Photo by Steve Johnson on Unsplash

I confess to being very much a non-expert on AI. I read pieces in the business press and random articles that more knowledgeable friends pass along. But I can’t say I have any knowledge of the nuts and bolts of the technology. 

However, I can understand costs and the story there does not look very good for the United States AI industry. It looks like the latest offerings from China offer comparable speed in computing at a small fraction of the cost. According to this piece on the new MiniMax M2 Model, it can deliver performance that is comparable to the cutting edge U.S. models, at just 8 percent of the cost. This system is also open source. That makes it cheaper to adopt and alter than proprietary models. 

The need to develop efficient techniques was forced on China’s AI companies by the decision to deny them the most advanced computer chips produced by Nvidia and other chip makers. It now looks as though this prohibition may not have been a smart path from a competitive standpoint, even if it meant that the United States might have better AI in some abstract sense. Given the enormous need for electricity and water by data centers, it is a great thing that China seems to have designed systems requiring far less computing power, and therefore less electricity and water.

It is also worth noting that, while access to low-cost electricity may be a real constraint on the progress of AI in the United States, that is not likely to be a problem in China. Electricity there is plentiful, and the average price is less than half as much as it is here. 

With tech stocks hitting ever higher levels (Nvidia NVDA just crossed $5 trillion in market capitalization), it’s hard to believe there is not a serious bubble in the U.S. markets. Having lived through two major bubbles in my adult life, I am used to the optimists coming up with ever whackier stories. 

For example, for Nvidia’s current stock price to make sense, the company would have to earn about 10 percent of all projected corporate profits five years out. That’s not impossible, but I wouldn’t bet on it. I heard similar projections about Cisco Systems and other tech companies in the 1990s bubble. 

I also remember being assured in the 00s that house prices could never plummet because they are not traded like stocks on an exchange. (I believe Federal Reserve Board Chair Alan Greenspan once pushed that line.) They also said that we had never seen the sort of nationwide plunge in house prices I was predicting. I pointed out that we had never seen the sort of nationwide run-up in prices that the country saw from 1996 to 2005.

Anyhow, it sure looks like a bubble to me. It will be ironic if the success of China’s AI is what ends up being the proximate cause of its bursting.  

More By This Author:

The Economics Of Trade And Trump’s Little Water Pistol
No, Economists Didn’t Get It Wrong On Trump’s Tariffs
The Trump Boom: A Legend In His Own Mind