The structure of physical AI investment centered on the AX sector, together with hypotheses on the benefits to individual companies as the technology moves into real-world implementation, has been laid out in connection with METI’s FY2027 budget request. The government aims to create sustained growth investment centered on AI and semiconductors, pointing to market opportunities through 2040.
The Ministry of Economy, Trade and Industry’s FY2027 budget request totals ¥7.7859 trillion, emphasizing a policy shift away from reliance on supplementary budgets and toward integrating large-scale growth investment from the initial stages. The biggest structural change is the newly established “Strong and Prosperous Japan” Investment Framework, which identifies AX as the top-priority growth driver. The framework totals approximately ¥1.9913 trillion and covers areas including AI, semiconductors and robotics. This is expected to enable planned implementation over multiple fiscal years and strengthen the framework for long-term fiscal support.
The technology scope of the “physical AI platform” and “AI robotics field deployment” is defined as encompassing autonomous intelligent systems that perform inference based on real-world sensor data and drive actuators. The development of vision-language-action (VLA) models integrating video, tactile and LiDAR data, in addition to LLMs, is considered essential under the multimodal foundation model development program. Attention is also focused on the NEDO-led FRONTia project. The application requirements call for real-time coordination with autonomous control units and adaptability to field environments.
The policy is expected to have vertically integrated effects spanning robotics, sensors, semiconductors, digital twins and systems integration. FANUC, Yaskawa Electric, Harmonic Drive Systems, Nabtesco, Renesas, Socionext, Hitachi, NTT DATA, PFN and Mujin are positioned as major potential beneficiaries. In the digital twin and systems integration sectors in particular, Lumada and autonomous data-integration platforms are expected to serve as key integration infrastructure.
The roadmap through 2040 envisions three stages of evolution. Phase 1 covers the establishment of foundation models and point deployments, Phase 2 moves to mass production and autonomous coordination across production lines, and Phase 3 transitions to social infrastructure and autonomous coexistence. The targets and timelines for automation will vary by industry, including manufacturing, logistics and nursing care, with the goal of creating an end-to-end unmanned logistics network.