East Asian Technology Intelligence
Japan & China tech news — translated, contextualized, and delivered often.
Free. Unsubscribe anytime.
3 Takeaways This Issue
- Nvidia’s expansion of sovereign AI partnerships with SoftBank and Sakura Internet in Japan shifts the country’s industrial focus from traditional hardware manufacturing to domestic data center infrastructure.
- China’s 2025 National Science and Technology Awards prioritized Star Market-listed semiconductor firms like AMEC and Piotech, targeting specific production bottlenecks to bypass Western lithography restrictions.
- The joint physical AI pilot by SoftBank and Yaskawa Electric demonstrated a model training process where industrial robots optimize their own assembly line movements over time, reducing initial programming hours by 40 percent.
Core Move
Beijing’s Top Tech Honors Reveal the Playbook for Subsidizing its Chokepoint Counter-Offensive #
Beijing’s National Science and Technology Awards show the Chinese government’s industrial policy. The awards signal exactly where China wants to replace imports and invest to close technology gaps. This system does not just recognize past work. It is a live blueprint to guide state resources, mainly in manufacturing and deep tech where Western firms lead.
Star Market companies like Oulin Biosciences and Leaderdrive won big awards for filling international gaps. This shows that Beijing uses its awards to highlight firms that help it become self-sufficient. The awards also signal provincial governments and state funds to send cash and contracts to these winners. Western awards celebrate market success or science, but China’s awards focus on national strategic needs.
SUPCON Technology won a top award for industrial automation. This sector is a major chokepoint, as German and Japanese specialists often control it. Beijing’s praise for SUPCON shows its fast push to make its own industrial control tools, from hardware to software. This move protects Chinese factories from global political pressure. It looks like Japan’s post-war industrial rebuild, but it applies to digital supply chains and stems from rivalry with the US.
Some Western observers think these awards show strong market-driven innovation, which is a mistake. These companies are technically skilled, but the state drives their work to break foreign bottlenecks. Their success comes from state demand and subsidies, not from open global markets. This makes them tough rivals at home, but their exports will depend on global politics.
Western firms in parts manufacturing and industrial automation will face tougher rivalry in China. They should also expect more rules and less market access as Beijing favors its own firms. To track this shift, you can watch SUPCON’s new contracts with state-owned firms. You can also follow local government research grants to award winners, and watch for new Chinese export controls on these tech goods.
🗾 Japan Radar #
What Japanese media is reporting that Western outlets miss
Japan is shifting AI from digital chatbots to physical industrial hardware and sovereign infrastructure to secure its supply-chain independence.
🗾 AI & Machine Learning
Anthropic Integrates ‘Claude Fable 5’ into Subscriptions: Max and Team Premium Plans Included
Anthropic announced on July 17 (US time) that its top-tier AI model, ‘Claude Fable 5,’ will become a standard feature in its Max and Team Premium paid plans starting July 20. Users on these plans can access Fable 5 for up to 50% of their usage limit without additional cost. For Pro and Team Standard plans, Fable 5 will be available via ‘usage credits’ charged separately from the monthly fee, with a one-time grant of $100 in credits. Anthropic’s shift from a temporary, credit-based offering to a tiered subscription model for Fable 5, as detailed in Japanese tech media, reflects a common pattern in the AI industry: initial cautious rollout followed by consolidation into established pricing. The demand for Fable 5 was ‘difficult to predict,’ according to Anthropic, which suggests that even top-tier LLM providers are still figuring out how to monetize their most advanced models while managing capacity. The local coverage highlights the user frustration during the extension period, which is a valuable signal about the practical challenges of deploying and pricing cutting-edge AI.
For Western readers: Western businesses using or evaluating advanced LLMs should expect similar capacity and pricing volatility from other providers, as even major players struggle to forecast demand and manage infrastructure costs for their most capable models. Assume that your AI provider’s roadmap for free trials and tiered access will shift, and build that into vendor risk assessments. 🗾 Robotics & Automation
SoftBank and Yaskawa Electric are collaborating on “Physical AI” to develop robots that improve their performance through reinforcement learning after purchase, rather than degrading over time. At SoftBank World 2026, SoftBank CTO Ryuji Wakikawa described this shift as a “major paradigm shift,” explaining that robots will now learn from real-world data, synthesize it with NVIDIA’s Cosmos, evaluate in Omniverse simulations, and then apply improvements to physical robots. The core of this partnership is an attempt to address a long-standing problem in industrial automation: how to make robots adaptable to real-world variability without constant human reprogramming. By integrating an automated learning loop, SoftBank and Yaskawa are trying to operationalize a vision where robot performance is a function of data and continuous iteration, not just initial engineering. This is a pragmatic, execution-focused approach consistent with Japan’s industrial strengths.
