Musk has once again poured cold water on today’s AI data centers. François, a well-known figure in the AI world, cited recent views on stock market volatility.
He believes that companies like SanDisk, SK Hynix, and Micron are rare buying opportunities. He thinks spot prices for GPUs are still rising, and much of the high-end memory supply has already been booked through 2029, so demand for computing power has not peaked.
But Musk pointed to two rather serious issues from a practical perspective.
First, manufacturers can’t just buy GPUs and call it a day. They also have to build and operate massive computing clusters, which is extremely challenging from an engineering standpoint.
Previous analyses have noted that building these systems involves engineering challenges such as providing supporting power and cooling, as well as ensuring reliability. Some analysts have even conducted on-site research.
He also thinks cash will be extremely tight next year, given the current state of the market. Even if the whole industry is delayed by just a few months—for example, manufacturers pay upfront for computing power, but the equipment takes a few months to start generating stable profits—the resulting working capital shortfall alone could reach hundreds of billions of dollars.
So while demand for AI computing power is indeed surging, the pressure on cash flow next year will be enormous.
$SNDK $SKHY
He believes that companies like SanDisk, SK Hynix, and Micron are rare buying opportunities. He thinks spot prices for GPUs are still rising, and much of the high-end memory supply has already been booked through 2029, so demand for computing power has not peaked.
But Musk pointed to two rather serious issues from a practical perspective.
First, manufacturers can’t just buy GPUs and call it a day. They also have to build and operate massive computing clusters, which is extremely challenging from an engineering standpoint.
Previous analyses have noted that building these systems involves engineering challenges such as providing supporting power and cooling, as well as ensuring reliability. Some analysts have even conducted on-site research.
He also thinks cash will be extremely tight next year, given the current state of the market. Even if the whole industry is delayed by just a few months—for example, manufacturers pay upfront for computing power, but the equipment takes a few months to start generating stable profits—the resulting working capital shortfall alone could reach hundreds of billions of dollars.
So while demand for AI computing power is indeed surging, the pressure on cash flow next year will be enormous.
$SNDK $SKHY



