There has never been anything like this. This time it’s different. All investors are going to be rich. In every economic bubble in history, from Tulips to railroads to Bitcoin, the same feverish speculation abounds. It doesn’t really matter how fantastic the technology becomes, there will be a cycle of boom, bust and reset.
Even if Open AI achieves its goal of creating AGI only investors in that company will benefit. The rest of the market will be pretty much as they are. Having built a directory of over 4,000 AI tools, it’s pretty clear that there is something bubbling in the air.
VC money is pouring into AI startups without a clear outcome of payback. That’s one reason why there are so many of them. A lot of wealthy people and companies are very excited about this paradigm shift.
And it certainly is a paradigm shift. The productivity enhancement of tools like ChatGPT or Midjourney is already massive. The cost reduction on video production is going to be extreme by the end of next year. In the long run, we’re very bullish on AI.
But we’re cautious and resistant of short term hype. We don’t believe ChatGPT will be killing Google any time soon, and there are some early warning signs.
Adoption vs availability
Adoption will likely lag behind the outcomes in 2025. I really thought we’d be further along by now, but it just shows – it’s not happening that fast. Productivity will only be enhanced for the minority who use specific tools, and there is a major obstacle in learning.
Getting a workforce who have only really just got used to digital to change their setup again is not something that will happen overnight. I doubt ChatGPT will get beyond 500m regular users by the end of next year. That is a long long way from ubiquity.
Chipmakers
At the same time major suppliers to the AI boom have been on an absolute rampage. Nvidia is an astonishing stock market darling. Other semiconductor chip makers have seen similar upward booms. But there is evidence of the market cooling.
While Nvidia and TSM remain high, Dutch chip maker ASML has pointed to disappointing orders, and a rapid fall in its stock price. The clamour for AI infrastructure investment is cooling off.
Are LLMs really that good?
A mild gloom has appeared over the chatbot market. They could soon eat all the training data on the web, what then? If the web becomes more AI generated, will ChatGPT train on its own output? A further problem highlighted by some recent Apple research is that LLMs can’t reason and plan, which is a pitfall for the so called ‘agentic web.’
A correction is likely
So in some sense it would be unsurprising to see some corrections next year. The US economy is likely to boom given the worst inflationary pressure is over, and the Trump administration promises tax cuts, but that doesn’t get away from the problems of the AI market.
Too much infrastructure investment. Too many possible tools. No one killer app. Not enough adoption.
This time it may well be different. The AI revolution could have more impact on humanity, for better or worse, than the Industrial Revolution beginning 250 years ago.
But it won’t beat the standard market economics of supply and demand in the short term. The supply is astonishing. Demand is really quite weak considering. A cold wind could blow soon.
