According to Qualcomm's investor day last week, the company is pivoting away from competing with Nvidia on AI model training and instead targeting the broader AI inference market, valued at billions of dollars. As enterprises deploy more AI agents—software that performs complex tasks via AI models—inference workloads are becoming the larger computational burden. Research presented by Google, Microsoft, and leading universities in April 2026 shows AI agents require roughly 1,000 times more inference computing than humans for software development tasks.
Qualcomm is accelerating this shift through strategic acquisitions. The company completed its purchase of Alphawave Semi in December 2025 and recently announced deals for Modular to secure AI inference and software technologies. These moves add to its earlier Nuvia acquisition, enabling Qualcomm to build a complete AI data center platform spanning high-speed inference chips, CPUs, interconnect technology, and software—positioning it to challenge Nvidia's dominance. Product rollouts begin in fiscal 2027, with Microsoft potentially among early customers.