For Western readers: Western automation companies should note that this model of continuous robot improvement, leveraging cloud AI infrastructure and simulation, could significantly alter the total cost of ownership and flexibility expectations for industrial robots, potentially shifting competitive dynamics in high-mix, low-volume manufacturing. Cross-Regional Analysis · AI & Machine Learning2 STORIES
Nvidia Powers Japan’s Sovereign AI Push as Tech Giants Lead Industrial Revival Nvidia CEO Jensen Huang’s high-profile visit to Tokyo underscores Japan’s aggressive, state-backed drive to build its own ‘sovereign AI‘ infrastructure, highlighted by a new 44-company consortium including SoftBank, Sony, and Honda. Supported by massive government procurement of Nvidia’s next-generation Rubin chips, this national initiative aims to leverage proprietary domestic data to revitalize Japan’s world-class manufacturing and robotics sectors. Collectively, these developments signal Japan’s strategic pivot toward reducing reliance on foreign software models while securing its place in the global hardware supply chain.
Why it matters: The repeated visits by US tech leaders aren’t just sales trips; they reflect Japan’s willingness to spend serious money on AI and chips, as well as its strategic importance in the US-led effort to contain China’s technological ambitions. While Western media tends to focus on the ‘AI race’ and the competitive aspects, Japan’s domestic framing prioritizes economic security and industrial revitalization, aiming to rebuild its manufacturing base, not just consume foreign tech.
For Western readers: If you are a Western enterprise AI provider or a semiconductor firm, understand that Japan’s ‘sovereign AI’ push means your Japanese partners will prioritize domestic integration and IP transfer over simply buying off-the-shelf solutions, so structure your joint ventures accordingly.
🇨🇳 China Watch #
China’s technology moves, framed for Western readers
China is bypassing consumer AI hype to aggressively localize the industrial, semiconductor, and software supply chains driving national self-reliance.
AI & Machine Learning
Hangzhou Xunsun Intelligence CEO: AI Software Factory Redefines Software Production From Requirements to Running Systems Hangzhou Xunsun Intelligence, a Chinese AI software factory, announced its platform can automate software development from initial requirements to fully functional systems using large language models (LLMs). CEO Chen Yin emphasized a ‘human-machine co-creation‘ model, suggesting AI handles repetitive coding while humans oversee architecture and critical design decisions. The company aims to enhance software development efficiency and lower costs within the Chinese market. Xunsun’s “AI software factory” claims indicate China is not just a consumer of AI models but is actively pursuing their application to industrial processes like software development. Their framing around ‘human-machine co-creation’ is a pragmatic approach that acknowledges the current limitations of fully autonomous AI, prioritizing practical deployment over theoretical breakthroughs. It’s an incremental step towards a more automated engineering workflow, not a revolution.
For Western readers: Western software firms should evaluate their internal development processes for areas where similar AI tooling could incrementally reduce manual effort and accelerate delivery, rather than waiting for a fully autonomous AI developer that won’t appear soon. Semiconductors & Hardware
Chinese manufacturers are advancing in batch verification of high-purity PFA (perfluoroalkoxy alkane), a critical material for advanced semiconductor manufacturing. This development helps China localize key components for its domestic semiconductor process nodes. The move aims to reduce reliance on foreign suppliers for crucial materials in chip production. The focus here is on securing the upstream chemical supply chain, not just equipment or IP. Chinese firms are proving they can develop and qualify these specialized materials, which means export controls targeting end-products or high-level equipment will continue to face circumvention attempts through domestic material substitution. This is an unglamorous but critical piece of the puzzle for Beijing’s tech ambitions.
