This week’s signal
U.S. President Trump Signs Executive Order on AI Security, Allowing Government Inspection of Cutting-Edge Models up to 30 Days Before Public Release
President Trump’s latest executive order creates a voluntary framework for AI model inspection. This indicates the U.S. government sees AI security as a shared responsibility. That view will increasingly shape market access and strategic partnerships. For Western firms, particularly those with East Asian interests, this isn’t just about compliance; it’s about competitive positioning in a changing international environment.
The immediate reaction from Tokyo will likely be cautious, but attentive. Japan’s regulatory approach to AI differs from the U.S. and China. It has focused on ethical guidelines and industry-led standards, not top-down security review mandates. Japanese technology giants often operate in a consensus-driven business environment. They typically favor self-regulation and close collaboration with government through industry associations.
However, this executive order introduces a new dynamic. Even with its “voluntary” clause, the order introduces a more direct, state-centric security prerogative. This approach is less familiar in their domestic context. The U.S. government is willing to inspect cutting-edge models up to 30 days pre-release, even if not mandatory. This establishes a new expectation. This indicates that the line between commercial innovation and national security is increasingly blurred, especially for frontier AI.
This development reflects broader East Asian trends. National security concerns are increasingly overriding purely commercial interests in technology development. In China, the integration of AI into state surveillance and industrial policy has long been explicit. Japan’s approach is fundamentally different, but it cannot ignore the security implications from its key ally. The underlying U.S. concern is balancing innovation with the potential misuse of powerful AI models. This issue has been seen in past debates around hypothetical models like Anthropic’s ‘Mythos’. It will now resonate in Japan’s own national security dialogues.
The next question is how Japanese firms will adapt. This applies particularly to firms integrating AI offerings into global supply chains. It also concerns firms partnering with U.S. companies. Will they proactively engage with this “voluntary” framework to demonstrate trustworthiness? Will they gain a competitive advantage? Or will they view it as an implicit barrier? This could lead them to deepen collaborations with European partners who may adopt a different regulatory approach. The choices made in Tokyo will reveal much about the future alignment of AI innovation and security across the Pacific.
🗾 Japan Radar
🗾 AI & Machine Learning
Microsoft announces its first in-house inference model “MAI-Thinking-1”, trained from scratch without distillation
At Build 2026, Microsoft unveiled MAI-Thinking-1. This is its first proprietary inference model, developed by the Microsoft AI Superintelligence team. This 35-billion parameter text model reportedly outperforms Anthropic’s Claude Sonnet 4.6 in general quality. It also matches Claude Opus 4.6 on coding benchmarks such as SWE-bench Pro.
Its ‘zero-distillation’ approach trains the model using commercially licensed, clean data. The model is optimized for Microsoft’s Maia 200 AI accelerator. This hardware reportedly offers 1.4 times the power efficiency of NVIDIA’s GB200. This shows Microsoft is becoming more independent and ambitious in developing core AI models. It could reduce Microsoft’s dependence on partners like OpenAI. This will also increase competition with other large AI companies.
The ‘zero-distillation’ and clean data approach directly answers company worries about data origin. It also addresses how reliable models are. This makes MAI-Thinking-1 a trustworthy choice for businesses. It also competes with NVIDIA’s hardware efficiency by using Microsoft’s own custom silicon.
🗾 AI & Machine Learning
Microsoft Announces Android-Based AI Agent Platform ‘Solara,’ Unveils Snapdragon-Powered Badge-Type Device
Microsoft unveiled ‘Project Solara’ at Build 2026. This new AI agent-first platform is built on an Android Open Source Project (AOSP) base. Its goal is to shift computing from app-centric to agent-centric interactions. The company showcased concept reference devices. Qualcomm helped develop some, like a wearable badge. MediaTek developed others, like a desk-based companion device. Solara is designed for enterprise use. It integrates with Microsoft services like Intune and Entra ID. Major companies will pilot it in the coming months.
