OK first, I have no skin in the game, but what is remarkable is that I had completely understated the amount invested in the AI sector. To give you a sense of how large the investment in the sector is, the Bank of England in a recent study indicated that for AI investment to meet the current stated rate of return, the entire sector would need to generate nearly $2.5 trillion in revenues per annum. To give you a sense of perspective, it's equal to 140% of the total revenues from the global tech sector. BTW, that's a lot of money!
Today, right now, (and from the same study) corporate executives use AI for about 100 minutes per week, they pay small amounts and have generally indicated that they have not seen any real value to the product. Still they pay the minimum (a few hundred dollars a year). The second group are small businesses that have been able to scale up their operation using AI instead of human staff. Again, not nothing, but not a fortune either. They, on average, pay $20 a month for their AI services.
In the late 1990 vendor finance for telecom providers was an essential component of the business. Everyone did it, and it was part of the game. The problem was that vendor finance was underreported and when the "shit hit the fan" it took down giants (Northern Telecom was the most visible). It makes a lot of sense for Nvidia to provide vendor finance to many datacentre, the problem is if they don't all succeed what will be the impact on Nvidia. Right now the market is pricing all the AI players as if they are going to be the winners of the AI contest. However, there can only be one winner, and it may not be the one with the best technology (aka Betamax Vs VHS).
How come no one talks about that risk, simply the SEC has removed the rules that separated the research from the investment banking (that occurred in late 2025) So you have one broker that is pricing the shares of SpaceX at $800 (currently trading around $114). The logic is "all the potential AI revenues that SpaceX will generate. Part of the problem is that the stories for the other providers are identical – yesterday the profits for OpenAI were leaked – the numbers were not good, revenues were apparently declining (I have not seen the numbers myself).
The Financial Times shows the "AI bubble" to be the biggest ever. The issue is not that AI will not generate revenues, there are too many sectors that can benefit from AI, the problem is that everyone is priced as the #1 winner, and no one knows the links between all the participants. That's the problem. Google for example has guaranteed the income of a datacentre. If the data center goes bust google is the one paying.
On the other side of the equation Berkshire just invested $10 billion in an AI stock issuance. Again that is a lot of money but lets not forget that they BH is sitting on more than #200 billion of free cash Finally, and on a positive note, for years everyone has been criticising the tech sector for sitting on piles of money. Now people are upset that they are actually investing!
Comments