For Western readers: Western semiconductor material suppliers should expect increasing competitive pressure in China, particularly in high-purity chemical segments, as Chinese alternatives gain qualified status within domestic fabs. Semiconductors & Hardware
TCL CSOT Showcases Limitless Breakthrough XR Displays at SID Display Week 2026
Chinese display maker TCL CSOT presented its latest Extended Reality (XR) display technologies at SID Display Week 2026, including micro-LED and micro-OLED solutions tailored for AR/VR applications. This showcases their push into advanced display segments critical for next-generation immersive devices. Chinese display manufacturers like TCL CSOT are not just volume producers anymore; they are moving up the value chain into advanced display technologies. This shift impacts the competitive landscape for Japanese and South Korean display firms, pushing them to innovate faster or risk market share loss in emerging XR sectors.
For Western readers: Western hardware companies developing AR/VR devices should anticipate increased sourcing options and potentially aggressive pricing from Chinese suppliers like TCL CSOT for cutting-edge display components, but also consider supply chain resilience given geopolitical tensions. AI & Machine Learning
ByteDance Denies Autonomous Driving Ambitions, Focuses on Frontier AI ByteDance has publicly denied media reports claiming it plans to enter the autonomous driving sector, specifically driverless logistics, under its cloud platform, Volcano Engine. The company clarified it is conducting early-stage research in ‘Physical AI’ and frontier AI large model technologies, but not for smart driving. ByteDance is pushing back on the prevailing narrative that every major Chinese tech player must enter autonomous driving. Their ‘Physical AI’ focus, distinct from full-stack vehicle development, could position them as a foundational model provider for other AV companies or logistics operators, rather than a direct competitor.
For Western readers: If you are investing in or partnering with Chinese autonomous vehicle startups, ByteDance is unlikely to emerge as a direct competitor for vehicle platforms or logistics services; instead, watch for them to offer AI model capabilities to existing players. Startups & Funding
Marvel-Tech raises over RMB500 million for gas turbine manufacturing in China Shanghai-based Marvel-Tech, a developer of gas turbines and industrial waste-heat recovery systems, has secured over RMB500 million across two financing rounds. These funds will support building a manufacturing and assembly center, as well as a high-power unit performance testing center for their gas turbines in China. The investment enables Marvel-Tech to build 1GW of manufacturing and delivery capacity for advanced gas turbines that can run on green fuels. This is a practical step towards China’s energy independence and supports the rapid expansion of domestic AI data centers with locally produced power infrastructure.
For Western readers: Western suppliers of gas turbines and industrial power generation equipment should expect increasing competition in the Chinese market, particularly for data center and industrial applications, as domestic alternatives mature and scale up.
🔺 The Triangle #
Where US, Japan, and China technology interests intersect
The battle for AI supremacy is shifting from cloud-based frontier models to hardware-level control of the global silicon supply chain.
Semiconductors & Hardware
TSMC Boosts 2026 Expansion Budget, Adds $100B to U.S. Investment TSMC has increased its 2026 capital expenditure budget significantly, with a substantial portion of this new investment directed towards its fabrication facilities in the United States. This expansion signals TSMC’s commitment to scaling its advanced manufacturing capabilities beyond Taiwan, particularly in response to geopolitical pressures and customer demands for geographical diversification. TSMC’s increased investment outside Taiwan, particularly in the US, reflects an acceleration of the ‘de-Taiwanization’ of advanced chip manufacturing that has been discussed for years. This isn’t just about raw capacity; it’s about re-engineering critical supply chains under geopolitical pressure, which impacts everyone in the chip ecosystem, including East Asian equipment and materials suppliers.
For Western readers: If your business relies on advanced semiconductors, anticipate that the shift in TSMC’s capital expenditure towards US facilities will lead to potentially higher costs and slower ramp-ups for certain nodes outside Taiwan, despite the overall increase in capacity. Semiconductors & Hardware
Micron has secured long-term memory supply agreements with seven automotive industry suppliers, including East Asian firms DENSO, Astemo (a Hitachi Astemo joint venture), and Hyundai Mobis. These agreements are aimed at providing stable access to advanced memory and storage as vehicles become more AI-enabled, improving production planning and collaboration for future platforms. The deals signal a concerted effort by leading Japanese and Korean automotive suppliers to lock in critical memory supply, prioritizing stability and predictability over spot market pricing. It’s a pragmatic response to a volatile semiconductor market, reflecting a deeper understanding of supply chain fragility, which is a major concern for East Asian manufacturers that are often highly exposed.
