-
Power semiconductors will define how well and how quickly the global economy adopts Edge AI and benefit from its promises. That’s why the race is stiffening among chipmakers to offer the most innovative power management components and systems. Who is winning?
-
What’s at stake: The stakes for power semiconductor makers in the Edge AI market are immense, encompassing billions of dollars in potential revenue, leadership in technological innovation, and a pivotal role in shaping the future of intelligent, energy-efficient devices. Those who can deliver the most advanced, dependable, and sustainable solutions stand to define the next era of electronics.
Artificial intelligence has rapidly evolved from a futuristic concept to a transformative force reshaping every sector of the global economy. But while much of the early excitement around AI centered on massive data centers and cloud-based supercomputers, the focus in recent years has shifted toward the “edge,” the network’s frontier, where data is generated and decisions must be made instantly.
Edge AI, which refers to the deployment of AI models on local devices rather than in distant clouds, is favored to drive everything from smart cameras and autonomous vehicles to industrial robots and wearable health monitors. As the intelligence of these devices grows, however, so too does the complexity of their power requirements. The race is on among semiconductor manufacturers, device makers, and system integrators to deliver energy-efficient, high-performance solutions that will define the next era of intelligent technology.
Edge AI devices are a diverse and rapidly expanding category. They include smart security cameras that can detect suspicious behavior in real time, industrial sensors that monitor equipment health and predict failures, wearables that track vital signs and alert users to anomalies, drones that analyze crops or inspect infrastructure, and even compact servers that enable AI-powered analytics in remote locations. “We are on the verge of a significant transformation at the tiny edge,” said Remi El-Ouazzane, president, microcontrollers, digital ICs, and RF products group at STMicroelectronics in recorded statements earlier this year.
The range of Edge AI devices is staggering. In manufacturing, edge-enabled robots and vision systems are optimizing assembly lines and ensuring quality control. In healthcare, portable diagnostic devices are using AI to interpret medical images and monitor patients in real time. In transportation, autonomous vehicles, and advanced driver-assistance systems (ADAS) rely on edge processing to make split-second decisions. Even in the consumer space, smart home devices and personal assistants are increasingly running AI models on-device, enabling faster and more private interactions.
The Edge AI hardware market is forecast to surge to $58.90 billion by 2030 from $26.14 billion in 2025, with the supporting power electronics market projected to reach $76 billion by 2032. This growth is fueled by the proliferation of smart devices, ranging from autonomous vehicles and industrial robots to wearables and smart cameras, which require increasingly sophisticated and efficient power solutions.
The transition to wide-bandgap materials like silicon carbide (SiC) and gallium nitride (GaN) is accelerating. These materials enable higher efficiency, smaller device footprints, and better thermal management, which are essential for Edge AI’s demanding applications.
Who’s who?
Infineon Technologies and STMicroelectronics currently lead the global rankings for power chip suppliers to Edge AI device manufacturers, with ON Semiconductor, ROHM, and Vishay also in the top tier. Maintaining or improving this position means investing in R&D, manufacturing scale, and strategic partnerships.
Geopolitical tensions and supply disruptions have made diversification and supply chain robustness a top priority. As Jochen Hanebeck, CEO of Infineon, noted, adapting to global complexities and ensuring reliable supply to key markets like China and the U.S. is crucial for sustained leadership.
Edge AI devices demand ever greater computational power within strict energy and thermal budgets. The ability to deliver high-performance, energy-efficient power semiconductors is a key differentiator. Innovations like Infineon’s 300-mm GaN wafers and STMicroelectronics’ 8-inch SiC production lines are examples of how suppliers are pushing technological boundaries.
Collaborations with AI platform leaders are opening new avenues for energy-efficient, on-device AI, making power semiconductor makers essential partners in the broader AI ecosystem. The rapid expansion of Edge AI is creating new, high-margin revenue streams for power semiconductor makers. However, price competition, especially as SiC and GaN technologies mature, could pressure margins unless offset by scale and innovation.
