Asia-Pacific Chip Selloff Drags A-Share ChiNext and STAR Down 3%; Qwen Go Launch Nears, India Restarts Rupee Defense

On October 13, Asia-Pacific tech stocks weakened broadly under the dual impact of intensified memory-chip competition and a cooling AI trade: China's ChiNext and STAR Composite indices fell about 3%, Hong Kong's Hang Seng Tech Index dropped more than 2%, and South Korea's KOSPI once plunged 5.1% and triggered a circuit breaker. Meanwhile, the commercialization of personal AI assistants, India's rupee defense, and a major internal rotation in US equities formed the day's other three main threads in global financial markets.^[1][3]

Asia-Pacific Chip Rout: Memory Competition and Cooling AI Trade Converge

On October 13, A-shares opened broadly lower. As of publication, the Shanghai Composite fell 1.36%, the Shenzhen Component fell 2.49%, and the ChiNext Index fell 3.21%, losing the 2,900 mark; the STAR Composite fell more than 3%, completely erasing its year-to-date gain. As AI hardware and semiconductor chips fell broadly, SMIC, Hygon Information Technology and AMEC all dropped around 4%. Hong Kong stocks opened lower and stayed weak, with the Hang Seng Index down 0.78% and the Hang Seng Tech Index down 2.24%; heavyweight internet and chip stocks sold off sharply, with GigaDevice, Cambridge Technology and others falling. Zhongji Innolight's H-shares hit a record low since their Hong Kong listing, down nearly 20% in October; Hong Kong's optical module index continued to deepen losses, and Cambridge Technology fell nearly 8% again.^[1]

However, A-shares did not fall across the board. Dip-buying funds gradually entered, with liquor, coal, agriculture, real estate, oil and gas, and shipping sectors outperforming; Huangtai Liquor hit the limit up, while Guanghui Energy, COSCO Shipping Energy and China Merchants Energy Shipping all rose about 4%; in coal, Antai Group approached the limit up. In commodities, most domestic futures fell, but fuel oil rose nearly 4%, crude oil rose 2%, and the container shipping index rose more than 1%. In bonds, the 30-year Treasury bond futures main contract rose 0.08%, the 10-year rose 0.04%, the 5-year rose 0.01%, and the 2-year was flat.^[1]

South Korea and Japan fell more sharply. On Tuesday, the KOSPI's decline once widened to 5.1% to 6,061.82; the Korea Exchange activated the Sidecar mechanism at 9:47 a.m. local time, pausing program selling for five minutes; Samsung Electronics and SK Hynix both fell more than 6%, with SK Hynix down nearly 8%, heading for a fifth straight decline. Japan's SoftBank Group fell more than 8%, while Tokyo Electron, Advantest and other AI supply-chain names each fell more than 5%. The MSCI Asia-Pacific Index fell 0.9% to 270.78, approaching its lowest since mid-August; Nasdaq 100 futures fell 0.5% in Asian morning trading.^[3]

The direct variable weighing on memory chips came from changes in the competitive landscape. ChangXin Memory Technologies President Cao Kanyu announced at an industry event over the weekend that the company had made key progress on a next-generation 4F² memory cell architecture and planned to launch server memory products based on that architecture by the end of 2026; the design uses a proprietary vertical gate-all-around transistor to suppress leakage and targets high-margin DRAM used in AI servers. Combined with an overnight rout in US memory chips and the early termination of Samsung and SK Hynix's largest-ever buyback (about $40 billion), valuation expectations for the Korean chip giants came under pressure.^[1][3]

Personal AI Agents Accelerate: Qwen Go to Launch, Agent Monetization Hopes Rise

The personal agent race has a new product. Hard AI has learned exclusively that Qwen Personal Agent 'Qwen Go!' is about to launch. In a user-shared experience video, it entered a flight booking platform to find a delayed flight and, after negotiating with customer service, secured RMB 70 in cash compensation; it also searched Taobao for orders eligible for price protection and recovered RMB 52 in price difference for the user. The first batch of use cases will also include car maintenance quotes and appointments, sports venue bookings, and calculating the final price of a refrigerator by combining subsidies, trade-in recycling and delivery-installation fees.^[2]

