During his keynote at Computex 2025 in Taipei today, Qualcomm president and CEO Cristiano Amon announced that the company will re-enter the data centre market with the development of its own specialised CPU chips. The new processor will incorporate technology from NVIDIA, enabling tight integration with the latter’s AI chips.
NVIDIA’s graphics processing units (GPUs) have become essential in data centres for training large-scale AI models that power services such as chatbots. These GPUs are typically paired with central processing units (CPUs), a segment currently dominated by Intel and AMD.
By leveraging NVIDIA’s technology, Amon said Qualcomm’s upcoming chip will be able to communicate efficiently with NVIDIA’s GPUs – an important capability given the latter’s prominence in AI infrastructure. He also revealed that the company had recently formed a partnership with Saudi Arabian AI firm Humain to support development of the new data centre CPU.
As mentioned earlier, this isn’t Qualcomm’s first foray into the data centre space. In the 2010s, the company developed an Arm-based CPU that was tested by Meta Platforms. However, the initiative was ultimately shelved amid legal hurdles and cost-cutting measures.
