The number of companies developing AI processor chips now numbers well over a hundred, according to new research.
Grow a new Arm: UK advisory body wants investment in local AI chips
READ MORE
Ever since OpenAI unleashed ChatGPT over two years ago, interest in training and deploying AI models has ballooned, with graphics processing units (GPUs) becoming the accelerator of choice to drive all that effort, particularly those manufactured by US chip biz Nvidia.
But these are not the only options when it comes to AI training or especially inference – the actual use of the trained models – with other options available such as the neural processing units (NPUs) embedded in some desktop and laptop CPUs.
Other manufacturers want to get in on the action, perhaps with an eye to the fact that Nvidia became the first publicly traded company to pass $4 trillion in market value last month.
According to Jon Peddie Research (JPR), there are no fewer than 121 companies now talking about producing AI processors, ranging from tiny chips targeting embedded applications and Internet of Things (IoT-class) devices, right up to hyperscale datacenter products.
JPR’s Q3 2025 AI Processors Market Development Report finds that the US currently leads, as you might expect, with at least 59 of those companies hailing from the land of the not-so-tariff-free. China is some way behind with 14 companies, while most other nations have just a handful at most.
While America remains out in front, China’s DeepSeek and Huawei continue to push their own advanced chips, and India has also announced an indigenous GPU program, with production targeted for 2029, JPR says.
Within the US itself, California and Texas are the hotbeds of development activity, with California home to no fewer than 42 of the AI chip firms based in the country.
Taken together, all these firms have attracted more than $13.5 billion in startup funding, the research operation says, with many of them raising $100 million or more in the past year alone.
But JPR president Dr Jon Peddie cautions that a shake-out is likely to see many of these fledgling AI silicon hopefuls fall by the wayside over the next several years.
“AI processors are experiencing a Cambrian explosion, reminiscent of the 3D graphics boom of the late 1990s and the XR (extended reality) wave of the 2010s,” Dr Peddie says. “We expect rapid consolidation in the coming years, with the 121 players we track today shrinking to around 25 survivors by the end of this decade.”
The shake-out might also come sooner, as many customers are beginning to report that there is little or no return on investment for the billions they have sunk into AI development, believing (or perhaps hoping) it would deliver a leap in productivity, as The Register reported recently.
That hasn’t stopped UK advisory body the Council for Science and Technology (CST) advising the government that Britain should position itself as a world leader in designing AI chips over the next decade. ®