#闪迪长期协议成焦点，开盘表现待验证 The core catalyst of this market round is the long-term supply agreement between SanDisk and leading cloud providers becoming the focus of market competition. The company has secured 8 large-scale long-term contracts, with an average contract period of 4 years and a maximum of up to 5 years. These contracts have locked in over half of the shipments for 2027 and two-thirds for 2028, set price floors, and are backed by customer prepayments as guarantees. Even if spot storage prices fall later, the performance baseline is supported, and the worst-case gross margin can still remain high. The biggest market imagination: storage is gradually shifting from a strongly cyclical product to a stable cash flow growth valuation target, which is the underlying logic behind the recent large capital inflows. However, there are also significant disagreements, which become variables at the opening: Some institutions worry that the long-term agreements also cap the price ceiling. If NAND spot prices continue to rise sharply, SanDisk cannot capture all the upside profits, limiting the profit ceiling. Once the market starts trading this negative factor, short-term profit-taking pressure will appear. Two major opening scenarios ✅ Optimistic scenario: capital continues to trade valuation re-rating, opening with a surge. The long-term agreements bring earnings certainty, capital continues to pay a premium, and the storage sector strengthens in tandem. ⚠️ Cautious scenario: the positive news is realized, opening surges then falls back. After a prior strong rise, some capital believes the long-term agreement benefits are already priced in and uses the news to sell off. Signals to watch during trading 1. Opening volume, only a volume surge with price increase has sustainability 2. Storage sector linkage (Hynix, Micron strength or weakness in sync) 3. Mid-session sentiment of the US tech market

Sandisk continues to surge, can it still come down? 🚀 Four major drivers behind Sandisk's explosive rally this round 1. Major long-term guidance at Investor Day reshapes valuation The latest investor conference provided a 2028-2030 outlook: revenue to maintain mid-to-high double-digit growth, gross margin to stay around 80%, and plans to return all excess cash to shareholders. This dispelled the market's previous concerns about "peak performance," leading institutions to reprice the stock, no longer viewing it purely as a cyclical stock. 2. AI inference boom drives explosive demand for enterprise flash AI large model inference and vector databases bring massive storage demand, with cloud providers aggressively purchasing large-capacity enterprise SSDs. The company's data center business revenue is skyrocketing, no longer relying on traditional USB drives and consumer storage cards. They launched a new generation of ultra-high capacity enterprise drives, securing numerous cloud provider orders. 3. Global NAND supply tightness and chip price increases The industry generally expects the storage shortage to continue until 2027, with slow upstream capacity expansion and rising flash chip prices. Storage manufacturers have strong profit elasticity; even slight price increases can significantly boost net profits. The storage sector collectively strengthens, driving positive sentiment. 4. Capital inflows fuel a short-term short squeeze rally Previously, many shorts bet on a pullback after the surge, but after continuous rises, shorts were forced to cover. Combined with institutional buying, this amplified the short-term upward momentum, trading volume surged sharply, accelerating the short-term rally. ⚠️ Core risks not to be ignored 1. Essentially still a storage cyclical stock; if major manufacturers expand production as planned, chip prices will fall, and profits will decline rapidly. 2. Fierce competition from peers; Samsung, SK Hynix, and Micron are also increasing investment in the AI storage track, which will squeeze profit margins. 3. Current stock price volatility is extremely high; after a short-term surge, a deep correction could occur at any time. 📌 Signals to watch for in the future • Spot price trends of NAND flash • Overseas cloud capital expenditure guidance • Changes in institutional ratings and target prices

#海力士扩产提速，资本开支能否兑现回报 1. Overview of the current expansion plan SK Hynix recently finalized a massive investment totaling 54 trillion KRW (approximately $38 billion) to build two major wafer fabs: Yongin Y2 and Cheongju M17, accelerating the pace of capacity deployment. • Yongin Y2 fab: Focused on HBM and high-end DRAM, expected to start cleanroom production by June 2029 • Cheongju M17 fab: Dedicated to NAND flash and advanced packaging, production to begin by the end of 2028 Capital expenditure for 2026 has been raised to over 40 trillion KRW, with the vast majority of funds directed toward the AI high-bandwidth memory segment; there are no large-scale expansion plans for ordinary consumer-grade memory. 2. Supporting logic for achievable returns 1. Short-term sustained supply-demand gap for high-end products HBM is currently in short supply, with 2026-2027 capacity already locked in by long-term contracts from overseas cloud providers. The new capacity targets AI servers and inference scenarios, matching the current strong demand. High-end product gross margins remain high, providing ample profit space. SK Group anticipates 2027 to be the peak of memory shortages, with new capacity coming online just in time to meet incremental demand. 2. Customer long-term contracts secure revenue baseline Major manufacturers have pre-signed multi-year supply agreements with top clients like Google, Nvidia, and Microsoft, locking in shipment volumes and prices in advance. Even if the market fluctuates in the future, these orders guarantee basic cash flow, significantly reducing the impact of cyclical volatility. 3. Industry oligopoly structure and capacity discipline After the last cycle's sharp downturn, Samsung, Hynix, and Micron reached a consensus to prioritize profitability and no longer expand general memory capacity indiscriminately. New capacity is concentrated in high value-added AI memory, avoiding the low-end capacity glut of the past, thus reducing the risk of price wars compared to historical cycles. 3. Potential risks, returns may not materialize 1. Long investment return cycle The two new fabs will not release capacity until 2028-2029, with construction and ramp-up periods lasting several years. Large capital expenditures will continue to erode current cash flow, with heavy interest and depreciation burdens. By the time capacity comes online, the market landscape may have changed. 2. Risk of AI demand falling short of expectations The risk is concentrated around the 2028-2029 capacity release window: if global major players slow AI capital spending or large model iterations improve efficiency, reducing memory required per task, the new capacity could become a burden, repeating the storage industry's historical cycle of "overcapacity following expansion." 3. Competitors simultaneously increasing investment Samsung and Micron are also ramping up HBM expansion, with concentrated production in a few years. The high-end segment will gradually enter a phase of increased supply, squeezing product profits and making it difficult for gross margins to permanently maintain current very high levels. 4. Key signals to monitor for the market outlook ✅ Optimistic signals: sustained strong HBM orders, continued signing of long-term contracts, AI compute capital expenditure remains robust ⚠️ Risk signals: leading cloud providers reduce procurement budgets, major memory manufacturers further increase expansion, AI technology significantly reduces memory consumption In the short term, expansion is a positive sentiment; in the medium to long term, whether returns can be realized essentially bets on whether the structural AI memory dividend can sustain until the new fabs' capacity is released.

