ABI Research Projects TinyML AI Chipsets to Reach 4.1B by 2031

ABI Research said it expects TinyML AI chipset shipments excluding personal and work devices to rise through 2031. The forecast points to increased embedded AI deployment and changes in how vendors plan for inference demand across edge and cloud environments.

In the company’s projections, the TinyML segment grows at a 37% compound annual growth rate to surpass 4.1 billion units, with related revenue above US$7.8 billion. ABI Research tied the outlook to the shift from experimentation to scaled deployment, with industrial IoT and other far-edge environments referenced in the report.

ABI Research’s market data indicated that the TinyML segment remained led by MCUs through the decade, while NPUs posted the fastest growth at a 90% CAGR. It also projected edge AI growth of 17% in Europe and 16% in North America through 2031, alongside 18% in Asia-Pacific, where shipments were projected to exceed 721 million by the decade’s end. For cloud AI, it cited training demand as cluster sizes increased and inference growth associated with token generation, multimodal generative output, and reasoning models and agentic AI workloads.

ABI Research said the report found that growth differed across device categories, with medium- to low-priced smartphones facing near-term pressure from higher DRAM prices. It reported that manufacturers including Xiaomi, vivo, and OPPO reduced 2026 sales forecasts, while premium smartphones were described as more resilient. It also cited heterogeneous SoC architectures gaining share, with Qualcomm, MediaTek, Apple, AMD, and Intel optimizing AI workloads across CPUs, GPUs, and NPUs for greater efficiency and broader framework support.

“The AI chipset market is fragmented but maturing at the same time,” said Paul Schell, Senior Analyst at ABI Research. “TinyML is gaining real traction as enterprises push intelligence closer to sensors and endpoints, while cloud and premium device segments continue to absorb the most advanced AI workloads. Vendors that can balance performance, power efficiency, and developer accessibility will be best positioned to win.” “Over the next several years, competitive advantage in AI semiconductors will come from architectural fit, not just raw compute,” Schell said. “The strongest suppliers will be those that align silicon roadmaps with deployment realities, whether that means ultra-low-power inference at the far edge, premium on-device AI experiences, or scalable cloud platforms for training and orchestration. That shift will create new openings for both established chipset leaders and specialist challengers.”

Provided by Globe Newswire on behalf of ABI Research. Click to read original content.