Image: Falcon Northwest
It’s raining money, at least if you’re selling graphics cards. According to recent industry research, the market for graphics processing units is set to hit about $100 billion in 2024. It’s a massive shift in a market segment that was already growing quite quickly.
But don’t get too excited, gamers. We aren’t the ones buying. Almost all of that growth is driven by an explosion in AI infrastructure, according to Jon Peddie Research in a new market report, which confirms that Nvidia is reaping most of the benefit in this change.
That’s hardly surprising if you’ve even been glancing at the news lately. Selling AI hardware for data centers, just a couple of years after selling some of the same hardware for cryptocurrency generation, has driven Nvidia to become the most valuable company on the planet. (By market cap, anyway. There are other ways to measure corporate value, but that’s an argument for bigger brains than mine.)
For a bit of broad context on this $100 billion number, the estimated worldwide revenue for smartphones in 2023 was about $500 billion according to Statista, and vehicle sales totaled about $2.1 trillion for the top 20 manufacturers.
Nvidia is definitely the go-to vendor for any company that wants to rapidly scale up flexible, AI-focused data centers, if you have the scratch. Nvidia’s next-generation AI hardware is projected to cost between $30,000 and $40,000 per unit at the top tier. That’s the kind of “big iron” that makes an RTX 4090 look affordable, and it’s partly why the headline of this article is in dollars rather than concrete chips or cards.
But, of course, Nvidia isn’t the only player in the market right now, even if it’s absolutely dominating at the moment. Intel, AMD, Qualcomm, and Broadcom are all nipping at Nvidia’s heels, and even some surprise players like Apple and Meta (Facebook) get a mention in the short form of Jon Peddie’s report.
It’s worth pointing out that selling chips for AI data centers isn’t the only way to make some scratch in GPUs. Naturally, you’ve got both discrete and integrated graphics for PCs, smartphones, and tablets, game consoles, and even more specialized hardware going into cars, VR headsets, and wearables like smartwatches. As the report says, “GPUs have become ubiquitous and can be found in almost every industrial, scientific, commercial, and consumer product made today.” Total volume for the AI portion of the market is actually comparatively low, even if it’s driving the lion’s share of the profits.
Long story short: People are still going to play games and stream video, even if they’re talking to ChatGPT to help stem the crushing tide of digital isolation while they do it. Someone has to make and sell that hardware.
It’ll be interesting to see how the focus on high-margin business-to-business sales affects Nvidia’s output in 2025, and how the rest of the market adapts to that. After all, Nvidia can only make so many chips, and it’s not going to be too concerned with the $500 profit on a top-of-the-line RTX 5090 Ti Super-Duper Magnum MAX if it can more reliably make $10,000 selling a chip to the next AI startup that’s splashing around in buckets of angel investor cash. That could open up some space for the likes of AMD, Intel, and others to finally try and sell more GPUs to consumers, especially at the mid-tier and low-end.
But this is all navel-gazing. The only thing we can say for sure is that there’s going to be a lot of money made selling GPUs in the near future.
Author: Michael Crider, Staff Writer, PCWorld
Michael is a 10-year veteran of technology journalism, covering everything from Apple to ZTE. On PCWorld he’s the resident keyboard nut, always using a new one for a review and building a new mechanical board or expanding his desktop “battlestation” in his off hours. Michael’s previous bylines include Android Police, Digital Trends, Wired, Lifehacker, and How-To Geek, and he’s covered events like CES and Mobile World Congress live. Michael lives in Pennsylvania where he’s always looking forward to his next kayaking trip.
