The global edge AI chips market was valued at USD 16.80 billion in 2025. It is estimated at USD 20.60 billion in 2026 and is forecast to reach USD 106.00 billion by 2036, growing at a CAGR of 17.8% over the 2026-2036 forecast period. The report examines edge and cloud workload splits, chip architectures, software ecosystems, AI content per device, regulation, trade policy, pricing, the value chain, and competition.
Edge AI chips run AI models on or close to the device that produces the data instead of in a central cloud data center. The study covers SoCs with integrated NPUs for phones, PCs, tablets, wearables, and XR devices; standalone accelerators in M.2, PCIe, and module formats; automotive AI processors for driver assistance, automated driving, and smart cockpits; edge GPUs and AI modules for robots and industrial machines; FPGAs; and AI-capable microcontrollers for battery-powered sensing. Standalone accelerators, automotive processors, and AI modules are counted at full value, while SoCs with integrated NPUs are counted only for the share of their value attributable to AI.
The market is moving from basic features such as camera enhancement and wake words to generative and physical AI, and Microsoft's Copilot+ PC specification, introduced in 2024, sets a new performance floor for laptops by calling for NPUs of at least 40 trillion operations per second. Supplier results show the shift: Qualcomm's automotive sales hit a record USD 3.96 billion in fiscal 2025, 36% higher than a year earlier, and its IoT sales grew 22% to USD 6.6 billion, while Ambarella, as reported in its fiscal 2026 results, earned 80% of its USD 390.7 million in revenue from edge AI SoCs. Robot adoption adds to demand, with 542,000 new industrial robots installed globally in 2024, according to the International Federation of Robotics.
Growth is held back first by device constraints: phones, glasses, and wearables run on small batteries with passive cooling, so chips must deliver results within tight power and thermal budgets, and sustained AI workloads can force throttling. Larger models need more DRAM, and memory became far more expensive in late 2025, when Samsung was reported to have lifted prices on some memory chips by up to 60%; this may lead makers to hold back AI features in cheaper models. Each vendor's NPU has its own compiler, runtime, and tools, so developers must adapt models for every chip and operating system. Consumer devices are price-sensitive, and automotive processors must meet AEC-Q100 and ISO 26262, with design wins taking two to four years to reach production. Further challenges include fitting multi-billion-parameter models into limited memory, protecting on-device models and data under rules such as UN Regulation No. 155 and the EU Cyber Resilience Act, a shortage of edge AI engineers, and U.S. export controls combined with China's push for domestic suppliers.
The main opportunities lie in accelerators that bring generative AI to PCs, cameras, retail and factory systems, and on-site servers, from suppliers such as Hailo and Axelera AI; AI in smart glasses and XR, where Qualcomm reported having 30 XR designs either shipping or in development during fiscal 2025 and Meta launched its Ray-Ban Display glasses in September 2025; edge infrastructure and telecom AI; and developer ecosystems, illustrated by Qualcomm's acquisition of Arduino and its community of 30 million users. Key trends include small language models running locally, heterogeneous SoCs that devote more die area to the NPU, integrated physical AI platforms for robots such as NVIDIA's Jetson Thor, and the spread of RISC-V and chiplet-based designs.
The chip type analysis spans SoCs with integrated NPUs, standalone accelerators, automotive AI processors, edge GPUs & AI modules, FPGAs, and AI-enabled MCUs. SoCs with integrated NPUs are expected to hold the largest share in 2026, as they ship in billions of smartphones and hundreds of millions of PCs each year, NPU performance rises with every product generation, and platforms from Apple, Qualcomm, MediaTek, Samsung, and Google make on-device AI a standard feature. Standalone edge AI accelerators are expected to grow at the highest CAGR, as PCs, cameras, factory systems, and edge servers running generative AI need more performance than an integrated NPU can supply, served by products such as Hailo's Hailo-10H.
The application analysis covers smartphones, PCs, wearables & XR, smart cameras, automotive (ADAS & automated driving and intelligent cockpit), robotics & drones (including humanoid robots), industrial automation, smart home & IoT, and edge servers and gateways. Smartphones are expected to account for the largest share in 2026, reflecting their very high shipment volumes, the move of generative AI assistants into flagship and mid-range models, and the growing AI-attributable value of each mobile SoC. Robotics & drones are expected to grow at the highest CAGR, supported by robot computing platforms such as NVIDIA's Jetson Thor, launched in 2025 for humanoids and other machines, heavy investment in humanoid developers, and edge AI SoCs from suppliers such as Ambarella for drones and autonomous machines.
