East Asian Technology Intelligence
Japan & China tech news — translated, contextualized, and delivered weekly.
Weekly. Free. Unsubscribe anytime.
3 Takeaways This Week
- Western tech leaders must immediately audit developer environments using OpenAI’s Codex CLI to prevent a silent software bug from destroying high-performance SSDs via 37 terabytes of excessive write operations in just three weeks.
- Global tech giants should secure power partnerships early as SoftBank’s telecom unit expands into the U.S. with plans to rent AI-tailored cloud compute at an unprecedented 10-gigawatt scale.
- Institutional investors should replicate the strategy of foreign funds who poured a record $60 billion into Japan’s stock market during the first half of the year by targeting the country’s AI-powered factory robot manufacturers and niche tech suppliers.
This week’s signal
What? 37 Terabytes in 21 Days? Beware of ‘That AI Tool’ Eating High-Performance SSDs: 886th Lap #
<
Western enterprises using AI tools, especially coding assistants, must quickly check how these tools use physical resources. A recent report from Japan showed that OpenAI’s Codex CLI caused extreme wear on solid-state drives. The tool wrote 37 terabytes of data in just 21 days. This is not just a simple technical glitch. It is a strong warning against trusting even the best AI applications without proof. It also shows a real financial risk, which is the physical cost of software defects.
Japanese media, especially ITmedia AI+, did not view this as a debate about AI safety or ethics. Instead, they saw it as a practical issue of hardware losing value due to software flaws. This view is important. Many Western stories focus on risks like runaway AI. In contrast, the Japanese view focuses on immediate costs and operations. For a Japanese company, this is a direct hit to profits because they must replace expensive, high-performance drives too early. This shows the deep risk aversion in Japanese enterprise tech. It also shows their focus on proven performance and reliability. The message is that even popular, helpful AI tools require careful testing and caution.
The cause was a telemetry feature turned on by default and changes to SQLite logging levels. This proves the main issue is not the intelligence of the AI, but its basic software engineering. This fits a broader trend in East Asia. In this region, practical use and stable operations matter more than theory. Silicon Valley favors fast launches and public updates. The Japanese market prefers strong, predictable performance.
Western firms should audit their AI tools immediately. Beyond security and privacy, the cost of unexpected hardware damage is real and large. We must watch how fast AI makers fix these heavy resource flaws. We must also watch if companies start to demand more openness and control over resource use from AI sellers. This could become a new factor when businesses choose which AI software to buy.
🗾 Japan Radar #
What Japanese media is reporting that Western outlets miss
🗾 AI & Machine Learning
What? 37 Terabytes in 21 Days? Beware of ‘That AI Tool’ Eating High-Performance SSDs: 886th Lap
A software bug in OpenAI’s coding assistant, Codex CLI, is causing excessive write operations to users’ solid-state drives (SSDs), significantly shortening their lifespan. Reports indicate that the tool wrote approximately 37TB of data in just 21 days, largely due to a telemetry feature enabled by default and a change in SQLite logging levels. Japanese media is framing this as a practical issue of hardware wear and tear stemming from software flaws, rather than philosophical AI risks. It emphasizes the need for caution and verification even with popular, seemingly beneficial AI tools, aligning with a more pragmatic, risk-averse approach common in Japanese enterprise IT.
For Western readers: Western businesses and individual developers using AI coding assistants should actively monitor their hardware’s health and resource usage to prevent premature SSD failure and unexpected costs. 🗾 AI & Machine Learning
Meta CEO Mark Zuckerberg stated in an internal meeting on July 2nd (local time) that AI agent development over the past four months has “not accelerated as expected.” He also noted that the large-scale organizational restructuring, which involved significant layoffs and redeploying 7,000 employees to AI teams, was not as “clean” as it should have been and that management misjudged the timing of the changes. Zuckerberg expressed optimism that Meta would start seeing greater benefits from its substantial AI investments within the next 3-6 months. Japanese media, like ITmedia AI+, often scrutinize global tech giants for their strategic execution, providing local readers with insights into whether such ambitious AI shifts are truly delivering as promised. The article’s focus on internal communication and the role of ‘Claude Code‘ in management’s optimism offers a deeper look into the operational realities behind the AI race, resonating with a business culture that values careful planning and execution.
For Western readers: Western businesses and investors should temper expectations for immediate, dramatic returns from large-scale AI transformations, acknowledging the significant operational hurdles and human capital challenges even top-tier companies encounter. Enterprise & Cloud
SoftBank plans to rent AI compute in U.S. at 10-gigawatt scale SoftBank’s telecom unit head, Junichi Miyakawa, announced plans for a new AI-tailored cloud venture, 49% owned by its parent group, that aims to rent AI compute capacity in the U.S. at a 10-gigawatt scale. Miyakawa anticipates this venture could generate profits “on a different order of magnitude,” indicating SoftBank’s aggressive push into the high-growth AI infrastructure market. SoftBank’s substantial investment in U.S. AI compute infrastructure demonstrates Japan’s ambition to secure its position in the global AI race, which is critical for national competitiveness against China and the U.S. Western media frames this as a significant business play by a major Japanese conglomerate.
