22 Carefully Selected Edge AI Semiconductor Stocks | Where Are the True Leaders Among Japanese Stocks That Can Compete Globally in the “AI at the Point of Use” Coming After the Cloud AI Bubble?

Looking back at 2026, the AI market showed clear signs of losing momentum, while the center of gravity in AI continued to shift from the cloud toward the edge. AI at the point of use is expected to involve processing within devices that meets requirements for low latency, low cost and data confidentiality, making it a practical option for applications such as on-site cameras and robots.

Renesas Electronics is one of Japan’s leading edge AI semiconductor companies. Its RZ/V series, which incorporates DRP-AI, covers 0.5 TOPS to 80 TOPS. DRP-AI3 uses design innovations to improve performance per unit of power, while the mid-range RZ/V2N delivers high performance in a compact package. Although the company fell into the red in fiscal 2025, revenue and profit are expected to recover in fiscal 2026, and the company has disclosed that it expects to recognize a gain on the transfer of SiTime.

Socionext, meanwhile, focuses primarily on custom SoCs, with its order backlog for advanced semiconductors reaching a record level. Its strength lies in designing dedicated chips around the power and cost constraints of each device. For the fiscal year ending March 2026, revenue is expected to increase, but operating profit and net income are expected to remain in the red. The ramp-up of mass production is expected to mark the point at which financial performance begins to recover. At the same time, attention is warranted regarding the risk of delays in large custom projects and the risks associated with dependence on major customers.

This article breaks down the entire edge AI value chain, from microcontrollers, sensors, power and packaging to trading companies and software, into 22 stocks, and examines the phase in which AI at the point of use moves toward mass production.