Qualcomm has officially confirmed that it is conducting comprehensive technical evaluations of Samsung Foundry’s 2nm process technology for integration into its future chip lineup. The mobile semiconductor giant also highlighted its deepening alliance with Samsung Electronics, spanning both smartphone ecosystem collaboration and next-generation LPDDR6 memory deployment to drive the on-device Agentic AI era.
Speaking at the Snapdragon Summit 2026 in Maui, Hawaii, Chris Patrick, Senior Vice President and Head of Mobile Handsets at Qualcomm, stated that the company is currently engaged in in-depth work regarding Samsung’s 2nm manufacturing technology. The engineering teams are actively assessing how this advanced process can be applied to upcoming mobile chipsets. Although the recently launched Snapdragon 8 Elite Gen 6 is fabricated by TSMC, Patrick emphasized that Samsung Foundry remains an enduring core partner and will continue to play a pivotal role in Qualcomm’s future silicon strategy alongside TSMC.
Highlighting the device ecosystem, Patrick identified Samsung Electronics—the creator of Galaxy smartphones—as Qualcomm’s strongest global partner in the mobile space, backed by an alliance spanning decades. Patrick emphasized that autonomous, on-device Agentic AI will spark a major smartphone replacement cycle. Because AI agents can interpret user intent and autonomously execute complex native system functions that users previously overlooked, consumers are expected to transition to higher-tier smartphones sooner to gain superior experiences.
Executing complex Agentic AI workloads on compact mobile hardware requires advanced memory solutions. To resolve severe computational bottlenecks, Qualcomm is collaborating closely with Samsung Electronics’ memory division to commercialize LPDDR6 DRAM. Engineers from both enterprises are actively addressing key design challenges, including novel signaling architectures and complex memory channel optimizations.
Furthermore, Qualcomm is evaluating Processing-in-Memory (PIM) and High-Bandwidth Computing (HBC) technologies. As data processing volume during Large Language Model (LLM) inference dictates response generation speeds, Patrick noted that traditional SoC architectures centered on external memory are ripe for re-evaluation. By bringing computational tasks directly adjacent to the memory modules, PIM effectively mitigates bandwidth latency, serving as a critical R&D frontier for Qualcomm to boost end-to-end performance in increasingly complex edge AI environments.
–Qualcomm and Apple Extend Global Patent Licensing Deal Effective April 2027