This is a major strategic change for Microsoft. The company is leveraging Android for its next-gen AI devices. It also expands its enterprise AI ecosystem beyond Windows. This move intensifies competition with Google and Apple in the AI device space. This shows the growing importance of purpose-built edge AI hardware. It also strengthens Microsoft’s partnerships with Qualcomm and MediaTek. This signals a broader industry move towards specialized AI agents and devices.
🗾 AI & Machine Learning
Advantech Pushes ‘WEDA’ to Forefront, Reduces Edge AI Model Development Time by 86%
Advantech demonstrated its WEDA (WISE-Edge Developer Architecture) solution at COMPUTEX Taipei 2026. This system integrates edge AI development, deployment, and operation using Advantech’s hardware. WEDA is provided free with Advantech products. It leverages NVIDIA’s AI agent platform NemoClaw to reduce development time for edge AI models. Advantech showcased a complex demo involving real-time visitor tracking and VLM/LLM integration with physical devices.
This shows the intensifying competition in industrial edge AI. Advantech, a major Taiwanese hardware provider, is deeply integrating NVIDIA’s ecosystem. For Western businesses, this means innovation in industrial AI deployment is speeding up. Integrated hardware-software solutions are becoming more important. This could affect supply chain decisions and competitive strategies in manufacturing and automation.
🗾 Semiconductors & Hardware
TSMC Confident in Sustained Growth from Expanding AI Utilization; Continues Business with Tokyo Electron After Shareholder Meeting
TSMC CEO C.C. Wei predicts strong company growth for the next few years. At its annual shareholder meeting, he pointed to increased AI utilization as a key driver for its advanced technology and manufacturing value. Wei also confirmed that TSMC’s second factory in Kumamoto, Japan, is proceeding as planned. This plant will produce 3nm-equivalent advanced semiconductors, and there is potential for acceleration.
TSMC will continue doing business with Tokyo Electron, despite a Taiwanese subsidiary of the latter being convicted in a trade secret case involving TSMC. Wei emphasized the incident was an employee’s error, not a company policy violation.
This decision and TSMC’s expansion plans carry implications for Western businesses. TSMC plays a critical role in the global AI supply chain, and it is committed to expanding advanced manufacturing capacity, including in Japan. The choice to maintain relations with Tokyo Electron, a key Japanese semiconductor equipment supplier, demonstrates the deep interdependence in the East Asian semiconductor ecosystem. It also reflects the strategic importance of these supply relationships, especially amidst intensifying US-China tech competition.
🗾 Enterprise & Cloud
Autodesk Rolls Out AI Assistant Features for Major Products, Including MCP for Fusion
Autodesk announced the tech preview of ‘Autodesk Assistant’ for key products like Fusion, Inventor, Moldflow, and Vault. This aims to integrate AI into design and manufacturing workflows. The AI assistant is designed to understand 3D design data and operational context. It goes beyond simple chat to actively support and automate tasks.
Additionally, Autodesk released Model Context Protocol (MCP) for Fusion. This protocol enables external AI integration and customized workflows, and it includes a partnership with Anthropic’s Claude for Fusion.
This move by a major US CAD/CAM software provider shows the global trend of AI integration into specialized enterprise software. This integration is a critical factor for boosting productivity and addressing labor shortages in manufacturing. Such labor shortages are a concern shared across developed economies, including Japan. The emphasis on agent-based AI and open protocols for external AI integration reflects a strategic change. These solutions are more flexible, context-aware, and ecosystem-friendly. They could also set industry standards.
