ChatGPT might bring about another GPU shortage - sooner than you might expect

ChatGPT
(Image credit: CHUAN CHUAN via Shutterstock)

Unless you live in a snug, deep cave, you’ll definitely have heard of ChatGPT. The AI-based text tool developed by the non-profit OpenAI has become prolific across the internet and foreshadows an explosion of AI development and mainstream use.

In order to create and maintain the huge databases of AI-analysed data that ChatGPT requires, the tool’s creators apparently used a staggering 10,000 Nvidia GPUs to train the AI to produce interesting and quirky responses. 

With the increasing popularity of AI, tech giants like Microsoft and Google are eager to integrate it into their own services. Forbes made an estimate as to how much it would cost to integrate AI into every single Google search, estimating the 4,102,568 Nvidia A100 GPUs needed would run the tech giant a whopping 100 billion dollars - paid to Nvidia, of course, which is a really hard number to wrap your head around even if it is just a guess on Forbes’ part.

The gamers among us will probably read this and sweat just a little, as it spells out bad news for the people that are looking to buy GPUs anytime soon. We may see a graphics card shortage akin to the recent one caused by high demand from crypto-mining, as companies will need plenty of GPUs to train AI cores to make the most of this newfound craze.

Analysis: Hello darkness, my old friend

This shortage scare will probably only affect Nvidia card sales since they have the most advanced AI cores on the market right now - however, that might change. Should other companies like AMD and Intel see the high demand and potential profit pool they may push harder to match Nvidia in the AI arena.

On the other hand, if Nvidia shifts more attention toward the AI sphere - and Team Green certainly wants to - it could leave an opening for AMD to fill up the gaming space for consumers. A shortage of Nvidia GPUs for gamers could push people towards AMD graphics cards.

Should the demand for GPUs in the future reach incredibly high levels the availability of those GPUs will suffer greatly for the average gamer. We won’t know just how far the demand could crawl and the number of GPUs that’ll be needed to train the AI models. 

As mentioned above, we already had to deal with a massive GPU shortage not too long ago thanks to miners and scalpers in the cryptocurrency space, which gives us an idea of what a fresh GPU shortage could look like. Crypto miners bought up as many GPUs as they could to mine currencies like Ethereum, and whatever stock wasn’t being used to mine was bought by scalpers who massively inflated the prices.

The shortage only eased up last year with the ‘crypto crash’ and Ethereum merge, and we're still seeing Nvidia graphics cards sell for far above MSRP. If Nvidia does commit to producing more GPUs for AI training purposes, expect graphics card prices to rise again.

Muskaan Saxena
Computing Staff Writer

Muskaan is TechRadar’s UK-based Computing writer. She has always been a passionate writer and has had her creative work published in several literary journals and magazines. Her debut into the writing world was a poem published in The Times of Zambia, on the subject of sunflowers and the insignificance of human existence in comparison.

Growing up in Zambia, Muskaan was fascinated with technology, especially computers, and she's joined TechRadar to write about the latest GPUs, laptops and recently anything AI related. If you've got questions, moral concerns or just an interest in anything ChatGPT or general AI, you're in the right place.

Muskaan also somehow managed to install a game on her work MacBook's Touch Bar, without the IT department finding out (yet).