AMD's Dr. Lisa Su has predicted that the AI GPU market could grow to $500 billion in 2028, or 'roughly equal to annual sales of the entire semiconductor industry for 2023'

It's becoming increasingly difficult to turn on the TV, walk through our doors, or even glance at our phones, without being bombarded with promises of a brave, new world of AI. It's not only about AI tools or end-user benefits. There are also endless claims of AI market growth, and money-making opportunities.

Lisa Su, AMD's CEO, has once again put her name in the hat for AI market hopefuls. And not without merit either, I should add. Su stated in AMD's Q3 earnings call for 2024 (via Seeking Alpha) that "In the Data Center, we expect AI accelerator TAM to grow at a rate of more than 60% per year, reaching $500 billion by 2028." This is roughly equal to the annual sales of the entire semiconductor industry for 2023.

It's not surprising that companies who stand to gain from a successful AI industry are often the ones who hype it up and predict the most amazing successes. It's hard not to agree, but it is also difficult to disagree. The AI industry has grown rapidly over the past few years, with a market value of almost $200 billion in 2017.

The market is still there, bubble or no bubble, until bursts.

Su's remarks need to be understood in a slightly different context. She's referring to the "TAM" (total addressable markets) for AI GPUs that is predicted in 2028. This is the maximum possible market value for the AI GPU market: an ideal number. Su's comparison to the semiconductor market in 2023 was a good one, with sales of about $500 billion. But these were actualsales and not TAM.

Su is not saying that the AI GPU will be as large in 2028 as it is today. But the numbers help to convey the magnitude of the predicted AI GPU explosion. It's important to remember that GPUs are made of silicon, and if the AI market grows, so will the semiconductor market.

AMD's AI GPU business, which has been running for less than a year, is now as large as its CPU division, and its stock price has more that doubled since the middle of 2022. AMD's GPU business is booming because GPUs are synonymous with AI in the wider world. Nvidia's Blackwell will take the lion's portion, with AMD's CDNA 4 taking the proverbial crumbs.

It's no surprise either that AMD's gaming revenues are down year-onyear, but this explosive growth and record revenues haven't been affected. Even though today's largest consoles are equipped with AMD graphics, gaming is still peanuts in comparison to AI and the rest.

As PC gamers, we still have ourpeanuts to consider. Let's hope the next generation of gaming graphics cards from AMD and Nvidia can bring some excitement back to the gaming industry. If Su's AI predictions are accurate, then we can hope that continued investment in GPU architectures will lead to AI accelerators and eventually trickle down to our gaming GPUs. They may be hand-me-downs, but they are still powerful. I'll take it.