By performance class, the report covers chips below 5 TOPS, from 5 to 50 TOPS, and above 50 TOPS. The 5 to 50 TOPS class is expected to hold the largest share in 2026, since it covers most smartphones, PCs built to the 40 TOPS Copilot+ requirement, and smart cameras, which together make up the bulk of edge AI unit demand. The above 50 TOPS class is expected to grow at the highest CAGR, as automated driving, robotics, and edge generative AI need much higher compute.
By process node, the study covers 5 nm & below, 6-16 nm, and above 16 nm. Leading-edge nodes of 5 nm and below are used for flagship phone and PC processors and advanced automotive and AI SoCs, where performance per watt matters most, and Ambarella's move to 5-nanometer AI SoCs shows how specialist suppliers are also adopting them. Mature nodes remain important for microcontrollers, industrial chips, and cost-sensitive IoT devices. The report assesses demand across the three node groups without singling out the largest or fastest-growing one.
The end-use analysis covers consumer electronics and automotive, along with industrial & manufacturing, security & surveillance, healthcare, retail, and telecommunications. Consumer electronics is expected to account for the largest share in 2026, because smartphones, PCs, and wearables carry NPUs in very large volumes and device makers compete on AI features. Automotive is expected to grow at the highest CAGR, as driver assistance and cockpit systems need rising AI compute per vehicle and design wins secured in earlier years move into production.
Asia-Pacific is expected to be the largest regional market in 2026, as most of the world's phones, PCs, cameras, drones, electric vehicles, and industrial equipment are built there and the region hosts major chip designers and foundries. China leads output of phones, electric vehicles, cameras, and drones, and domestic chipmakers such as HiSilicon, Rockchip, and Horizon Robotics, which went public in Hong Kong in October 2024, are growing. MediaTek and TSMC in Taiwan, Samsung in South Korea, and Renesas and Sony in Japan supply chips and components, and according to the International Federation of Robotics, China accounted for more than half of global industrial robot installations in 2024.
North America is expected to grow at the highest CAGR during the forecast period, backed by U.S. strength in AI PCs, self-driving technology, robotics, and chip design, with Qualcomm, Apple, NVIDIA, Intel, AMD, and Ambarella at the forefront and U.S. humanoid developers driving demand for physical AI platforms. Europe holds a significant share, as BMW, Mercedes-Benz, and Volkswagen buy automotive AI processors, NXP, Infineon, and STMicroelectronics supply automotive and industrial chips, and the European Commission's Chips Act aims to mobilize more than EUR 43 billion for the region's semiconductor sector. In the Middle East & Africa, Israel is a major design center, home to Hailo and Mobileye, while the UAE and Saudi Arabia use AI cameras in smart city projects. In Latin America, Brazil and Mexico are sizable device markets, and Mexican vehicle output adds automotive demand.
Participants range from mobile and PC processor leaders and automotive chip suppliers to robotics platform providers, accelerator startups, microcontroller makers, IP licensors, and Chinese domestic vendors. They compete on energy efficiency, NPU strength, memory handling, software and developer support, safety and security features, price, and long-term supply commitments. Leading players are raising NPU performance, building developer communities, winning automotive programs, moving into robotics, XR, and edge infrastructure, and enabling generative AI on devices. Notable moves include Qualcomm's purchase of Arduino and its Snapdragon Ride Pilot platform with BMW, NVIDIA's Jetson Thor launch, STMicroelectronics' STM32N6 microcontroller with an integrated NPU, and NXP's 2025 agreement to acquire edge AI accelerator developer Kinara.
Key companies profiled in the report include Qualcomm Incorporated (U.S.), Apple Inc. (U.S.), MediaTek Inc. (Taiwan), Samsung Electronics Co., Ltd. (South Korea), Google LLC (U.S.), Intel Corporation (U.S.), Advanced Micro Devices, Inc. (U.S.), NVIDIA Corporation (U.S.), Mobileye Global Inc. (Israel), Horizon Robotics (China), Ambarella, Inc. (U.S.), Hailo Technologies Ltd. (Israel), Axelera AI (Netherlands), Synaptics Incorporated (U.S.), NXP Semiconductors N.V. (Netherlands), STMicroelectronics N.V. (Switzerland), Infineon Technologies AG (Germany), Renesas Electronics Corporation (Japan), Texas Instruments Incorporated (U.S.), HiSilicon (Huawei) (China), Rockchip Electronics Co., Ltd. (China), and Arm Holdings plc (U.K.).
