Crisis of Overproduction

 Fears of an "AI bubble" are back in the news this week, with investors starting to show some alarm about the level of indebtedness of the big AI companies and the mind-boggling and possibly unrealistic scale of future earnings they would need to generate ever to pay it off. 

Some analysts have started to talk about this as the risk that we may end up with a ruinous over-supply of AI infrastructure, after the current capital build-out, without ever generating enough consumer demand to actually pay for all these AI services we are developing. 

As one of the guests on the latest episode of Ezra Klein's podcast puts it: "There isn’t a major economic capex episode in Western economic history that didn’t end in a rolling series of crises, vastly inflated asset prices and a vast oversupply of capacity of whatever the thing was — railroads, fiber-optic cable, take your pick."

He goes on: "so that’s the bubble. The bubble is — once again, we’re vastly oversupplying something because of the availability of capital and the eagerness of the suppliers of the capital to continue building out in anticipation of future demand."

Of course, the most dogmatic economists in the neoclassical tradition might assert that oversupply is a non-issue. After all, human wants are infinite. 

But while this may be true in some cosmic sense—it is human nature always to find new sources of dissatisfaction and hence new desires—it is just as plainly the case that human wants for specific commodities are not in fact unlimited. 

Economists themselves were the first to introduce the concept of diminishing returns, after all. It is in fact possible to have enough labubus to finally exhaust one's interest in buying more. And now, our society is forced to grapple with the question of whether the same is true of AI services. 

Marxian theorists have long warned that this tendency to oversupply a commodity during a period of frenzied speculative interest (the classic definition of a bubble) creates an incentive for capitalist states to engage in imperialism and militarism—in order to artificially create a market for goods abroad that can no longer find enough demand at home. 

As William Appleman Williams argues in his Tragedy of American Diplomacy, for example, this quest to find markets for over-supplied goods (produced by excessive U.S. productive capacity) was a fundamental driver of U.S. foreign policy from the time of the "Open Door" policy on.

So too, Norman Mailer—in his socialist novel of the 1950s, Barbary Shore—written under the heavy influence of Trotskyist polemics at the time—argues that the United States was gradually coming to approximate the bureaucratic despotism of its rival the USSR, because of the internal contradictions of monopoly capitalism. 

The state capitalism of the era, he argued, was leading to a crisis of overproduction. This left the state with the choice of either accepting economic recession—or of using the military and the police state to create a market by force where none previously existed. 

As Mailer writes: "The productive capacity of monopoly has become so tremendous, its investment in machinery so great in comparison to the labor force it could exploit, that only the opening of the entire world market could solve its search for investment and profit even temporarily." 

He goes on: "monopoly cannot continue its operations on an adequate scale […] there is no choice but to engage in the production of armaments or suffer economic collapse."

This would be my big fear about the future trajectory of the AI industry. Most of our discussion about the AI bubble centers exclusively on the risk of "economic collapse." We fear that the bubble will eventually burst, investors will lose confidence that AI companies will ever generate enough revenue to justify their capital expenditures, and the global economy will go into recession. 

But suppose the U.S. government realizes this risk—and tries to head it off by artificially creating a market through wars of imperial conquest? We already know that AI has military and domestic surveillance applications. We also know the Trump administration has started multiple wars of aggression and built an enormous apparatus of domestic surveillance, detention, and repression. 

Along with the rest of the world, to be sure, I too fear the consequences of what happens if the AI bubble "bursts," because this technology never finds an outlet in sufficient consumer demand to justify its construction. 

But I worry even more about what happens if the bubble never bursts, because the U.S. military resorts to the old tactic of imperialism and militarism to create markets by force for the goods that it cannot sell at home due to ruinous overproduction. 

Mailer includes a prophecy in his novel, after all, that reads to me like a pretty accurate description of the way things are going under the second Trump administration: 

More money than goods to buy, an inflation of vast proportions can be prevented only if wages are reduced and exploitation increased. The result is a diminishment in the will to work and a drop in the velocity of industrial performance. Discontent is everywhere. The first examples of random sabotage, motivated by no more than brute exasperation begin to multiply. The police system which had been already expanded at the moment of entering the war, when hundreds of thousands of people politically suspect had to be found and imprisoned, now received a new levy. The police are everywhere, within the unions, in the military, at the seats of government power; they have almost reached the point where they co-exist with all of society. State profit and state surveillance, state-enforced poverty and state-endowed wealth.

Uh oh. I fear what's coming.