The semiconductor industry is undergoing a structural shift driven by AI demand: McKinsey has sharply raised its forecast for the global semiconductor market in 2030, but shipment growth has not been significantly upgraded, with price increases becoming the main driver of market expansion.
McKinsey's latest research raises its baseline forecast for the global semiconductor market in 2030 from $1.6 trillion in autumn 2025 to $2.3 trillion, an increase of about 44%. Over the same period, global wafer shipments are expected to maintain an average annual growth rate of about 7%, broadly in line with previous forecasts. This means the added market value comes more from rising ASPs for advanced-node chips and memory rather than from pure shipment growth.
AI infrastructure investment is the core driver of this change. The five major hyperscale cloud service providers plan capital expenditure totaling about $800 billion in 2026 and further rising to $1 trillion in 2027, with the vast majority going into AI infrastructure, continuously pushing up demand and prices for advanced-node chips and memory.
In the first half of 2026, global semiconductor sales grew year on year at the fastest pace since the mid-1980s, and July hit a 40-year high; shortly afterward, global semiconductor market revenue exceeded $1 trillion for the first time.
Market forecast raised 44%, while shipment growth has not notably accelerated
McKinsey raised its CAGR forecast for the global semiconductor market from 2025 to 2030 from 13% to 19%, well above the historical average of 9% from 2014 to 2024. Its forecast range is $1.7 trillion to $2.6 trillion, with a baseline scenario of $2.3 trillion.
The forecast includes fabless companies, OEMs developing their own chips and vertically integrated chip companies in the statistics, and adds more detailed data on Chinese companies, so the forecast itself may be higher than those of some other institutions. Another major analysis firm also expects the market to reach $2.3 trillion by 2030, while two other institutions forecast a range of $1.7 trillion to $2 trillion.
More notable is that revenue growth and shipment growth are now "clearly decoupling." McKinsey expects overall wafer volume CAGR of about 7% by 2030, broadly similar to its autumn 2025 forecast, and it has not been raised in tandem with the sharp upward revision to market size.
This means that the increase in market size from $1.6 trillion to $2.3 trillion in 2030 mainly comes from higher ASPs and changes in product mix rather than a large increase in chip shipments. If AI demand continues to expand, the semiconductor industry's long-standing growth model dominated by capacity cycles may change.
Advanced-node ASPs rise against historical trends
Server and data center chips will become the fastest-growing area. McKinsey expects revenue in this market to rise from $330 billion in 2025 to $1.2 trillion in 2030, with a CAGR of 29%; wireless communication chips rank second, with an expected revenue increase of about $205 billion and a CAGR of 10%.
In advanced nodes, McKinsey defines 7nm and below as advanced processes and notes that their ASP trend has reversed historical patterns. In the past, wafer ASPs usually declined continuously in the first two years after a product launch, for example with 28nm process wafers falling about 10% annually and 150nm process wafers falling about 25%.
Now, however, 3nm and 5nm wafer ASPs are rising at 2% or more per year, mainly driven by strong AI and high-performance computing demand and limited advanced capacity. It is expected that by 2030, higher ASPs, shipment growth and product mix changes will together create about $710 billion in incremental value for advanced-node chips.
DRAM prices have risen sixfold cumulatively, with memory contributing $560 billion in incremental value
Price changes in memory are even more pronounced. AI model training and inference require large amounts of high-capacity, high-bandwidth memory, and HBM in particular is constrained by supply. Since early 2023, DRAM prices have risen about sixfold cumulatively, and ASPs have returned to their 2011 highs; based on the price level corresponding to the same historical shipment growth rate, current DRAM ASPs are about 550% higher.
Supply will be difficult to release quickly in the short term. McKinsey expects that if some long-term supply expansion measures proceed smoothly, memory supply may begin to catch up with demand around 2027, after which prices will tend to decline, though they may still remain above historical norms.
Commercial HBM gross margins are expected to fall from about 80% in 2024 to about 50% in 2030, mainly due to intensifying competition in the accelerator market rather than lower ASPs. Even so, memory is still expected to contribute about $560 billion in incremental market value, becoming an important source of AI-driven semiconductor growth.
Mature-node processes see volume growth but falling prices, with relatively limited contribution to growth
Mature-node and other chips will also benefit from the expansion of AI infrastructure, including field programmable gate arrays (FPGAs) that can flexibly adapt to AI architectures, interconnect and dataflow components, and data center power electronics, while segments such as photonics and power devices are also expected to grow relatively quickly.
Overall, however, growth in this segment is far weaker than in advanced-node chips and memory. Shipments will continue to increase, but ASPs are expected to keep declining, and by 2030 this segment is expected to contribute only about $100 billion in incremental value.
From this perspective, McKinsey's latest upward revision is not simply a higher semiconductor shipment forecast, but a reassessment of how the AI supply-demand gap affects chip prices and product mix. With shipment growth not significantly upgraded, rising ASPs for advanced-node chips and memory are becoming the main force pushing the semiconductor market toward more than $2 trillion.
Comments