#消费动能转弱，9月政策仍受通胀制约 Current end-consumer demand recovery is slowing, with weak consumer willingness and domestic demand repair falling short of previous market optimism. The market originally expected a round of easing policies in September, but inflation stickiness has become the biggest constraint, locking the space for significant easing, making it difficult for policies to fully stimulate the economy. 1. Signals of weakening consumption momentum 1. Discretionary consumption remains weak; major consumption categories like automobiles and home appliances show weak recovery, daily consumption is mainly essential, with insufficient willingness for excess spending. 2. Residents tend to save and are cautious about future income expectations; even small consumption subsidies have limited pull effect. 3. The service sector's impulse rally has faded; cultural and tourism consumption declined after summer, lacking new consumption growth points. 2. How inflation constrains September easing Inflation has not fully declined to a range that allows worry-free easing; some price components remain resilient: • Large interest rate cuts and increased stimulus could trigger price rebounds, bringing new inflationary pressure. • Therefore, monetary policy will be cautious, with a lower probability of large rate cuts; more use of fiscal tools and targeted support rather than broad easing. Simply put: although domestic demand is weak, prices have not yet given policy the "green light" to loosen fully. 3. Major asset scenario analysis 1. Stock market: expectations for a broad bull market cool down, shifting to structural trends. Pro-cyclical consumer sectors are unlikely to see widespread rallies; funds will continue to cluster around AI and high-growth sectors. 2. Bond market: easing expectations are lowered, yield decline space narrows, the one-sided bull market phase ends, shifting to oscillation. 3. Commodities: weak domestic demand suppresses downside space, but inflation floor prevents sharp drops, resulting in overall range-bound oscillation. 4. Two key indicators to watch going forward 1. CPI inflation data for August-September; if inflation falls, policy easing space will reopen. 2. Types of policies implemented in September: if mainly special bonds and industrial subsidies, it indicates targeted easing; if rate cuts and reserve requirement ratio cuts appear, it means constraints have been lifted.

The reason for SanDisk's surge has been found, but unfortunately, I shorted it and am currently stuck in a losing position. Last night during the US stock market session, SanDisk surged as high as +17%, closing with a strong gain of 13.67%, directly driving the entire storage sector to collectively rebound, with SK Hynix and Micron also rising sharply. 1. Direct trigger for the surge: Investors' heavy long-term guidance released today 1) Provided long-term performance targets: Revenue growth maintained at mid-to-high double digits from 2028 to 2030, long-term gross margin target at 80%, free cash flow margin at 50%, with profit expectations directly revised upward. 2) Shareholder return commitment: After capacity investments are completed, all remaining cash flow will be returned to shareholders (via buybacks and dividends), dispelling market concerns about disorderly capacity expansion despite profits. 3) Enhanced AI inference storage narrative: The company believes AI is shifting from training to inference, which will trigger a new wave of flash memory demand; the enterprise flash memory market size is expected to rise significantly by 2030, and expectations for the implementation of HBF (High Bandwidth Flash) technology are heating up, unlocking growth potential. 2. Underlying market logic: Previous oversell + macroeconomic tailwinds resonating 1) After recent earnings reports, the market worried that the storage cycle had peaked, causing the stock price to continuously pull back, accumulating a large amount of short positions and oversold chips; once the positive news came out, shorts covered heavily, amplifying the rally. 2) US PPI data came in below expectations, combined with CPI meeting expectations, the market priced in reduced short-term Fed rate hike pressure, marginally improving liquidity conditions for growth stocks, and capital flowing back into the AI hardware sector. 3) The logic of long-term supply agreements is being repriced by capital: 3-5 year long-term supply contracts lock in orders, weakening the strong cyclical nature of the storage industry, and the valuation midpoint is expected to rise. 3. Current risk points (cannot be ignored) 1) The large single-day bullish candlestick is event-driven; short-term profit-taking is heavy, and pre-market and intraday volatility the next day will be significant, making it easy to spike and then fall back. 2) The long-term targets are a 3-year outlook; short-term quarterly earnings have not been raised, and some of the positive news has already been priced in. 3) The storage sector as a whole remains highly tied to US Treasury yields and Federal Reserve policies; once hawkish statements return, the sector will quickly come under pressure. 4. Key signals to watch going forward 1) Whether it can hold the high point of this rebound; if it quickly falls back to the launch platform, it will be a pulse-type rally. 2) Whether the storage sector forms sustained linkage (with SK Hynix and Micron continuing); if only SanDisk rallies alone, the rebound's sustainability is limited. 3) The Jackson Hole central bank meeting and Fed statements will determine the major direction for tech growth.

Today I made two short trades and am currently holding a short position on Micron. I was really nervous at the start, a bit thrilling, but now I'm closer to my 10,000 yuan goal. Keep it up, going to sleep first, will share trading insights tomorrow. Wishing all the teachers great wealth 😆