This article is more than 1 year old
With AI hype driving up server GPU prices, will cloud costs rise next? We reckon so
And what next for cooling, datacenter placement, and more – tune in and find out direct from our vultures
Register Kettle This week Timothy Prickett Morgan over at our sister site The Next Platform wrote a fantastic in-depth analysis of the effect this latest AI hype is having on datacenter GPUs.
You can catch the article here. In short, the demand for machine-learning accelerators to train, test, and run models is driving server-grade GPU prices through the roof, and we're likely to see rises in cloud GPU costs and a rationing of resources, too.
For instance, we've seen top-end Nvidia silicon selling for $40,000 a pop. As Tim wrote: "The situation is worse than the used car market here in the United States, and it is a kind of inflation that comes from too much demand and too little supply.
"A situation that vendors love when they know they can't make enough units anyway. The hyperscalers and cloud builders are rationing access to GPUs among their own developers, and we would not be surprised to see price increases for GPU capacity in the cloud."
For this week's Kettle we got together with Tim to discuss the situation, where this leaves IT buyers, and more, and you can watch it all below in less than 20 minutes.
Clockwise, from top left in the thumbnail, we have Timothy with El Reg's Tobias Mann, Chris Williams, and regular host Iain Thomson.
Get an idea of what our vultures really think, not just about GPU prices but also why cloud giants build datacenters in seemingly some of the strangest places, cooling technology and choices, the constraints facing these growing large language models, the hardware needed for recommendation systems, and more. ®