For Western readers: Western automotive suppliers and OEMs should recognize that key East Asian competitors are securing long-term memory access, potentially giving them a supply advantage in an increasingly competitive, AI-driven automotive sector. Do not assume ‘just-in-time’ procurement will remain viable for critical components like advanced memory. Semiconductors & Hardware
Infineon, LS ELECTRIC Partner on DC Power Infrastructure for AI Data Centers Infineon Technologies AG and South Korea’s LS ELECTRIC have signed an MoU to co-develop high-efficiency direct current (DC) power solutions for AI data centers and next-generation power grids. This partnership aims to address the escalating power demands driven by AI workloads and distributed power networks. The collaboration combines Infineon’s semiconductor expertise with LS ELECTRIC’s system integration capabilities. The focus on DC power infrastructure, solid-state transformers (SSTs), and solid-state circuit breakers (SSCBs) points to a more fundamental re-architecting of power delivery, not just incremental efficiency gains. LS ELECTRIC’s goal to become a ‘total solutions provider’ in the global AI data center market shows how South Korean firms are aiming to move up the value chain beyond components into integrated systems. This is more than just an efficiency play; it’s about control over core infrastructure layers.
For Western readers: Western data center operators should assume new infrastructure standards and supply chains for AI power will increasingly originate from East Asian partnerships rather than being solely driven by Silicon Valley, creating new integration challenges for global deployments. Semiconductors & Hardware
Edge AI Hits 25% of Smartwatch Shipments in 1Q 2026 Edge AI-capable smartwatch shipments globally grew 70% year-on-year in Q1 2026, reaching 25% market penetration, driven by demand for advanced health monitoring. While Apple dominates this segment with approximately 90% of shipments, Chinese brands like Huawei are upgrading their silicon to compete in the smartwatch AI market. This growth is largely fueled by on-device processing of health data, reducing reliance on cloud services. The focus on health-driven edge AI in smartwatches, particularly for conditions like sleep apnea and diabetes, moves beyond simple tracking into diagnostic support. Huawei’s efforts with its Kirin W80 chip show a clear strategy to match Apple’s on-device AI capabilities, but the reality is that Apple still controls nearly all of the current market in this category. For now, it’s more about China trying to catch up in a high-value niche rather than leading it.
For Western readers: Western chipset manufacturers should recognize the increasing pressure to integrate low-power neural accelerators into their wearables platforms to remain competitive, as Chinese players like Huawei are explicitly targeting this high-margin health application space with vertically integrated solutions. Semiconductors & Hardware
Tower Semiconductor to Invest $3 Billion in Japan SiPho and SiGe Capacity 📊 Featured Chart
Source: Tower Semiconductor
Tower Semiconductor, an Israeli foundry, plans a $3 billion expansion of SiPho, SiGe, and packaging capacity in Japan, with the Japanese government contributing $1 billion. This investment includes adding capacity at the former Panasonic fab in Arai and constructing a new 300mm fab adjacent to its Uozu Fab 7. The investment targets Silicon Photonics and Silicon Germanium, foundational technologies for high-speed optical interconnects in AI data centers. Tower’s focus on these specialized processes, rather than general-purpose logic, aligns with Japan’s strategy of carving out niches in areas where it can leverage existing materials and precision manufacturing expertise. The 2028-2029 revenue and profit projections are ambitious, suggesting Tower expects rapid scaling of these niche technologies.
For Western readers: If you are involved in AI data center architecture or optical component sourcing, understand that Japan is making a serious play for SiPho and SiGe manufacturing, backed by significant government capital, which could diversify your supply options beyond traditional players by late 2027. 🧩 Pattern This Issue
China: Beijing awards Star Market firms to bypass materials and software chokepointsJapan: Nvidia anchors METI-backed sovereign AI plans to revitalize industrial roboticsChina: Suppliers advance batch verification of high-purity PFA for local semiconductor nodes
East Asia is systematically decoupling its industrial AI and semiconductor supply chains from Western vulnerabilities, meaning Western tech firms face rapidly closing market access as Japan and China build self-reliant hardware-software ecosystems.
AsiaAI.FYI · Written by Dick Weisinger ·