TSMC Estimates Semiconductor Market Value to Reach $1.5 Trillion by 2030, with COUPE as a Keyword
TSMC's Senior VP Chang Hsiao-chiang stated that while mobile phones drove semiconductor growth in the past decade, AI will be the future driver. He projects the global semiconductor market to reach $1.5 trillion by 2030, with AI and HPC contributing 55%.
📋 Article Processing Timeline
- 📰 Published: May 14, 2026 at 13:55
- 🔍 Collected: May 14, 2026 at 14:02 (7 min after Published)
- 🤖 AI Analyzed: May 15, 2026 at 07:35 (17h 33m after Collected)
(Central News Agency, Hsinchu, 14th) Chang Hsiao-chiang, TSMC's Senior Vice President of Global Business and Deputy Co-COO, stated today that while mobile phones were the main growth driver for semiconductors over the past decade, artificial intelligence (AI) is expected to be the future driving force. He predicts that the global semiconductor market value could reach $1.5 trillion by 2030, with AI and high-performance computing (HPC) contributing the largest share at 55%.
TSMC's Taiwan Technology Forum took place today at the Sheraton Hsinchu Hotel. Chang said that AI is developing much faster than imagined, has already changed the entire industry, and continues to evolve rapidly. AI has a significant impact on semiconductors and may be the most important and influential technology in human history. While mobile phones drove semiconductor growth in the past decade, the future driver will be AI.
Chang stated that the foundry innovation model, which separates IC design from manufacturing, has accelerated semiconductor innovation by entrusting complex and expensive manufacturing to foundries. Currently, almost all AI accelerators are supplied by IC design houses and foundries.
Chang mentioned that AI architecture will have different layers. NVIDIA CEO Jensen Huang often uses a "five-layer cake" analogy to describe the AI ecosystem, which includes power, data centers, chips, models, and applications.
Chang pointed out that if we break down the chip, it can be subdivided into three layers. The first layer is computation, the second is heterogeneous integration and 3D ICs, and the third, crucial for the future, is photonics and optical interconnects.
Regarding COUPE (Compact Universal Photonic Engine), Chang said that in the future, computation and communication in AI systems will no longer be just an extension of electronic signals; photonic technology will also become a vital direction. Through miniaturized and universal photonic engines, it is hoped to support the high-speed, low-power data transmission and system interconnects required for the continuous expansion of AI systems.
He pointed out that by 2030, the global semiconductor market value is expected to reach $1.5 trillion. Of this, 55% will be contributed by AI and HPC, 20% by smartphones, 10% by automotive, and 10% by IoT.
**CoWoS Explained**
CoWoS is an advanced semiconductor packaging technology that can be broken down into CoW (Chip-on-Wafer) and WoS (Wafer-on-Substrate). CoW refers to chip stacking, and WoS involves placing the stacked chips on a substrate, which can increase the data transfer speed between chips.
Through advancements in advanced packaging technology, chip manufacturers hope to continue improving chip performance, bypassing the physical limit bottlenecks gradually being encountered in the 3-nanometer process.
Chang stated that mobile phones continue to drive semiconductor innovation. In the second half of the year, phones using chips produced with TSMC's 2-nanometer process will be available. The radio frequency (RF) chips in these phones have also advanced to 6nm, image processors to 12nm, and wireless communication technology to 4nm.
Chang said that smart glasses have great development potential, currently using 4nm process technology and are just beginning to develop. The high-voltage display technology within them has advanced to the FinFET architecture, and it is expected that smart glasses will become available at a reasonable price in the future.
As for physical AI, Chang stated that physical AI is robotics, especially humanoid robots, which require a brain, sensors, and microcontrollers to control movement.
Chang said that humanoid robots represent a new frontier for semiconductor applications, much like when mobile phones first appeared 20 years ago. Humanoid robots may be the smartphones of the future.
Chang stated that by 2030, the foundry industry's value will be about $500 billion, the semiconductor market will reach $1.5 trillion, related electronic equipment about $4 trillion, and the information industry value will be $15 trillion, influencing a global economy of $150 trillion.
Chang said that Taiwan has the strongest AI supply chain, and TSMC has many partners, such as Quanta. He looks forward to working together with everyone to create a better AI future. (Editor: Chang Liang-chih) 1150514
TSMC's Taiwan Technology Forum took place today at the Sheraton Hsinchu Hotel. Chang said that AI is developing much faster than imagined, has already changed the entire industry, and continues to evolve rapidly. AI has a significant impact on semiconductors and may be the most important and influential technology in human history. While mobile phones drove semiconductor growth in the past decade, the future driver will be AI.
Chang stated that the foundry innovation model, which separates IC design from manufacturing, has accelerated semiconductor innovation by entrusting complex and expensive manufacturing to foundries. Currently, almost all AI accelerators are supplied by IC design houses and foundries.
Chang mentioned that AI architecture will have different layers. NVIDIA CEO Jensen Huang often uses a "five-layer cake" analogy to describe the AI ecosystem, which includes power, data centers, chips, models, and applications.
Chang pointed out that if we break down the chip, it can be subdivided into three layers. The first layer is computation, the second is heterogeneous integration and 3D ICs, and the third, crucial for the future, is photonics and optical interconnects.
Regarding COUPE (Compact Universal Photonic Engine), Chang said that in the future, computation and communication in AI systems will no longer be just an extension of electronic signals; photonic technology will also become a vital direction. Through miniaturized and universal photonic engines, it is hoped to support the high-speed, low-power data transmission and system interconnects required for the continuous expansion of AI systems.
He pointed out that by 2030, the global semiconductor market value is expected to reach $1.5 trillion. Of this, 55% will be contributed by AI and HPC, 20% by smartphones, 10% by automotive, and 10% by IoT.
**CoWoS Explained**
CoWoS is an advanced semiconductor packaging technology that can be broken down into CoW (Chip-on-Wafer) and WoS (Wafer-on-Substrate). CoW refers to chip stacking, and WoS involves placing the stacked chips on a substrate, which can increase the data transfer speed between chips.
Through advancements in advanced packaging technology, chip manufacturers hope to continue improving chip performance, bypassing the physical limit bottlenecks gradually being encountered in the 3-nanometer process.
Chang stated that mobile phones continue to drive semiconductor innovation. In the second half of the year, phones using chips produced with TSMC's 2-nanometer process will be available. The radio frequency (RF) chips in these phones have also advanced to 6nm, image processors to 12nm, and wireless communication technology to 4nm.
Chang said that smart glasses have great development potential, currently using 4nm process technology and are just beginning to develop. The high-voltage display technology within them has advanced to the FinFET architecture, and it is expected that smart glasses will become available at a reasonable price in the future.
As for physical AI, Chang stated that physical AI is robotics, especially humanoid robots, which require a brain, sensors, and microcontrollers to control movement.
Chang said that humanoid robots represent a new frontier for semiconductor applications, much like when mobile phones first appeared 20 years ago. Humanoid robots may be the smartphones of the future.
Chang stated that by 2030, the foundry industry's value will be about $500 billion, the semiconductor market will reach $1.5 trillion, related electronic equipment about $4 trillion, and the information industry value will be $15 trillion, influencing a global economy of $150 trillion.
Chang said that Taiwan has the strongest AI supply chain, and TSMC has many partners, such as Quanta. He looks forward to working together with everyone to create a better AI future. (Editor: Chang Liang-chih) 1150514