For Western readers: Western AI companies and cloud providers will face increased competition and potential partnership opportunities from a well-capitalized Japanese player in the AI compute market. Semiconductors & Hardware
Foreign Investors Scoop Up Half-Year Record $60bn in Japan Stocks, Driven by AI and Niche Companies
Overseas investors poured a record 9.7 trillion yen ($60 billion) into Japanese equities in the first half of 2026, marking the highest net buying ever for a half-year period. This surge is attributed to improvements in capital efficiency and a robust base of chip-related companies benefiting from the global AI boom. The record foreign investment signals strong confidence in Japanese corporate governance reforms and its strategic positioning within the AI and semiconductor ecosystem. Western media widely reports on Japan’s market resurgence, often framing it as a safe haven and growth opportunity amidst global tech competition.
For Western readers: Western investors and technology firms should recognize Japan as a crucial hub for AI-enabling technologies and a significant target for capital deployment, offering diversification and exposure to a robust, reforming market. Robotics & Automation
Japan Eyes AI-Powered Comeback in Factory Robot Race with China, Europe
Japanese industrial robot manufacturers are strategically investing in AI and open collaboration to reclaim their once-dominant global market share, which has declined to 40%. Companies like Yaskawa Electric are integrating AI-powered machines into new factories, while Kawasaki Heavy is partnering with Nvidia and plans a significant bond issuance for AI investments. Japan’s initiative highlights a national strategy to secure a leadership position in ‘physical AI,’ an area where it perceives a competitive advantage. Western media generally frames this as a race, but local coverage also emphasizes the collaborative aspects and domestic policy support, such as the $65 billion public-private investment target.
For Western readers: Western businesses, particularly those in automation, AI, and manufacturing, should anticipate increased competition and potential collaboration opportunities with Japanese firms as they aggressively re-enter the global market with advanced AI-driven solutions.
🇨🇳 China Watch #
China’s technology moves, framed for Western readers
Startups & Funding2 STORIES
Kuaishou’s Kling AI Valued at $18B After Huge $3B Funding Round Chinese short-video giant Kuaishou is reportedly spinning off its AI video generation platform, Kling AI, following a massive $3 billion funding round that values the entity at $18 billion. Supported by major domestic tech and cloud giants like Tencent and Alibaba Cloud, this pre-IPO financing positions Kling AI for independent commercial operations ahead of a potential 2027 listing.
Why it matters: The substantial valuation and investor interest, including Tencent, highlight China’s strategic focus on leading in AI foundational models and applications, especially in areas like video generation where local giants see immense market potential. Western media frames this as a continuation of China’s aggressive investment in AI, contrasting with a more cautious approach in some Western markets due to regulatory scrutiny.
For Western readers: Western investors and AI companies should recognize the scale of investment and rapid development in China’s generative AI space, which could foster new market leaders and technological advancements impacting global competition. Semiconductors & Hardware
Inside Huaqiangbei’s AI Hardware Boom: Four Kings and Budget Products Conquering the Mass Market Shenzhen’s Huaqiangbei district, renowned for electronics, is experiencing a boom in budget-friendly AI hardware, driven by ‘Four Kings’ chipmakers — Huawei Ascend, Sophon, T-Head, and Horizon Robotics — and smaller players. These manufacturers are rapidly developing AI-driven embedded devices, edge computing solutions, and surveillance products, catering to a diverse domestic market that prioritizes affordability and rapid deployment. The rapid commercialization of budget AI hardware in Huaqiangbei signifies China’s strategic pivot towards pervasive AI applications, utilizing domestic chip architectures to meet internal demand and reduce reliance on foreign technology. Western media often underreports the volume and ingenuity of these localized solutions, focusing more on high-performance AI rather than mass-market adoption.
For Western readers: Western businesses, particularly in industrial AI, IoT, and surveillance, will face increasing competition from cost-effective and functionally robust Chinese AI hardware, potentially impacting market share in developing economies. Semiconductors & Hardware
China’s Domestic AI Compute Revolution Reaches a Tipping Point China is experiencing a significant surge in domestic AI compute capabilities, driven by a national imperative to overcome hardware sanctions. local AI chip manufacturers and data center operators are rapidly expanding, indicating a shift towards greater self-sufficiency in high-performance computing necessary for advanced AI development. The acceleration of China’s domestic AI compute capacity directly challenges US efforts to constrain its technological progress, demonstrating remarkable resilience and strategic adaptation. While Western media often focuses on the challenges posed by sanctions, this article highlights the domestic innovation and infrastructure build-out within China.
For Western readers: Western AI companies and semiconductor firms face a more formidable, self-reliant Chinese competitor and a potentially shrinking market for their high-end AI compute solutions within China. Policy & Regulation
German government pledges to defend trade, signalling tougher China stance
Germany’s ruling coalition has adopted a tougher stance on trade defense, signaling a potential shift in European Union policy towards China. This move, part of a 34-point economic revival package, aims to protect German companies from unfair trade practices, particularly by addressing issues like Chinese industrial subsidies and trade imbalances. This German policy pivot matters for East Asian tech as it could lead to increased EU anti-dumping and anti-subsidy measures, impacting Chinese tech exports. Western media frames this as a substantial shift in German policy, driven by domestic economic concerns and international pressure.
