Morgan Stanley has released a research report maintaining an "Attractive" rating on the Greater China semiconductor sector, with the core investment thesis centered on the promising outlook for AI-driven semiconductors and the critical role of memory innovation. The firm projects the global semiconductor industry market could reach approximately $1.5 trillion by 2030, with AI semiconductors contributing roughly half of that total.

According to the report, the total addressable market (TAM) for AI semiconductors is expected to hit around $753 billion by 2030, representing a compound annual growth rate of approximately 30% from 2023 to 2030. In a supply-chain-data-driven optimistic scenario, the TAM for cloud AI semiconductors could expand to $485 billion by 2026.

The bank notes that the outlook for cloud semiconductors is becoming clearer, pointing to a combined 87% year-over-year increase in capital expenditure among the four major cloud service providers—Amazon.com (AMZN), Alphabet (GOOGL), Microsoft (MSFT), and Meta Platforms, Inc. (META)—during the second quarter of 2026. Their capital expenditure-to-EBITDA ratio has already surpassed 70%.

Morgan Stanley estimates that global cloud capital expenditure could approach $1.4 trillion in 2027, based on the top 14 publicly listed global cloud service providers and excluding sovereign AI investments, with memory expected to account for nearly half of total global cloud capex that year. Utilization rates for logic semiconductor wafer foundries could reach 80% in the second half of 2026; however, excluding memory and NVIDIA (NVDA) AI GPU revenue, non-AI semiconductor growth is projected to decline in 2026.

The report also highlights that both DRAM and NAND are currently in a supply shortage, keeping memory prices elevated. With sustained robust AI demand, Taiwan Semiconductor Manufacturing (TSM) could expand its CoWoS capacity to 260,000 wafers per month by 2028. HBM consumption is forecast to reach approximately 44 billion Gb in 2027, with NVIDIA remaining the primary HBM consumer. Overall 2.5D packaging capacity is still expected to grow 50% year-over-year by 2028.

The bank also outlines key memory innovation directions, including process node migration, design, packaging, peripherals, and integration, estimating the memory innovation TAM excluding HBM at approximately $25 billion.