Qualcomm plans to enter the data center AI chip market with three main product lines to challenge Nvidia by year-end: custom hyperscaler chips, a dedicated CPU for agentic AI workloads, and an AI accelerator tailored for workloads with bespoke architectures.

The custom hyperscaler chips are expected to ship soon and target demanding AI workloads in data centers.

Qualcomm is pursuing a shift into data center AI chips to compete with Nvidia and capture a share of the growing data-center AI workload market.

In India, Qualcomm is leaning on close collaboration with the Tata Group for packaging and system-in-package work, with potential domestic manufacturing initiatives and increased local sourcing.

Qualcomm’s partnerships with Tata Group and Dixon Technologies in India are part of a broader strategy to use India as a manufacturing hub supported by government initiatives.

The company highlights a close manufacturing relationship with the Tata Group for packaging and system-in-package efforts, and potential collaboration with Dixon Technologies for smartphones and wearables, signaling rising local manufacturing activity.

Qualcomm aims to offer unique capabilities to distinguish itself in the data-center and hyperscaler segments.

Qualcomm anticipates an improved financial trajectory after the June quarter as memory chip price pressures ease and smartphone supply constraints loosen.

The company signals potential growth in the June quarter, with margins and revenue likely to rebound as supply constraints ease.

Qualcomm believes the data-center AI chip market can support multiple players, emphasizing a strategy of workload-specific, optimized solutions with a new architecture.

CEO/COO Akash Palkhiwala says Qualcomm has developed technologies with strong relevance to data centers and plans to enter with unique offerings at an appropriate time.

Palkhiwala emphasizes a strategy of differentiated architectures and solutions tailored to specific AI workloads, signaling a move away from a one-size-fits-all approach in data-center accelerators.