News

According to recent reports, Alphabet, Google's parent company, held approximately $548 million worth of Qualcomm shares in the third quarter of 2026. The reported investment comes as Qualcomm seeks to reduce its dependence on the smartphone chip market and accelerate its expansion into the AI data center business.

Qualcomm and Alphabet already have an established business relationship. Qualcomm processors have long powered Android smartphones, while the two companies also cooperate in the automotive sector and have recently expanded their collaboration into PC products. However, Alphabet's reported shareholding does not necessarily indicate that Google has committed to purchasing Qualcomm's data center chips. It should also not be interpreted as confirmation of a strategic investment specifically targeting Qualcomm's data center business.

Qualcomm is actively developing its data center business as part of a broader revenue diversification strategy. At its investor day in June 2026, the company outlined a target of generating $15 billion in data center chip revenue by fiscal 2029. It also raised its fiscal 2029 non-handset revenue target to $40 billion. Qualcomm is positioning AI inference as a key entry point into the market, developing its Dragonfly portfolio of AI accelerators and server CPUs, alongside advanced memory technologies.

On the technology front, Wccftech reported on November 2 that Qualcomm had completed validation of its first-generation High Bandwidth Cache (HBC) silicon at its San Diego laboratory. HBC technology stacks low-power LPDDR memory dies on an AI accelerator base die using wafer-to-wafer bonding. This approach shortens the distance between processing units and memory, helping improve data transfer efficiency.

According to Qualcomm, HBC can deliver up to 18 times the effective memory bandwidth of LPDDR5X. Its bandwidth per unit of power consumption is also claimed to exceed five times the company's projected HBM4 specification benchmark.

Regarding its product roadmap, Qualcomm plans to integrate first-generation HBC technology into the Dragonfly AI250 accelerator, with samples expected in the first half of 2027. The accompanying Dragonfly Rack is designed as a fully liquid-cooled AI inference system, featuring 38.6TB of memory capacity and an effective bandwidth of 7.4PB per second.

Qualcomm is also developing the Dragonfly C1000 server CPU based on its custom Oryon architecture. According to the company's announcement with Meta on June 24, 2026, the first-generation C1000 CPU is expected to enter production in the second half of 2028. Meanwhile, the Dragonfly AI300 represents the next generation of Qualcomm's AI inference accelerator roadmap and is expected to incorporate second-generation HBC technology in 2028.

Customer partnerships are another important part of Qualcomm's data center expansion. In September 2026, Qualcomm reached an agreement with Amazon that could enable Amazon to purchase up to $60 billion worth of Qualcomm AI data center chips and related products in the future. Microsoft and Meta have also become customers for Qualcomm's data center processors.

Despite these developments, Qualcomm still faces several challenges before its data center business can generate substantial revenue. New products must undergo customer validation, software platform integration, and long-term reliability testing before large-scale deployment. As a result, its ability to convert announced partnerships and technological progress into sustained commercial growth will be an important factor in determining its competitiveness in the AI infrastructure market.