Edge AI chips run AI models on or close to the device that produces the data instead of in a central cloud data center. The study covers SoCs with integrated NPUs for phones, PCs, tablets, wearables, and XR devices; standalone accelerators in M.2, PCIe, and module formats; automotive AI processors for driver assistance, automated driving, and smart cockpits; edge GPUs and AI modules for robots and industrial machines; FPGAs; and AI-capable microcontrollers for battery-powered sensing. Standalone accelerators, automotive processors, and AI modules are counted at full value, while SoCs with integrated NPUs are counted only for the share of their value attributable to AI.
The market is moving from basic features such as camera enhancement and wake words to generative and physical AI, and Microsoft's Copilot+ PC specification, introduced in 2024, sets a new performance floor for laptops by calling for NPUs of at least 40 trillion operations per second. Supplier results show the shift: Qualcomm's automotive sales hit a record USD 3.96 billion in fiscal 2025, 36% higher than a year earlier, and its IoT sales grew 22% to USD 6.6 billion, while Ambarella, as reported in its fiscal 2026 results, earned 80% of its USD 390.7 million in revenue from edge AI SoCs. Robot adoption adds to demand, with 542,000 new industrial robots installed globally in 2024, according to the International Federation of Robotics.
Market Dynamics
On-device generative AI in phones and PCs is the largest growth driver, as device makers now run summarization, writing help, image editing, live translation, and assistants locally to improve speed, privacy, and offline use and to cut cloud costs; the first Copilot+ PCs shipped in June 2024, Apple Intelligence launched in October 2024, and Qualcomm introduced its Snapdragon X2 Elite and Snapdragon 8 Elite Gen 5 platforms in 2025. AI-based driver assistance and autonomy form a second driver, as navigation-on-autopilot, driver monitoring, and AI cockpits need powerful onboard processors, with NVIDIA's DRIVE Thor, Mobileye's EyeQ, Horizon Robotics' Journey, and Qualcomm's Snapdragon Ride Pilot, developed with BMW, competing for design wins. Physical AI in robots, drones, cameras, and machine vision adds demand, as these systems must sense and act in real time, often without reliable connectivity. Privacy rules such as GDPR and the EU AI Act, lower latency, and lower cloud bills also favor local processing, especially as the International Energy Agency projects data center power use to reach about 945 terawatt-hours by 2030.Growth is held back first by device constraints: phones, glasses, and wearables run on small batteries with passive cooling, so chips must deliver results within tight power and thermal budgets, and sustained AI workloads can force throttling. Larger models need more DRAM, and memory became far more expensive in late 2025, when Samsung was reported to have lifted prices on some memory chips by up to 60%; this may lead makers to hold back AI features in cheaper models. Each vendor's NPU has its own compiler, runtime, and tools, so developers must adapt models for every chip and operating system. Consumer devices are price-sensitive, and automotive processors must meet AEC-Q100 and ISO 26262, with design wins taking two to four years to reach production. Further challenges include fitting multi-billion-parameter models into limited memory, protecting on-device models and data under rules such as UN Regulation No. 155 and the EU Cyber Resilience Act, a shortage of edge AI engineers, and U.S. export controls combined with China's push for domestic suppliers.
The main opportunities lie in accelerators that bring generative AI to PCs, cameras, retail and factory systems, and on-site servers, from suppliers such as Hailo and Axelera AI; AI in smart glasses and XR, where Qualcomm reported having 30 XR designs either shipping or in development during fiscal 2025 and Meta launched its Ray-Ban Display glasses in September 2025; edge infrastructure and telecom AI; and developer ecosystems, illustrated by Qualcomm's acquisition of Arduino and its community of 30 million users. Key trends include small language models running locally, heterogeneous SoCs that devote more die area to the NPU, integrated physical AI platforms for robots such as NVIDIA's Jetson Thor, and the spread of RISC-V and chiplet-based designs.
