A new wave of custom AI chips is testing Nvidia's grip on the market that powers artificial intelligence.
According to Silicon Canals, OpenAI's chip developed with Broadcom shows that Big Tech is moving beyond its heavy reliance on Nvidia. The outlet argues that the timing of the effort matters more than any single chip announcement — a sign that the industry's largest players are serious about building alternatives rather than making a one-off statement.
OpenAI is not alone. Nikkei Asia reports that Qualcomm is challenging Nvidia's AI grip with a chip that ditches HBM, the high-bandwidth memory that has become a costly and supply-constrained ingredient in today's leading AI accelerators.
As qz.com notes, a growing roster of companies is now trying to challenge Nvidia's dominance in AI chips. Together, these moves point to a broader shift: the buyers and partners who have fueled Nvidia's rise are increasingly designing or backing their own silicon.
Nvidia has held a commanding position in the hardware that trains and runs modern AI systems, which has translated into pricing power and long waiting lists for its most sought-after processors. Efforts by OpenAI, Qualcomm and others to field competitive chips could, over time, give customers more options and ease that bottleneck.
Why it matters: if these challengers gain traction, the cost, availability and control of the chips underpinning the entire AI boom could start to shift away from a single dominant supplier.