Qualcomm has announced a groundbreaking move to develop custom data centre processors that connect seamlessly with NVIDIA’s powerful AI chips, marking its return to the competitive server market.
This exciting news came out on Monday, during the Computex 2025 event in Taipei. Qualcomm aims to create central processing units (CPUs) that work hand-in-hand with NVIDIA’s graphics processing units (GPUs), which are widely used for training AI models like chatbots. As reported by Reuters, this collaboration is a big step for Qualcomm to tap into the booming AI infrastructure space.
Why This Matters?
These new CPUs will use Nvidia’s technology to ensure fast and efficient communication between processors. This is key for data centres handling heavy AI tasks.
Qualcomm’s CEO, Cristiano Amon, shared that linking their custom processors with Nvidia’s rack-scale architecture will bring high-performance and energy-efficient computing to data centres.
A Comeback for Qualcomm
Qualcomm isn’t new to this game. Back in the 2010s, they tried to build Arm-based CPUs for data centres with Meta Platforms, but faced budget cuts and legal hurdles.
Things changed in 2021 when they acquired Nuvia, a company with expertise in Arm processors, and hired former Apple chip engineers.
Now, they’re back with renewed focus, even signing a deal with Saudi AI firm Humain for custom CPUs.
Competing in a Tough Market
The data centre CPU space is crowded with giants like Intel and AMD, while Nvidia itself has entered with its Grace CPU.
Big cloud players like Amazon and Microsoft also make their own chips. By partnering with Nvidia, Qualcomm hopes to stand out with power-efficient AI solutions. Details on pricing or exact launch dates for these processors are not out yet.