🇨🇳 China Watch
AI & Machine Learning
Alibaba’s Qwen App Opens Platform to Third-Party AI Agents, Onboarding Major Chinese Brands like Luckin Coffee and Mixue
Alibaba-backed Qwen App is evolving from a general-purpose AI assistant into a platform for third-party AI Agents and Skills. This is an important development in China’s AI ecosystem. Major Chinese consumer brands are among the first to integrate branded AI agents. These include Luckin Coffee, Mixue, and China Eastern Airlines. Users can complete tasks like food orders and travel arrangements via natural language. The goal is to deepen Qwen’s integration into daily life, using Alibaba’s extensive service ecosystem. This development suggests Alibaba’s strategy to embed its AI capabilities into China’s large consumer market. It uses its existing e-commerce and service infrastructure.
For Western companies, this shows the unique platform-centric approach to AI adoption that Chinese tech giants are taking. This differs from more fragmented Western models. It also reveals the fierce competition for AI dominance within China.
Startups & Funding
Chinese AI Startup DeepSeek in Talks to Raise $7 Billion from Tencent, CATL, and Other Major Investors
Chinese AI startup DeepSeek is reportedly raising $7 billion in its first external funding round. Tech giants Tencent and battery leader CATL have committed funds. The National AI Industry Investment Fund may also be involved. This substantial investment round would value DeepSeek at up to $56 billion. This shows investors have confidence in China’s rapidly expanding AI sector.
This funding shows China is aggressively pushing to build domestic AI champions. This aims to reduce reliance on Western technology amidst intensifying US-China tech competition. State-backed funds and major industrial players like CATL are participating. This points to a national strategy to bolster foundational AI capabilities.
AI & Machine Learning
xAI Launches Global Recruitment Drive for Chinese AI Tutors
xAI is actively recruiting Chinese AI tutors globally to enhance Grok’s Chinese language capabilities. These include proficiency in Mandarin, regional dialects, and mixed Chinese-English expressions. This shows xAI’s strategic focus on the Chinese market. It is an important part of its global growth and multilingual AI development. The positions are open to remote workers worldwide, with competitive compensation for US-based hires.
This recruitment drive shows the intense competition for the vast Chinese-speaking market. Local language and cultural nuance are important for AI models to succeed there. This also shows the continuing trend of Western tech companies expanding their presence in East Asia. This happens despite potential geopolitical and data localization complexities.
Robotics & Automation
Unitree IPO Approved, Meituan-backed Group Emerges as Top Shareholder
Chinese robotics firm Unitree gained one of the fastest IPO approvals on Shanghai’s STAR Market. This shows strong domestic capital market support for advanced manufacturing. The company shipped over 5,500 humanoid robots. It plans to raise $620 million. Its significant shareholder base includes Meituan, Sequoia China, Tencent, Alibaba, and Ant Group.
This fast listing shows China’s strategic effort to develop domestic companies in fields like robotics. China is using its own capital markets and major tech companies to build industrial strength. The strong backing from Chinese tech companies helps build a national system. This system aims to compete globally in technology. It may also lessen reliance on Western tech and investment.
AI & Machine Learning
ByteDance’s Doubao to Launch Paid Plans in Late June, Link with Douyin E-commerce Push
ByteDance’s AI assistant, Doubao, will introduce tiered paid subscription plans in China by late June. The paid plans will offer advanced productivity features like presentation and video generation. Basic chatbot services will remain free. This move helps one of China’s leading tech giants monetize its AI capabilities. It also diversifies revenue streams.
Doubao plans to integrate with Douyin’s e-commerce push in Q3. The goal is to generate traffic and support ByteDance’s core business. This will happen in the competitive Chinese market.
This development shows Chinese tech firms are working to monetize their large language models. They are also deeply embedding AI into their core platforms, especially e-commerce. This is happening in a highly competitive domestic market.
It also reflects China’s broader effort to build a self-reliant and commercially viable AI ecosystem. This strengthens its place in the global AI race.
🔺 The Triangle
Cross-Regional Analysis
COMPUTEX 2026 Highlights Taiwan’s Pivotal Role in Global AI Industry Transformation
COMPUTEX 2026, themed ‘AI Together,’ shows Taiwan’s central position in the global AI supply chain. The event features 1,500 exhibitors from 33 countries and regions. The event focuses on AI’s shift from computing power to real-world deployment. This includes applications in AI servers, edge computing, robotics, and smart mobility. Taiwan’s comprehensive ICT supply chain supports its role as a key innovation hub.