Segment Analysis
The report analyzes the market by chip type, application, performance class, process node, end-use industry, and geography, helping stakeholders identify where AI silicon value is shifting.The chip type analysis spans SoCs with integrated NPUs, standalone accelerators, automotive AI processors, edge GPUs & AI modules, FPGAs, and AI-enabled MCUs. SoCs with integrated NPUs are expected to hold the largest share in 2026, as they ship in billions of smartphones and hundreds of millions of PCs each year, NPU performance rises with every product generation, and platforms from Apple, Qualcomm, MediaTek, Samsung, and Google make on-device AI a standard feature. Standalone edge AI accelerators are expected to grow at the highest CAGR, as PCs, cameras, factory systems, and edge servers running generative AI need more performance than an integrated NPU can supply, served by products such as Hailo's Hailo-10H.
The application analysis covers smartphones, PCs, wearables & XR, smart cameras, automotive (ADAS & automated driving and intelligent cockpit), robotics & drones (including humanoid robots), industrial automation, smart home & IoT, and edge servers and gateways. Smartphones are expected to account for the largest share in 2026, reflecting their very high shipment volumes, the move of generative AI assistants into flagship and mid-range models, and the growing AI-attributable value of each mobile SoC. Robotics & drones are expected to grow at the highest CAGR, supported by robot computing platforms such as NVIDIA's Jetson Thor, launched in 2025 for humanoids and other machines, heavy investment in humanoid developers, and edge AI SoCs from suppliers such as Ambarella for drones and autonomous machines.
By performance class, the report covers chips below 5 TOPS, from 5 to 50 TOPS, and above 50 TOPS. The 5 to 50 TOPS class is expected to hold the largest share in 2026, since it covers most smartphones, PCs built to the 40 TOPS Copilot+ requirement, and smart cameras, which together make up the bulk of edge AI unit demand. The above 50 TOPS class is expected to grow at the highest CAGR, as automated driving, robotics, and edge generative AI need much higher compute.
By process node, the study covers 5 nm & below, 6-16 nm, and above 16 nm. Leading-edge nodes of 5 nm and below are used for flagship phone and PC processors and advanced automotive and AI SoCs, where performance per watt matters most, and Ambarella's move to 5-nanometer AI SoCs shows how specialist suppliers are also adopting them. Mature nodes remain important for microcontrollers, industrial chips, and cost-sensitive IoT devices. The report assesses demand across the three node groups without singling out the largest or fastest-growing one.
The end-use analysis covers consumer electronics and automotive, along with industrial & manufacturing, security & surveillance, healthcare, retail, and telecommunications. Consumer electronics is expected to account for the largest share in 2026, because smartphones, PCs, and wearables carry NPUs in very large volumes and device makers compete on AI features. Automotive is expected to grow at the highest CAGR, as driver assistance and cockpit systems need rising AI compute per vehicle and design wins secured in earlier years move into production.
Regional Analysis
The report covers North America, Europe, Asia-Pacific, Latin America, and the Middle East & Africa, with analysis of 18 countries, the Nordic Countries, and five rest-of-region groups. The regional analysis considers device and vehicle manufacturing, chip design and foundry capacity, AI adoption, regulation, and trade policy.Asia-Pacific is expected to be the largest regional market in 2026, as most of the world's phones, PCs, cameras, drones, electric vehicles, and industrial equipment are built there and the region hosts major chip designers and foundries. China leads output of phones, electric vehicles, cameras, and drones, and domestic chipmakers such as HiSilicon, Rockchip, and Horizon Robotics, which went public in Hong Kong in October 2024, are growing. MediaTek and TSMC in Taiwan, Samsung in South Korea, and Renesas and Sony in Japan supply chips and components, and according to the International Federation of Robotics, China accounted for more than half of global industrial robot installations in 2024.
North America is expected to grow at the highest CAGR during the forecast period, backed by U.S. strength in AI PCs, self-driving technology, robotics, and chip design, with Qualcomm, Apple, NVIDIA, Intel, AMD, and Ambarella at the forefront and U.S. humanoid developers driving demand for physical AI platforms. Europe holds a significant share, as BMW, Mercedes-Benz, and Volkswagen buy automotive AI processors, NXP, Infineon, and STMicroelectronics supply automotive and industrial chips, and the European Commission's Chips Act aims to mobilize more than EUR 43 billion for the region's semiconductor sector. In the Middle East & Africa, Israel is a major design center, home to Hailo and Mobileye, while the UAE and Saudi Arabia use AI cameras in smart city projects. In Latin America, Brazil and Mexico are sizable device markets, and Mexican vehicle output adds automotive demand.
