# Former Digital Minister Argues Japan Can Compete in AI > Source: > Published: 2026-05-31 06:18:59.387450+00:00 # Former Digital Minister Argues Japan Can Compete in AI On May 31, former Digital Minister Masaaki Taira, who oversees cybersecurity and artificial intelligence policy within the ruling Liberal Democratic Party, told BS TV Tokyo's "NIKKEI Sunday Salon" that Japan still has opportunities to compete in the global AI sector, according to News On Japan. Taira warned that "trying to challenge the most advanced large-scale AI models head-on would be a path to defeat," and urged focus on areas of national strength. He pointed to **robotics/physical AI** and **highly specialized models trained on proprietary, nonpublic data** in manufacturing, healthcare, and nursing care as promising domains. He also noted Japan retains strengths in **semiconductor manufacturing equipment**, which supports the AI supply chain. ### What happened Reporting by News On Japan covers remarks made on the May 31 edition of BS TV Tokyo's "NIKKEI Sunday Salon" by former Digital Minister **Masaaki Taira**, who oversees cybersecurity and artificial intelligence policy within the ruling Liberal Democratic Party. Taira argued that Japan should avoid direct competition with the creators of cutting-edge large language models from U.S. and Chinese technology giants. "Trying to challenge the most advanced large-scale AI models head-on would be a path to defeat," Taira said. He identified two opportunity areas: **physical AI**, including robotics and AI foundation models for robots, and **highly specialized AI systems** built on proprietary data not available on the public internet. "The internet's data has largely been consumed already," he said, and he pointed to manufacturing, healthcare, and nursing care as sectors holding valuable nonpublic datasets. Reporting also notes Taira acknowledged Japan lagged the United States during the 1990s internet revolution and emphasized existing strengths in **semiconductor manufacturing equipment**. ### Editorial analysis - technical context Companies and research groups pursuing robot-focused AI typically combine control systems, perception models, and task-specific foundation models rather than competing with general-purpose large language models. Industry-pattern observations: developing robotics-oriented foundation models usually requires diverse multimodal datasets, simulation-to-real transfer pipelines, and close OEM partnerships. For highly specialized vertical models, success often follows organizations that can leverage proprietary operational data, controlled labeling processes, and domain-specific evaluation metrics rather than exclusively relying on web-scale pretraining corpora. ### Industry context Industry observers note that national ecosystems with strong manufacturing bases and specialized datasets can create differentiated AI propositions without matching frontier generalist models. Observed patterns in comparable markets show that focusing on verticalization, regulatory alignment for sensitive data (healthcare, eldercare), and integration with physical hardware often raises engineering complexity but also creates higher entry barriers for generalist competitors. ### What to watch For practitioners and observers, relevant indicators will include announcements of alliances between Japanese OEMs and AI vendors, releases of robot-specific model checkpoints or benchmarks, new data-governance frameworks enabling use of proprietary datasets, and investments in semiconductor manufacturing and sensor supply chains. Reporting to date does not include a government policy rollout tied to Taira's remarks; News On Japan does not report a formal policy announcement accompanying his commentary. ## Scoring Rationale Remarks from a former Digital Minister highlight strategic options relevant to national AI pathways and industry practitioners. The piece is notable for framing practical domains for competitiveness but does not announce policy changes or technical releases. Practice interview problems based on real data 1,500+ SQL & Python problems across 15 industry datasets — the exact type of data you work with. [Try 250 free problems](/problems)