Taiwan’s robust ICT ecosystem and its ability to integrate global technologies are important for East Asian technology. The island navigates competition and collaboration with China. It also serves as a key semiconductor and AI hardware supplier to both the US and Japan. Taiwan’s role in shaping AI deployment strategies affects regional technological leadership and supply chain resilience.
Semiconductors & Hardware
TWSE Highlights Taiwan’s Semiconductor and AI Leadership at COMPUTEX 2026
At COMPUTEX 2026, the Taiwan Stock Exchange (TWSE) emphasized Taiwan’s critical role in the global AI ecosystem. It showcased the country’s complete value chain, from semiconductor design to AI infrastructure. TWSE chairman Sherman Lin highlighted the country’s capital markets. He said they are essential for supporting R&D and global expansion in these capital-intensive industries. This also showed Taiwan’s role extends beyond manufacturing. The briefings featured key Taiwanese companies. These included those in ASIC, DRAM, advanced packaging, thermal management, power delivery, and AI server platforms. This reflected significant investor confidence as Taiwan’s market capitalization hit a record high.
Taiwan continues to solidify its important role in the US-Japan-China technology competition. This is especially true in semiconductors and AI supply chains. Taiwan can attract and deploy capital for cutting-edge technology R&D and manufacturing capacity. This is a major advantage. It influences the strategic decisions of major tech players in the US, Japan, and mainland China.
Semiconductors & Hardware
SK Hynix to Double HBM Output Amid Soaring Demand and US Listing Plans
South Korean memory giant SK Hynix plans to double its production capacity over the next five years. This involves a $67 billion investment through 2028. The move is driven by high demand for High Bandwidth Memory (HBM) chips.
The company is a leading player in the HBM market. It is also seeking a US listing to raise capital. This comes amidst a projected memory shortage lasting until 2030.
SK Hynix is a South Korean firm. However, its expansion directly impacts the global semiconductor supply chain. This includes chip manufacturers and AI developers in China and Japan. Its strong position in HBM is crucial for AI accelerators. This shows the important role East Asian companies play in the US-China technology competition.
AI & Machine Learning
HKUST and CalmCar Launch Physical AI Innovation Center in Hong Kong
The Hong Kong University of Science and Technology (HKUST) and Suzhou Calmcar Electronics Technology Co. Ltd (CalmCar) have partnered. They will establish a Physical AI Innovation Center in Hong Kong.
This center will develop full-stack physical AI technologies. It aims to overcome current AI limitations in understanding, reasoning, and decision-making for real-world applications. These applications include autonomous driving and robotics. This work establishes Hong Kong as a key center for national AI innovation.
This initiative supports China’s broader ‘AI+’ strategy. It leverages Hong Kong’s global connectivity and the Greater Bay Area’s industrial supply chain. This combination will aid advanced AI research and industrial deployment. The project shows a focused effort to build indigenous AI capabilities in critical areas. This will reduce reliance on Western technology and create a competitive East Asian AI environment.
Semiconductors & Hardware
TrendForce: Key East Asian Suppliers Drive Laser Diode Capacity Expansion for AI Data Centers
Sumitomo Electric and Mitsubishi Electric are key suppliers of EML and CW-DFB laser diodes. This is a growing market. These components are important for high-speed AI data center optical interconnects.
These East Asian suppliers are increasing production capacity. They are responding to rising global demand from hyperscalers such as NVIDIA and Google.
This shows Japan’s strong position in advanced optical components. This area is important for the US-Japan technology alliance to ensure reliable supply. China also wants to develop its own capabilities. These are for high-end optical modules for AI infrastructure.
AsiaAI.FYI ·
Written by Dick Weisinger ·
Subscribe