Major players are committing billions of dollars to expand their capacity and develop next-generation materials. For example, STMicroelectronics’ $2 billion to $2.3 billion capital expenditure plan for 2025 is aimed at meeting surging demand from electric vehicles and AI infrastructure. This transformation involves the increasing augmentation or replacement of our customers’ workloads by AI models. Currently, these models are used for tasks such as segmentation, classification, and recognition. In the future, they will be applied to new applications yet to be developed.”
Power requirements
The power demands of Edge AI devices are as varied as their applications. At the low end, simple sensor nodes running TinyML models may consume just a few milliwatts, allowing them to operate for years on a coin cell battery. At the high end, edge servers equipped with AI accelerators or GPUs can draw tens of watts, requiring sophisticated power management and cooling systems. In between are devices like smart cameras and wearables, which must balance the need for real-time AI processing with the constraints of battery life, size, and thermal management.
“It is a common misconception that AI is purely a big datacenter, power hungry application,” said Tom Hackenberg, principal analyst, memory and computing group, Yole Group, in a report. “This is no longer true. Today’s IoT edge applications are hungry for the kind of analytics that AI can provide. The STM32N6 is a great example of the new trend melding energy-efficient Microcontroller workloads with the power of AI analytics to provide computer vision and mass sensor driven performance capable of great savings in the total cost of ownership in modern equipment.”
Edge AI’s unique power profile is shaped by several factors.The choice of processor, whether a low-power microcontroller, a specialized AI accelerator, or a general-purpose GPU, has a profound impact on energy consumption. AI accelerators optimized for edge workloads, such as those from Nvidia, Google, or Apple, are designed to deliver high performance per watt, but even these must be carefully matched to the application’s needs.
The size and architecture of the AI model itself play a critical role. Smaller, quantized models (such as those used in TinyML) are far more energy-efficient than large, general-purpose neural networks. Advances in model compression and pruning are enabling more sophisticated AI capabilities within tight power budgets.
Many edge devices operate intermittently, waking only when needed to conserve energy. Power management strategies, including dynamic voltage and frequency scaling, intelligent sleep modes, and event-driven activation, are essential to maximizing battery life. Edge AI devices are often deployed in challenging environments, from factory floors to remote agricultural fields. Power supplies must be robust, dependable, and capable of handling fluctuating loads and harsh conditions.
Powering Edge AI
The transition to Edge AI brings a host of new challenges for device designers and semiconductor suppliers. Chief among these is the need to deliver ever-greater computational performance without exceeding the strict power and thermal budgets imposed by compact, often battery-powered, devices. Unlike data centers, which can rely on abundant power and advanced cooling, edge devices must operate within severe constraints.
As the intelligence of edge devices grows, so does their appetite for power. Yet many applications, especially those in remote or mobile settings, demand months or years of operation on a single battery charge. Achieving the right balance between performance and efficiency is a constant struggle. High-performance AI processing generates heat, which can degrade device reliability and shorten lifespan.
In small form factors, effective thermal management is a major engineering challenge. Power supplies must deliver consistent voltage and current despite rapidly changing workloads. Voltage fluctuations or power interruptions can cause AI models to fail or produce erroneous results.
Despite these challenges, the Edge AI revolution presents enormous opportunities for innovation. Infineon, the world’s largest manufacturer of power semiconductor components, has been at the forefront of this transformation. In 2025, Infineon became the first company to develop 300-millimeter GaN power wafer technology, a breakthrough that promises higher efficiency and greater scalability for applications ranging from data centers to electric vehicle charging stations.
The Bottom Line
To stand out in the power semiconductor market for Edge AI devices, companies must lead in advanced materials like SiC and GaN, integrate closely with AI ecosystems, and deliver energy-efficient, reliable products. Building resilient supply chains, offering strong technical support and tailored solutions, and maintaining rapid innovation are also essential. Excelling in these areas will help semiconductor firms secure leadership positions and become key partners in the fast-growing Edge AI sector.
Bolaji Ojo
Publisher and Editor in Chief, TechSplicit (formerly the Ojo-Yoshida Report)
This article was published by TechSplicit (formerly the The Ojo-Yoshida Report). For more in-depth analysis, register today and get a free two-month all-access subscription.
