Human Analyst’s Take on the Future of the AI Market

Computer graphics analyst JPR (Jon Peddie Research) recently published a report detailing the explosion of the AI processor market. 

“AI Processors Quarterly Update,” released in October 2025, reveals, “As of the end of the quarter, 137 suppliers—from industry giants to VC [venture capital]-funded start-ups—are chasing this $387 billion market. Presently, the market is entering a new phase marked by consolidation, strategic pivots, and differentiated bets on future compute paradigms.”

The report sums up the evolution of AI processors driven by computer vision, beginning with early attempts to develop programs that can recognize cats, dogs, cars, and other objects, followed by the GPU’s emergence as the de facto AI processor. In a DE 24/7 podcast episode, Jon Peddie, founder and president of JPR, compares the current AI market to the peak moment in Darwinian evolution. “All the species in the world kind of exploded at one time and tried to find food, a way to evolve and survive and carry on. And that’s where we are right now [with AI processors],” he says.

In an online commentary published in October 2025, analyst Gartner warns, “Agentic AI Supply Exceeds Demand, Market Correction Looms.” The firm believes “that agentic AI markets will consolidate in the short term as hype and fear of missing out give way to fundamental economics. In this AI Vendor Race, the losers of consolidation will be undifferentiated AI companies and their investors.”

Peddie also anticipates a survival-of-the-fittest period is coming. “No market can sustain 100-plus suppliers. But right now, it’s exciting as hell, because these 137 companies are trying to do something different, and the thing they want to do most of all is to be different from NVIDIA,” he says.

Challengers trying to unseat the giants in the processor market also face threats. Typically, “the big company sees what the little company is doing and will try to do it bigger, better, and faster—or it will acquire the little company,” Peddie explains. 

Jon Peddie, founder and president of JPR. Image courtesy: JPR

Not all acquisitions will lead to better unions. Sometimes culture mismatches and simmering rivalries prevent the teams from working together. But Peddie is optimistic. “Roughly 80% of those acquisitions have been successful, which is a high number,” he notes. 

In the processor market, Intel still maintains a comfortable lead over its competitors, but its main rival AMD is grabbing market share. In August 2025, citing data from a report by Mercury Research and AMD, PC Magazine reports, “AMD’s share of desktop CPU shipments reached 32.2% in Q2, a historic high and up from 23% a year ago, according to the chip tracking firm Mercury Research” (“AMD’s Desktop CPU Market Share Hits Historic High,” PCMag.com). 

The relationship between Intel and NVIDIA was complementary when the former focused on CPU and the latter GPU. But fractions arose when Intel began making attempts to enter the discrete GPU market, most memorably with the Larrabee project, and NVIDIA began making CPUs targeting the data centers. 

In September 2025, NVIDIA, now a $5 trillion AI hardware behemoth, decided to invest $5 billion in Intel, changing the duo’s dynamics. Peddie thinks the partnership is a win-win deal for both. 

“It allows Intel to pull back on its GPU business, and concentrate on integrated graphics,” Peddie points out. Also, together, NVIDIA and Intel can conquer enterprise businesses more easily. “If you’re a data center manager, and your visitors say, ‘Hey, I’m from Intel, and he’s from NVIDIA, and we’re here to help you.’ What are you going to say? You’d probably say, ‘Thank you. Can I get you some coffee?,’” he quips.

Peddie sees the introduction of AI tools in design and simulation software as a remarkable development, but he also cautions progress may be slower than expected, due to the legacy code predating the GPU proliferation. For simulation in particular, the software takes advantage of the GPU’s massive parallel processing capacity to speed up the jobs. 

“[Advances] will come in drips and drops. This little sub process will suddenly be working on a GPU, and then that one will be working on a GPU, and so on. The GPU will eventually take over; it won’t be that much longer. Along with it, we will begin to see LLM (large language model) 
interfaces,” he predicts.

For design and simulation software users, the new challenge is, “Whether you do it on a keyboard or with a microphone, it doesn’t matter. But you’re going to have to learn to speak [AI’s] language,” he says. 

The AI language is AI prompts, or the commands you issue to the AI tools like ChatGPU to execute a task, from drawing an image to running a simulation. “And this is a threat, because there’s going to be a group of people who just don’t want to learn it or can’t learn it, and they’re going to be the casualties,” Peddie adds. 

Note: This article is based on the podcast with Jon Peddie, available on-demand here.