Custom silicon reshapes AI chip market outlook

DataM Intelligence projects the global AI chip market will reach US$361.14 billion by 2035, expanding at a CAGR of 16.64% during 2026-2035. Growth is tied to generative AI adoption, larger data-center buildouts, rising training and inference workloads, and broader AI integration across edge devices, PCs, smartphones, automotive systems and robotics.

Recent North American developments show cloud and model companies moving closer to silicon design. Qualcomm and Amazon announced a long-term collaboration that includes a potential purchase of up to US$60 billion in AI data-center chips and related products, while OpenAI advanced its custom AI processor with Broadcom and TSMC. Meta plans to put its Iris AI chip into production in September 2026 and pursue a new chip every six months through 2027.

Asia-Pacific activity includes OpenAI and Samsung cooperation on next-generation chips and enterprise AI services, linked to memory supply for Stargate infrastructure. India’s IndiaAI Mission expanded shared compute capacity to more than 45,000 GPUs by June 2026, and Semicon 2.0 was approved with an outlay of ₹1,27,500 crore for chip design, fabrication, advanced packaging and the semiconductor ecosystem.

Technical priorities are shifting toward inference-first designs, high-bandwidth memory, low-precision computing, open Ethernet-based clusters, optical connectivity and chiplet-based rack-scale systems. NVIDIA, AMD, Intel, Google, Samsung, MediaTek, Qualcomm and Cambricon are competing on performance, energy efficiency, memory bandwidth and software ecosystems.