00:00 Speaker A
It is very important to consider the cost of AI, and this is what hyperscalers are paying. And let’s go to Wi-Fi Interactive. Here’s a nice graph of A of B. This dates back to 2007. Basically over the last 20 years, you’ve had semiconductor companies like Nvidia, Micron, Applied Materials, it’s been like a little basket. The white part is free cash flow. You can see the amount skyrocketing as they are getting paid a lot of money and a good profit for every tip. In the meantime, someone has to pay for it. That’s a hyperscaler. they are green. That’s Microsoft, that’s Alphabet, that’s Oracle, and a few others. Their free cash flow is plummeting. In other words, a huge transfer of wealth is occurring. In fact, A’s B calls this intergenerational wealth transfer. Now, the question is, what could go wrong here? Because hyperscalers think they can get paid. but
00:41 Speaker A
Torsten Slok, um Paulo’s chief economist, he has a very important chart. He shows that China’s AI models benefit more than the US. So the increased use of these Chinese models means that people are not paying as much money for hyperscalers. It’s just a theory. So if this continues, cash flow and profits could disappoint. And we’re not talking about this quarter, we’re talking about the next few years. A magnitude 7 decline could occur, straining balance sheets and increasing credit risk. But that’s what it is here. For this trade to work, many dominoes must fall or many pieces must fall in the right place. So I’ll post this diagram again. This is the key. This is your K for hyperscalers and semiconductor trading.
