Can Qualcomm take on Nvidia in the AI chip race?

Qualcomm unveiled two new AI accelerator chips designed to compete head-on with Nvidia, the current king of AI semiconductors. The announcement sent Qualcomm’s stock soaring 11%, as investors cheered the update. 

Context: until now, Qualcomm has been known for powering smartphones, not the data center. 

But with AI200 launching in 2026 and AI250 following in 2027, the chipmaker is betting big on data center computing. Both chips will come in full, liquid-cooled server racks, allowing up to 72 chips to work as one system, similar to setups used by Nvidia and AMD.

What’s the deal: the new chips build on Qualcomm’s existing Hexagon neural processing units (NPUs), the same tech that powers AI features in its phone chips. 

The company claims its rack-scale systems will be cheaper to run for cloud providers while delivering comparable power to Nvidia’s racks, which use around 160 kilowatts. 

Broader perspective: this marks a shift for Qualcomm from smartphones into a $6.7 trillion data center market expected by 2030. 

Nvidia still holds over 90% of the AI chip market, but competition is heating up with AMD, Google, Amazon, and Microsoft all developing their own accelerators.