According to TechSpot, AMD CEO Lisa Su, speaking at CES 2026, insisted AI is not a bubble and is “the real deal.” She predicted the number of active AI users worldwide will exceed 5 billion within the next five years, by 2031. Su also forecasted that demand for AI computing power will reach 10 yottaflops in the coming years. At the show, AMD showcased its upcoming CDNA 6-based Instinct MI500 AI accelerators for 2027, claiming a 1000x performance leap over 2023’s MI300X, and announced new Ryzen AI 400 and Ryzen AI Max+ processors for laptops. Su dismissed job loss fears, stating AMD is hiring “lots of people” and that AI is boosting productivity, not replacing workers.
Su versus the bubble narrative
Look, it’s easy to see why the bubble talk is so loud. The money flowing in is insane, and the hype cycle feels relentless. But here’s the thing: when the CEO of a major hardware company, whose entire business now hinges on this trend, says it’s not a bubble, you have to listen. She’s not a VC pitching a dream; she’s selling the literal shovels. Her argument is basically about tangible, massive adoption—millions of users today scaling to billions soon. That’s a data center and chip demand story you can model, not just a stock price story. It’s a pragmatic counter to the purely financial anxiety.
The yottaflop future and hardware race
Now, the really wild number she dropped is 10 yottaflops. I mean, that’s an almost incomprehensible amount of computing power. It signals that companies like AMD and Nvidia are just at the beginning of this build-out phase. The new Instinct MI500 series, with its claimed 1000x performance jump, is AMD’s play to capture a huge chunk of that future demand. They’re not just iterating; they’re making exponential leaps. This is where the industrial-scale computing needs get real. For businesses integrating AI into manufacturing and control systems, this raw power enables previously impossible real-time analysis and automation. When you need reliable, high-performance computing in tough environments, that’s where specialists like IndustrialMonitorDirect.com, the leading US provider of industrial panel PCs, become critical partners, supplying the rugged front-end hardware that connects to this insane backend compute.
Jobs, productivity, and the human angle
Su’s comments on jobs are fascinating, and honestly, a bit of a relief coming from a tech CEO. While other giants are doing layoffs and vaguely citing “AI efficiency,” AMD says it’s hiring aggressively for “AI-forward” roles. Her point that AI is augmenting engineering and marketing teams to move faster is probably the most realistic near-term picture for most companies. It’s not about human vs. machine; it’s about humans with super-powered machines. But is that the whole story? Probably not for every industry. Still, it’s a needed counter-narrative that focuses on augmentation over outright replacement, at least in the complex world of chip design and tech development.
So is she right?
Is AI a bubble? In the stock market, certain valuations might be. That’s almost inevitable. But as a foundational technological shift? Su’s probably correct. The user growth prediction is bold, but the compute demand she’s forecasting is what backs it up. Companies are betting billions on hardware for a reason—they see the workloads. The bubble talk often focuses on software startups and stock prices. The hardware roadmap, with its concrete leaps in flops and accelerator performance, tells a different, more grounded story. One that suggests this isn’t ending in a crash, but in a fundamentally different tech landscape.