Compared with earlier needs that had clear goals and short chains, Qwen Go upgrades its scope and service capabilities based on an agent architecture, enabling it to handle matters that require repeated checking, communication, and filtering and comparison. Users give goals and constraints, while the agent takes on specific operations in the process: checking orders, finding human customer service, and verifying discount rules. Overseas, Muse also uses a cloud-based operating environment; Meta disclosed that Muse has an independent cloud computer and browser, can fill out forms and communicate on behalf of users, and can continue working after the user closes the app.^[2]

Institutions are quantifying the commercial and compute increments brought by personal agents. In research dated October 7, Bank of America argued that agents help users complete tasks and purchases more smoothly and could expand overall online activity. Morgan Stanley estimates that China's consumer AI addressable market oriented toward individuals will be about RMB 300 billion by 2030 and could reach RMB 1.6 trillion by 2040, with main revenue coming from advertising and transaction commissions. In research dated October 6, Citi raised its CPU market forecast, expecting the global CPU market to grow from $29 billion in 2025 to $300 billion in 2030, with execution and orchestration workloads from personal agents as the basis for its judgment.^[2]

India Restarts 'Currency Defense': Central Bank Takes Over Oil Companies' FX Purchases

India's central bank introduced a package of rupee defense measures over the weekend. Starting October 12, 2026, the Reserve Bank of India opened a dedicated window to provide foreign exchange directly through designated banks to three state-owned refiners — Indian Oil (IOC), Hindustan Petroleum (HPCL) and Bharat Petroleum (BPCL) — completely bypassing the spot market. Goldman Sachs emerging market strategists Rishi Radia and Lu Sun said in a report titled 'India: RBI Reinforces Rupee Defense' that the measures could push the rupee higher by 1% to 2% in the short term but do not solve the fundamental problem of insufficient dollar inflows; the largest buyer that had previously pressured the rupee in the market each day has disappeared from the spot market.^[4]

Goldman Sachs estimates that, based on current oil prices of about $100 per barrel, the three refiners' annual dollar demand is about $120 billion. The report says bluntly that actual dollar demand has not changed, only shifted to being met by the RBI's foreign exchange reserves; pressure on the rupee in the spot market has eased, but at the cost of faster reserve depletion. As long as oil prices remain near $100, every barrel of crude India imports will directly consume foreign exchange reserves.^[4]

At the same time, the RBI revived a combination of derivatives controls briefly used during acute stress in April: it prohibits re-booking of cancelled rupee FX derivative contracts; it slashed the cap on FX derivative positions that can be established without prior proof of actual exposure from $100 million to $5 million, a 95% cut, applying to all authorized dealers and major FX derivatives exchanges; it imposed a 20% cash margin deposit requirement on purchase FX hedging contracts above $2 million and created a new Foreign Exchange Risk Reserve (FERR) mechanism. Goldman Sachs characterized this as 'a disguised increase in the cost of purchase FX hedging' and warned that rupee non-deliverable forwards (NDF) provided by independent offshore banks are not directly constrained, so some hedging demand could flow to the NDF market.^[4]

On the macro backdrop, the RBI recently raised rates by 25 basis points to 5.50% under a 'gradual tightening' stance; foreign investors have net sold about $31 billion of Indian equities this year, while domestic institutions have net bought about $72 billion. Goldman Sachs economist Santanu Sengupta expects the RBI to raise rates by another 25 basis points in December and by a cumulative 50 basis points more in the first half of 2027, pushing the policy rate to 6.25%. The rupee had previously approached a record low of 96.8, and the market will be further tested after liquidity drains from October 13 open market operation sales and October 20 tax payments.^[4]

Big Rotation in US Equities: Quality Factor Surges, Tech Net Sold

Beneath a calm surface, US equities are internally more divided than at any time since the dot-com bust. A latest report from Goldman Sachs flow strategist Lee Coppersmith shows that the GS Quality Factor has gained more than 16% over the past eight weeks, a move of a magnitude not seen since the 2020 COVID shock. The median S&P 500 constituent is now down 16% from its own high, while the index overall is less than 1% from a record high; by Goldman's preferred measure of market breadth, breadth has deteriorated to the lowest level since the dot-com bust.^[6]

