Let’s get this out of the way: we’re not saying we’re in a bubble, and we are absolutely not claiming we know when any bubble will burst. The following should not be construed as investment advice. Our last “Perspective” thought piece addressed some of the issues and concerns that bubbled to the surface around “Liberation Day,” and we thank those clients who allowed us to liberate some of their weekend for some live commentary and questions.

Artificial Intelligence promises nearly everything a civilization could want: a boost to our limited native brainpower, highly-capable robots (autonomous cars, drones, humanoid servants), advances in quantity and quality of life, and ultimately the most precious commodities of all: time and freedom. It’s not hard to imagine a Golden Age of leisure and work that requires more of our creativity and less drudgery. Indeed, some of the best science fiction movies have done just that (not the Alien series, though).

As the rapid surge of investment shows, this bounty will require more of almost everything: computing power, digital and energy infrastructure, and electric power for starters. There will be setbacks: electric utilities have existing commitments (notably to their residential customer base) and will have to add more capacity to power data centers.  Refining capacity for “rare” earth elements, and access to essential microchips are subject to significant geopolitical risks.

And those geopolitical risks are self-reinforcing. Foreign policy catch-phrases like “the End of History” and “G-1” are out of favor, and “Great Power Competition” is back in. Manufacturing capacity and expertise have been offshored to China for decades, and Taiwan is (forgive us) a mere chip shot away from a potential invasion. Can a nation rely on its main adversary for the supplies it needs to win a high-stakes race with that same adversary?

More questions: will all this capacity prove economically justifiable and how quickly? The buildout of the railroads through the 1890s and telecom infrastructure in the 1990s left in their wake durable and useful infrastructure–together with waves of bankruptcies. More efficiency in chip design and computing may mean we need less of everything else. Will national security interests (real or imagined) cause us to build more than economics would justify and inflate an even larger bubble? Will society find itself having underwritten a white elephant and left on the hook for even more debt?  Will this reallocation of capital further burden the consumer economy with higher prices, a housing shortage, and further neglect of (non-digital) infrastructure? Underlying these questions, will the efficiency that AI delivers come with more and better opportunities for those it displaces—a Golden Age—or will there be a deflationary bust (less of everything) and a dystopian level of inequality?

At Magnolia, we try not to predict what will happen–at least during working hours. What is happening right now is that current economic growth is highly-leveraged to the AI build out, as is the US stock market. The largest companies comprise an ever-larger share of market capitalization, profits, and returns. However, their revenues from AI so far are overwhelmingly costs to each other. The “hyperscalers” (as they’re now called) have been capital-light cash machines, but now they empty their tills into massive infrastructure projects–in some cases taking on substantial debt for the first time!

Will AI change the world? Count on it. For the better? LIkely a mixed bag. Will it “grow the pie” quickly enough to avoid an everything bubble? Will five attempts be enough to make the illustration look less like a bialy? Stay tuned.