At the recent media day hosted by Infineon in Shanghai, the company officially disclosed its latest strategic layout for 2026 and future business plans.
By analyzing its latest adjustments, one can clearly see the core logic behind this semiconductor giant's organizational restructuring and business focus shift, as well as how these initiatives address the most pressing realities of energy consumption and hardware transformation in current AI data centers and the physical AI sector.
Starting July 1, 2026, Infineon will restructure its original four business divisions into three major segments: Automotive Electronics, Power Systems, and Terminal Intelligence.
In response, David Poon, Senior Vice President of Infineon Technologies and President of Infineon Technologies Greater China, as well as Head of Consumer, Computing & Communications Business Greater China, stated that the essence of this adjustment is to integrate various product lines—such as medium/low-voltage and high-voltage power devices, microcontrollers (MCUs), and sensors, which were previously scattered across different departments—at the application level. From the customer's perspective, developing products previously required interfacing with multiple independent business units. After the organizational restructuring, customers can now engage with system-level solutions, which helps streamline supply chain processes and shorten product R&D cycles.
Architectural Shift in Data Center Power Supply Systems
Current market discussions on artificial intelligence (AI) mostly focus on the computing power of core chips. However, at the infrastructure level, the rising power consumption of server clusters is a physical limitation that must be addressed.
According to data disclosed by Infineon at the event, the power requirements for AI server racks are rising rapidly: the mainstream industry configuration was previously around 100 to 200 kilowatts, while the new generation of AI server racks has now increased to approximately 500 kilowatts, and is expected to further advance to the 1,000-kilowatt (1-megawatt) level within the next one to two years.
Under such high power density, chips have extremely low tolerance for voltage fluctuations (surge loads). Once power instability leads to a downtime, the cost of losses from server hardware and computing interruptions is exceptionally high.
Against this backdrop, Infineon has launched the "From Grid to Core" end-to-end solution, with the core focus on providing comprehensive power management across the entire chain. From power generation, transmission and distribution, and energy storage systems, to solid-state transformers and circuit breakers in data centers, and finally to the voltage regulation modules (VRMs) at the chip level that step down voltage from 50V to 12V and ultimately precisely to 1V, the goal is to optimize the power architecture and reduce overall system power consumption by more than 15%.
David Poon believes that future computing power competition will largely depend on the infrastructure's ability to control power conversion efficiency.
Component Penetration in Physical AI: Covering Automotive, Flying Cars, and Robotics
In the automotive electronics sector, the trend of software-defined vehicles (SDVs) is reshaping the distribution of semiconductor components.
Cao Yanfei, Senior Vice President of Infineon Technologies and Head of Infineon Technologies Automotive Business Greater China, revealed that Infineon's control and drive technologies, originally used within automobiles, are now extending into emerging physical AI fields such as flying cars and humanoid robots.
Taking humanoid robots as an example, the cost of their hardware and software control components accounts for more than half of the total machine cost. Currently, the value of semiconductor components Infineon can provide for each robot is approximately 500 US dollars. This includes motor drive control for robot joints, power devices to reduce temperature rise and power consumption, and electronic skin (e-skin) sensing technology for perception.
Furthermore, through the acquisition of Marvell's automotive Ethernet business and ams OSRAM's non-optical analog/mixed-signal sensor business, Infineon is transforming its established technologies in in-vehicle communication and perception into foundational standard components for robotics and automation equipment.
Pragmatic Considerations in Material Strategy
Regarding the selection of underlying semiconductor materials, although third-generation semiconductors such as silicon carbide (SiC) and gallium nitride (GaN) have received high market attention, Infineon has not blindly pursued the substitution rate of a single material in its internal technology roadmap, nor has it set its revenue share as a key performance indicator (KPI).
According to the logic it revealed, although the SiC business is expected to continue its strong growth this year, the core solution for the future remains a composite material combination of "Silicon + SiC + GaN." Different materials are suitable for different voltage, frequency, and power density requirements, and a mixed usage is necessary to achieve the optimal balance between cost and performance.
Additionally, David Poon mentioned that to ensure capacity flexibility, in addition to expanding its factories in Germany, Infineon is also building the world's largest and most efficient 200mm SiC power semiconductor wafer fab in Malaysia. Through the capacity allocation system of the "One Virtual Fab" model, it flexibly adjusts production lines according to the demand in different global application markets.
Industry Challenges and Localization Path
Addressing the recent price fluctuations in the semiconductor industry, David Poon stated in response to reporters' questions that current market volatility and the supply-demand landscape are influenced by numerous structural factors within the semiconductor value chain. The entire industry continues to face significant cost pressures across multiple segments, including energy, raw materials, transportation, and services, making price increases a relatively common industry phenomenon.
Regarding its localization strategy in the Chinese market, Infineon is advancing its "In China, For China" localization strategy into product development, customized services, and the translation of global leading innovations into tangible customer value. The newly launched Infineon Innovation Space and the local sample service center located in Wuxi have broken the past model where overseas headquarters defined products and the domestic team was only responsible for sales. This allows the Chinese team to directly define products based on local customer needs and complete trial production in local factories, effectively compressing the time-to-market for products.
Looking at the data, Infineon has ranked first in the global power semiconductor market for 22 consecutive years. At the technological intersection of increasing renewable energy installed capacity and a significant surge in AI computing power consumption, the company's path is to leverage its technical accumulation in power semiconductors and control chips, along with its end-to-end technology layout from "From Grid to Core" to physical AI applications covering new energy vehicles (NEVs) and robotics. Furthermore, it addresses the transformation demands brought by high-energy computing and physical AI through organizational restructuring that adapts to market and customer needs.