AI News

Yole Group: Global semiconductor revenue to approach $4 trillion by 2031, with memory becoming the largest growth engine

21ic Electronic NetworkUpdated Nov 7 13:06

AI summary

Yole Group forecasts that the global semiconductor device market will grow from about $700 billion in 2024 to nearly $4 trillion in 2031, with revenue growing more than 400% while shipments increase only about 60%. Memory leads with a growth rate of +43.2%, matching the logic and processor market from 2026 onward and continuing to be the largest market by revenue through 2031. Gary Huang points out that power is the biggest long-term constraint on AI: chip capacity can ramp up quarterly, but the power grid cannot.

Body · AI translation

Lyon, France / Shanghai, China, November 6, 2026 – The International Semiconductor Industry Group (ISIG) 2026 China Executive Summit recently concluded successfully in Shanghai, attracting more than 150 corporate executives and industry leaders from all links of the industry chain. Gary Huang, Group Vice President of Yole Group and Chief Customer Strategy Advisor for Asia, was invited to deliver a keynote speech, in which, based on the latest market data from Yole Group, he explained the shift in the semiconductor value landscape in the AI era: as the market scale continues to reach new highs, memory, interconnect and power supply are becoming key bottlenecks, and their strategic importance is rising in tandem with computing power.

Figure 1: Gary Huang, Group Vice President of Yole Group and Chief Customer Strategy Advisor for Asia, delivers a keynote speech at the ISIG 2026 China Executive Summit.

Value growth far outpaces volume growth, with prices driving semiconductors toward the trillion-dollar scale

In his speech, Gary Huang shared Yole Group's latest market forecast: the global semiconductor device market is expected to grow from about $700 billion in 2024 to nearly $4 trillion in 2031. Among application areas, telecommunications and infrastructure (AI data centers) are the main source of incremental growth.

By 2031, semiconductor device revenue will surge by more than 400% compared with 2024, while shipments will grow "only" by about 60%. The new growth is concentrated in core AI computing areas such as advanced logic chips and high-end memory. This is not a traditional cyclical rebound. AI demand is both sustained and continuously expanding, and its scale is pushing the semiconductor industry structurally onto a higher scale tier.

Figure 2: Global semiconductor device revenue forecast, 2024–2031 (by end market, in billions of US dollars). Source: Yole Group, Semiconductor Device Monitor, Q3 2026

Memory leads: three major markets, three growth trajectories

According to the content of Yole Group's presentation, the three major semiconductor device markets are in distinctly different investment cycles.

· Memory is growing fastest, with a growth rate of +43.2%. Memory has historically accounted for 25–30% of total industry revenue, but it has risen rapidly since 2025; in 2026 it will be on par with the logic and processor market and will become the largest semiconductor market by revenue, a position it will hold through 2031.

· The logic and processor market is also growing strongly, with a growth rate of +22.7%, but as the AI race gradually produces winners and losers, growth will slow somewhat, and some players will exit the market, merge or change their business models.

· Specialized and discrete devices follow a smoother path, with a long-term growth rate of around 6%.

Yole Group's analysis shows that AI-driven demand is reshaping the industry's revenue landscape, and memory will become the main growth engine of the next decade. Overall, many segments are benefiting from the AI wave, but the ones that truly achieve a leap in value are memory and processors.

Figure 3: Overview of semiconductor market drivers (memory, processors, specialized and discrete devices). Source: Yole Group, Semiconductor Device Monitor, Q3 2026.

Interconnect: copper cable still dominates inside the rack, while optical interconnect is poised to scale

Gary Huang said that, thanks to advantages in cost, supply and energy efficiency, copper cable is currently still the preferred interconnect solution for AI scale-up networks (that is, short-distance links connecting accelerators within a rack). Hyperscale cloud service providers are pushing copper cable to its physical limits by building larger, higher-power megawatt-class cabinets. Outside the rack, optical interconnect is already the standard solution, currently mainly using pluggable optical modules.

This landscape is about to change significantly. Between 2025 and 2031, the market for pluggable optical modules used in scale-out (inter-rack) applications will grow from $14.8 billion to $88.9 billion. Co-packaged optics (CPO) has even greater upside, especially in scale-up applications. CPO integrates the optical engine directly next to the switch chip or compute processor (XPU). At present, CPO is still constrained by high costs and an immature ecosystem, but as applications expand from switches to computing, its market size in scale-up applications is expected to grow from $44 million to $106 billion.

Gary Huang noted that to replace copper with optics inside the chip package (scale-in), performance still needs to improve by orders of magnitude. In addition, the technology still needs to be proven in terms of mass production capability, time to market and system-level reliability.

Gary Huang emphasized that what limits AI growth is not demand; the real constraint is the infrastructure needed to support AI development. Among all the pressure points, power is the one that requires the closest long-term attention, because the construction cycle for power generation and grid capacity is far longer than the ramp-up cycle for semiconductor capacity.

By 2030, the growth in power demand from AI data centers may outpace the speed at which new energy capacity comes online. In addition to increasing energy production capacity and improving the grid, accelerating the replacement of older data centers with more energy-efficient next-generation hardware may be a realistic path to delivering more computing power within the same power envelope. Power will remain a key long-term indicator for tracking the pace of AI growth.

"Chip capacity can ramp by the quarter, but the grid cannot," Gary Huang said.

Figure 4: Yole Group's portfolio of cutting-edge research products. Source: Yole Group, presentation at the ISIG 2026 China Summit.

At the end of his speech, the Group Vice President of Yole Group and Chief Customer Strategy Advisor for Asia concluded: "Yole Group's expert analysts do not believe that AI expansion will suddenly hit the brakes, but its growth will increasingly depend on coordination among energy, infrastructure and the semiconductor ecosystem. This requires timely insight so that we can capture signals before various market changes are reflected in revenue data."

Yole Group recently launched the quarterly-updated Semiconductor Device Monitor, covering 80 leading companies, more than 40 device categories and more than 25 wafer market segments, tracking changes in revenue, shipments and capacity. In addition, Yole Group offers the quarterly-updated Advanced Packaging Market Monitor, as well as a series of annual reports focusing on cutting-edge directions such as glass substrates, CPO and photonic packaging. These products provide companies with a channel for direct communication with Yole Group analysts, helping companies grasp cutting-edge trends in capacity planning, product roadmaps and competitive benchmarking early and seize market opportunities.

Source

21ic Electronic Network