
I read a recent Financial Times article about how Amazon, Google, Meta and Microsoft are on track to spend around $740 billion in capital expenditure this year, much of it on AI infrastructure. And as evident from the stock market tech sell off, investors are beginning to worry about whether these enormous investments will ever generate adequate returns.
Meanwhile, Elon Musk has been predicting a future in which AI and robotics create such abundance that goods and services become extraordinarily cheap, and money itself may eventually lose much of its relevance.
I find this to be an interesting contradiction. If AI delivers on its promise, many tasks that currently require expensive human expertise could be performed at a fraction of today’s cost. And as the technology improves and competition increases, even the cost of using AI itself could continue to fall.
But if that happens, how do the companies investing hundreds of billions of dollars today recover their investments, let alone earn attractive returns? Of course, falling prices can generate enormous new demand. But they can also destroy the profitability of the businesses that financed the original technology.
There is another dimension to this. If AI makes production increasingly independent of human labour, how will the income needed to consume this abundance be generated and distributed?
We could conceivably end up with extraordinary productive capacity, disappointing returns on capital and fewer opportunities for people to earn their livelihoods.
I do not know whether this is where AI is taking us. But I find the apparent contradiction between the enormous capital being committed today and the vision of near-free abundance tomorrow fascinating.
Could the greatest economic challenge of AI turn out to be not how much it can produce, but how the benefits of that production are distributed?