but the process will not follow the conventional script of the Dot-Com era. A report by the conservative think tank Center for Security Policy will be the battleground for a new “cold war” between the U.S. and China. The report argues that Chinese dominance would mean the Chinese Communist Party would be able to dictate “global technology standards worldwide”. The Trump Administration, and previously the Biden Administration, recognized these risks and took steps to retard the progress of Chinese AI research.

I offer an out-of-the-box view of how AI competition may evolve. Tracy Alloway at Bloomberg observed that while the aim of U.S. policy is to dominate the platform and own the global technology standard in the same way that U.S. social media companies owned their standards, Chinese companies are pursuing a strategy of commoditizing AI: “Wall Street still out there valuing US AI like it’s a $5,000 espresso machine with a supposedly infinite potential market. Meanwhile China’s already been handing out free Nespresso pods to anyone who’ll have them. Which is interesting because it means China is effectively pursuing a commodification strategy for AI (and makes the real U.S.-China tech competition about power availability rather than model or data sophistication).”

The U.S. hyperscale arms race has been to build the best LLM platform using their own databases based on a “winner take all” business model. Many companies and government organizations don’t want their proprietary data residing in public platforms and they want AI solutions based on their own data. Small language model is sufficient for these applications and they don’t need the enormous computing resources of LLMs. Those data-light platforms are all currently Chinese, which supports Alloway’s commoditization thesis.

One day, investors may wake up and realize that the U.S. hyperscalers were fighting the wrong war. The value-added in the AI value chain belongs to the implementers who pursued the commoditization strategy.
 

If this thesis correct, the shift is only starting and the peak is still a few years away.

On the other hand, Time Magazine’s choice of the Architects of AI as its “Person of the Year” is a magazine cover contrarian indicator that the top is near.