Competitive Landscape
The report reviews key growth strategies, competitive benchmarking, the competitive dashboard with industry leaders, market differentiators, vanguards, and emerging companies, and market share/rank analysis by key player for 2025, together with company profiles covering business overview, financial performance, product portfolio, strategic developments, and SWOT analysis.Participants range from mobile and PC processor leaders and automotive chip suppliers to robotics platform providers, accelerator startups, microcontroller makers, IP licensors, and Chinese domestic vendors. They compete on energy efficiency, NPU strength, memory handling, software and developer support, safety and security features, price, and long-term supply commitments. Leading players are raising NPU performance, building developer communities, winning automotive programs, moving into robotics, XR, and edge infrastructure, and enabling generative AI on devices. Notable moves include Qualcomm's purchase of Arduino and its Snapdragon Ride Pilot platform with BMW, NVIDIA's Jetson Thor launch, STMicroelectronics' STM32N6 microcontroller with an integrated NPU, and NXP's 2025 agreement to acquire edge AI accelerator developer Kinara.
Key companies profiled in the report include Qualcomm Incorporated (U.S.), Apple Inc. (U.S.), MediaTek Inc. (Taiwan), Samsung Electronics Co., Ltd. (South Korea), Google LLC (U.S.), Intel Corporation (U.S.), Advanced Micro Devices, Inc. (U.S.), NVIDIA Corporation (U.S.), Mobileye Global Inc. (Israel), Horizon Robotics (China), Ambarella, Inc. (U.S.), Hailo Technologies Ltd. (Israel), Axelera AI (Netherlands), Synaptics Incorporated (U.S.), NXP Semiconductors N.V. (Netherlands), STMicroelectronics N.V. (Switzerland), Infineon Technologies AG (Germany), Renesas Electronics Corporation (Japan), Texas Instruments Incorporated (U.S.), HiSilicon (Huawei) (China), Rockchip Electronics Co., Ltd. (China), and Arm Holdings plc (U.K.).
How This Report Helps
- Provides reliable market size estimates and long-term forecasts for the global edge AI chips market through 2036.
- Explains how on-device generative AI, driver assistance, and physical AI are raising AI silicon content per device.
- Assesses the impact of power limits, memory costs, software fragmentation, and export controls on chip demand.
- Identifies high-growth opportunities across chip type, application, performance class, process node, end-use industry, and region.
- Tracks emerging technologies such as small language models, edge generative AI accelerators, AI smart glasses, RISC-V, and chiplets.
- Benchmarks leading chipmakers on NPU performance, ecosystems, automotive design wins, and market position.
- Supports product roadmaps, design-win strategy, partnerships, and investment decisions for chipmakers, device OEMs, automakers, and investors.
Key Questions Answered
- How large is the global edge AI chips market today, and how will it grow through 2036?
- How much AI compute will phones, PCs, vehicles, robots, and cameras need, and how is AI content per device changing?
- What are the main drivers, restraints, opportunities, challenges, and trends shaping the market?
- How are rising memory prices and device power limits affecting on-device AI adoption?
- Which chip type, application, performance class, process node, and end-use industry segments will grow the fastest?
- How quickly will small language models and generative AI accelerators move to the edge?
- How are U.S. export controls and China's domestic substitution reshaping supplier positions?
- Which regions and countries offer the most attractive opportunities for edge AI chip suppliers?
- Who are the leading companies in the market, and what strategies are they using to compete?
Table of Contents
1. Introduction
2. Research Methodology
4. Market Overview
5. Edge AI Chips Market, by Chip Type
6. Edge AI Chips Market, by Application
7. Edge AI Chips Market, by Performance Class
8. Edge AI Chips Market, by Process Node
9. Edge AI Chips Market, by End-Use Industry
10. Edge AI Chips Market, by Geography
11. Competitive Landscape
12. Company Profiles
13. Appendix
LIST OF TABLES
LIST OF FIGURES
Companies Mentioned
- Qualcomm Incorporated
- Apple Inc.
- MediaTek Inc.
- Samsung Electronics Co., Ltd.
- Google LLC
- Intel Corporation
- Advanced Micro Devices, Inc.
- NVIDIA Corporation
- Mobileye Global Inc.
- Horizon Robotics
- Ambarella, Inc.
- Hailo Technologies Ltd.
- Axelera AI
- Synaptics Incorporated
- NXP Semiconductors N.V.
- STMicroelectronics N.V.
- Infineon Technologies AG
- Renesas Electronics Corporation
- Texas Instruments Incorporated
- HiSilicon
- Rockchip Electronics Co., Ltd.
- Arm Holdings plc