Tech stock flows have reversed sharply. Goldman Prime data show that tech stocks were the top net bought sector in four of the prior five weeks, but last week they suddenly became the most net sold sector, with selling on both the long and short sides; the tech long/short ratio has fallen to 1.8, in the 17th percentile over five years, while gross exposure remains in the 97th percentile over five years. At the same time, industrials saw their fastest buying pace since January (+1.8 standard deviations), almost entirely from new longs; financials also attracted inflows, and software showed relative strength. Consumer staples were net sold in seven of the past eight weeks, with the long/short ratio falling to the 3rd percentile over five years; biotech experienced its largest eight-week gross exposure compression in more than five years.^[6]

Although risk-off sentiment is strong internally, Goldman Sachs believes the current positioning structure actually provides potential upside support. Goldman's US equity sentiment indicator currently reads -1.3; global fundamental long/short funds' net leverage is in the 6th percentile of the past year (2nd percentile for US-only), and hedge funds' overall net leverage is in the 10th percentile. Bloomberg statistics on 13 instances since 1995 in which the S&P 500's top 10 constituents outperformed the equal-weight index by more than 8% show a median subsequent six-month return of +10.1% and a median 12-month return of +19.6%, both above the historical medians of +6.9% and +15%; the current performance gap is 855 basis points.^[6]

Goldman also flagged oil price and rate risks. After the 10-year US Treasury yield broke above 5%, Goldman's rates team believes the asymmetry in yields has skewed to the downside; Goldman's commodities team expects Brent crude to fall to $90 per barrel in December and average $80 in 2027, but stresses that near-term risks are clearly skewed to the upside, with more than $20 per barrel of risk premium still embedded in current oil prices. Nvidia's downside implied volatility has fallen to a low never seen before the COVID pandemic, offering a window for investors who want to hedge the risk of an AI trade unwind.^[6]

What to Watch: Earnings, CPI and Liquidity

Market attention is turning to upcoming macro data and earnings season. TSMC's quarterly earnings will be released this Thursday and are seen as one of the most important bellwethers for AI demand; US consumer price index (CPI) data will also be published this week, after US Treasury yields have risen to multi-decade highs. Jason Pride, head of investment strategy and research at Glenmede, said this week's CPI report is likely to be the most important data seen before the Fed's October meeting.^[3]

US large bank earnings season officially starts on Tuesday, with Goldman Sachs, JPMorgan, Citigroup and Wells Fargo all reporting. According to Bloomberg Intelligence data, analysts expect S&P 500 companies' third-quarter profits to grow about 25% year over year. Ameriprise chief market strategist Anthony Saglimbene believes bank stocks, big tech and energy stocks will be the three main threads for interpreting this earnings season.^[3]

Other assets: the dollar index edged down 0.1%, the 10-year Treasury yield fell 3.7 basis points to 5.209%; WTI crude fell 0.9%, Brent crude fell 0.9% to about $105.80 per barrel; spot gold extended overnight losses, falling 0.36% in Asia-Pacific trading on Tuesday to $4,157 per ounce.^[3]

Wall Street CN Calendar Opens Ninth-Year Presale

Wall Street CN's 2027 calendar presale has begun, marking the ninth consecutive year for the product.^[5]

Common Trends and What to Watch Next

Taken together, global markets are repricing around the true cost and execution capability of the AI trade: memory chip competition is quickly transmitting shifts in the supply landscape to relevant sectors in South Korea, A-shares and Hong Kong; personal agents are pushing the AI narrative into real app operations, compensation negotiations and transaction conversion, with institutions using long-term market forecasts to measure their compute and commercialization space. The RBI's measures have compressed FX purchase demand in the spot market, but reserve depletion and insufficient dollar inflows remain constraints; US equity funds are shifting from tech stocks to the quality factor, industrials and financials, showing that investors value quality and light positioning more under rate and oil price risks. Looking ahead, track TSMC's earnings, US CPI, US large bank results, and liquidity changes from India's October 13 open market operation sales and October 20 tax payments.^{[1][2][3][4][6]}