Welcome to the Monday newsletter, where we take a deeper look at what's happening in the AI industry.
Artificial intelligence can change the world and, at the same time, be inside a financial bubble.
Both can be true.
Hundreds of billions of dollars are being invested in chips, data centers, energy, infrastructure, and new AI models.
Companies are competing to build increasingly powerful models and acquire millions of users as quickly as possible.
The problem is that we still don't know whether all this investment will generate enough money and productivity to justify what we're spending.
Sam Altman, CEO of OpenAI, has acknowledged this directly. In his view, investors can be too excited about AI while the technology itself can genuinely be revolutionary.
“Are we in a phase where investors, overall, are too enthusiastic about AI? My view is yes. Is AI the most important thing that's happened in a long time? My view is also yes. Bubbles happen when smart people get too excited about something with real fundamentals.”
Sam Altman, CEO of OpenAI
That's the most important point: a technology can change the world forever and still generate a financial bubble.
It already happened with the internet.
In the late 1990s, investors poured enormous sums into internet-related companies. Many had huge valuations without proving they could generate profits.
Eventually, the dot-com bubble arrived. Many companies went bankrupt and enormous amounts of money were lost.
But the internet didn't disappear. Quite the opposite. It transformed communication, commerce, media, entertainment, and practically the entire economy.
Investors were right about the internet. In many cases, they were simply wrong about what particular companies were worth and how long the technology would take to deliver the results they expected.
Sundar Pichai, CEO of Google, thinks something similar could be happening with AI.
“There's some irrationality in the current AI boom. […] We can look back at the internet. There was clearly a lot of overinvestment, but none of us would question whether the internet was profound. I expect the same with AI.”
Sundar Pichai, CEO of Google
But AI raises another important question.
Heavy use of a technology doesn't necessarily mean it's creating a lot of economic value.
Someone can generate a hundred images they'll never use. A company can generate thousands of lines of code for a product nobody will buy.
We can create millions of videos, texts, and answers that consume computing power, electricity, and infrastructure while producing practically no economic return.
There was activity. But activity doesn't necessarily mean value.
Generating an image or asking ChatGPT a question might feel almost free to us, but someone is paying for the infrastructure behind that answer.
That's where the real problem appears. If operating costs remain enormous and the economic value produced doesn't grow enough, someone will eventually have to absorb the difference.
Andrew Ng, Coursera cofounder and former head of Google Brain, adds another important distinction: we shouldn't talk about all of AI as though it were a single bubble.
“AI isn't one thing, and different areas appear to be in a bubble to different degrees. The AI application layer is underinvested. Inference infrastructure still needs significant investment. On infrastructure for training models, I remain cautiously optimistic… but there could also be a bubble.”
Andrew Ng, Coursera cofounder and former head of Google Brain
In other words, we could be investing too much in some parts of the industry and too little in others.
The real test won't be how many people use AI. It will be how much value it produces.
Satya Nadella, CEO of Microsoft, frames it from exactly that perspective.
“To avoid an AI bubble, the benefits need to be distributed more broadly. AI needs to translate into real productivity and economic growth, not simply more spending.”
Satya Nadella, CEO of Microsoft
If a company uses AI and can now produce twice as much with the same resources, there's value.
If a programmer can build in a day what used to take a week, there's value.
If a small business can automate tasks that previously required five people, there's value.
If new products, industries, and businesses emerge that were previously impossible, there's value.
If that value ends up considerably greater than the cost of building and operating all this infrastructure, perhaps what we're seeing isn't simply a bubble.
Maybe we're paying upfront to build a new economic infrastructure.
But the opposite can also happen.
If, in a few years, costs are still too high, productivity gains were smaller than expected, and many companies only operated because investors subsidized their losses, we'll probably see bankruptcies, consolidation, and much lower valuations.
Even then, AI could still end up changing the world.
That's what makes this conversation interesting.
The bubble can burst and the technology can win.
The internet already proved both can happen at once.
Maybe the question isn't whether AI is real or whether it will change our lives. It's whether the economic value it ultimately creates will justify all the money we're investing today.
What do you think: are we building the infrastructure of the next great economic revolution, or are we inside an AI bubble?
Five hours writing it. Ten minutes to read it.
If anything here brings you value, those five hours were worth it.
Thank you for reading.