For Western readers: Western businesses and policymakers should anticipate a more assertive EU trade policy towards China, potentially resulting in new tariffs or trade restrictions on Chinese goods and technology.
🔺 The Triangle #
Where US, Japan, and China technology interests intersect
Semiconductors & Hardware
Apple Leverages Procurement Clout Amid Shortages; Chinese Brands Slash Production 📊 Chart of the Issue
Source: Electronics Weekly, Nikkei. Targets for calendar year 2026.
Apple is reportedly capitalizing on its superior procurement capabilities to increase smartphone market share in 2026, ordering components for 80 million iPhones for H2 launches, including 10 million Folding phones. In contrast, major Chinese smartphone brands like Xiaomi, Oppo, and Vivo have each drastically cut their 2026 production targets to under 100 million units due to severe memory and electronics component shortages. Apple’s ability to secure components amidst shortages, while Chinese rivals struggle, could shift market dominance further towards non-Chinese brands, impacting local suppliers and R&D. Western media frames this as an Apple success story, while local coverage in China would likely focus on the challenges for domestic brands and the broader implications for the supply chain.
For Western readers: Western component suppliers with strong ties to Apple may see sustained demand, while those reliant on Chinese brands could face reduced orders and revenue volatility. AI & Machine Learning
Anthropic Closes Loopholes for Chinese Access to Claude AI AI developer Anthropic is actively closing loopholes, including VPNs, relay services, and overseas accounts, that have allowed users in China to access its Claude AI. This action comes amidst persistent efforts by Chinese users to find new workarounds to access restricted Western AI services. Anthropic’s efforts reflect the intensifying US-China AI competition and the challenges Western firms face in enforcing geofencing on their cutting-edge AI, directly impacting the availability and use of advanced models within China’s tech ecosystem. Western media frames this as a clear enforcement of restrictions, whereas local Chinese discussions often focus on strategies for bypassing such controls.
For Western readers: Western AI developers must continuously invest in robust access control mechanisms to adhere to regulatory requirements and protect intellectual property, impacting their global market strategies and operational costs. Semiconductors & Hardware
NTU Singapore Launches Industry-Aligned IC Design Program Nanyang Technological University (NTU) Singapore has launched a new FlexiMasters in Integrated Circuit (IC) Design, developed in collaboration with the Singapore Semiconductor Industry Association (SSIA) and supported by the EDB. This program aims to address the growing demand for chip design talent in Singapore, driven by companies like AMD, Broadcom, and Qualcomm which have IC design operations in the city-state. The program strengthens Singapore’s role as a vital link in the global semiconductor supply chain, attracting further investment and talent to East Asia, distinct from the domestic focus seen in China’s talent development efforts.
For Western readers: Western fabless semiconductor companies with operations in Singapore will benefit from a more robust local talent pipeline, easing recruitment challenges for highly specialized IC design roles. Semiconductors & Hardware
Japanese fabless semiconductor company EdgeCortix successfully demonstrated its SAKURA-II Edge AI platform in a U.S. Air Force exercise, validating its ability to deliver high-performance, low-power AI inference in operational environments. The platform also received a Success Memorandum from the U.S. Defense Innovation Unit (DIU) for meeting all prototype project objectives, highlighting Japan’s growing role in critical defense technology collaboration with the US. This matters for the East Asian tech landscape as it showcases a Japanese firm successfully integrating advanced AI hardware into US defense systems, demonstrating Japan’s high-tech capabilities and strengthening its strategic alignment with the US. While local Japanese media might emphasize national pride and technological achievement, Western media will likely frame it more in terms of US defense capabilities and strategic partnerships.
For Western readers: Western defense contractors and technology firms should view EdgeCortix as a potential partner or competitor in the burgeoning market for resilient edge AI solutions for aerospace and defense applications. Workforce & Culture
73% of Southeast Asian Employees Feel Leadership Disconnected from Digital Needs: Lark Report
📊 Chart of the Issue
Source: Lark ‘The Paradox of Progress’ report
A report by Chinese workspace platform Lark reveals 73% of Southeast Asian employees believe their leadership is disconnected from actual digital needs, leading to a ‘Digital Tax‘ that erodes productivity. This disconnect fosters fear over AI and job security, with only 28% feeling empowered to innovate, despite a strong desire for AI to handle routine tasks. The findings underscore a crucial gap in enterprise digital transformation efforts across Southeast Asia, where employee experience is often overlooked. This perspective is critical for Chinese tech firms like Lark, which are major players in the region’s enterprise software market, contrasting with a Western focus that often prioritizes technological deployment over adoption.
For Western readers: Western companies operating or investing in Southeast Asia should be aware of these deep-seated employee concerns regarding digital transformation and AI, as they can significantly impact project success and ROI.
[AsiaAI.FYI](https://asiaai.fyi) ·
Written by Dick Weisinger ·