To The Moon

Last week, the tech market witnessed an epic rally, with the AI theme accelerating from a "red-hot summer" to a "booming autumn." Combined with tight supply-demand dynamics in the memory sector and a surge in AI application users, the $Invesco QQQ(QQQ)$ rose nearly 2%, From multi-billion-dollar AI capacity contracts to trillion-dollar order backlogs, and from memory price inflections to AI apps topping rankings, a flurry of key events unfolded in a single week—signaling the tech industry is entering an acceleration lane rather than peaking.

I. AI Main Theme: Clear Demand Inflection, Giant Moves Ignite the Market

1. Core Contracts & Earnings: Rewriting Industry Expectations

• $NEBIUS(NBIS)$ : On Monday, it signed a 5-year AI infrastructure capacity deal with $Microsoft (MSFT)$ worth $1.74B-$1.94B, sending its stock price surging 50% the same day. It later raised an additional $3B in funding to expand capacity, marking a "strong start" for AI infrastructure plays this week.

• $Oracle(ORCL)$ On Tuesday, it reported a staggering $45.5B in Remaining Performance Obligations (RPO), a 359% year-over-year (YoY) jump. After the market repeatedly verified the press release to confirm it was not a typo, the stock soared over 35%—a move evoking memories of $NVIDIA(NVDA)$ ’s "legendary rally" post-ChatGPT in 2023, underscoring corporate confidence in long-term AI infrastructure investments.

2. Institutional Bullishness & Demand Inflection: AI Momentum Continues

AI demand is shifting from "isolated outbreaks" to "widespread adoption," with revenue opportunities expanding from hardware to software and applications. Additionally, $CoreWeave, Inc.(CRWV)$ s presentation was particularly notable, noting: "That core issue in the AI space not only persists but is actually worsening—and what we’ve observed over the past 4-6 weeks is yet another critical inflection point in demand." This further confirms the high prosperity of the AI industry chain.

II. Memory Sector: Tight Supply-Demand Drives Prices, Leading Players Diverge

This week, the memory sector was driven by a dual catalyst—"improved demand from hyperscale data centers + ORCL’s earnings boost"—with DRAM server prices continuing to climb. While performance among sector players diverged, the overall trend remained strong:

1. $Micron Technology(MU)$ 40% Monthly Surge, Price Hike Plan Launched

• Its stock price has soared 40% in September alone, catching short-sellers off guard who bet on a 2026 oversupply of High-Bandwidth Memory (HBM).

• According to local Asian reports, Micron plans to raise prices of its entire product lineup (DDR4, DDR5, LPDDR4, LPDDR5) by 20%-30% and suspend current quotations—directly reflecting tight supply-demand conditions in the traditional DRAM market.

2. $SanDisk Corp.(SNDK)$ 70% September Gain, Supply-Demand Gap to Persist Until 2026

• Outperforming Micron by a wide margin, its stock price rose 70% in September. Its CEO stated at a Goldman Sachs conference: "We expect the memory market to remain in short supply through 2026" and has already initiated a 10% price hike in some markets. Expectations of a supply-demand gap have become the core driver of its stock price.

3. $NVIDIA(NVDA)$ ’s New Chip Reshapes Memory Landscape, Samsung ($Samsung$) Emerges as Top Winner

• Rubin CPX Chip Launch: NVIDIA unveiled its new Rubin CPX chip this week. By "separating pre-fill and decode tasks + optimizing GPUs for different workloads," it achieved a 6x overall speedup while reducing token costs to drive broader adoption. According to Semianalysis, the chip is a "game-changer" for AI inference, widening the gap between NVIDIA and competitors in rack system design to a "canyon-sized" margin.

• GDDR7 Demand Surges: Since the CPX chip replaces expensive HBM with on-board GDDR7, NVIDIA has asked Samsung to double its GDDR7 supply. As the first manufacturer to mass-produce GDDR7 in H2 2024, Samsung—though lagging in GDDR5 and HBM 3/4—stands to reverse the "lagging in the GPU space" narrative with its leadership in GDDR7. Additionally, Samsung is expected to deliver HBM4 customer samples to NVIDIA as early as late September, further benefiting from tight DRAM/NAND market conditions.

• Valuation Upside: Using a 2026 Sum-of-the-Parts (SOTP) valuation—assuming 9x EBITDA for Memory, 8x for Foundry, 3x for Display Panels, 6x for Mobile, and 5x for Consumer Electronics—Samsung’s stock still has 60% upside (excluding potential outperformance in the memory business).

III. AI Applications: Gemini Takes the Top Spot, $Alphabet(GOOGL)$ s AI Narrative Fully Reversed

The AI application segment saw a "phenomenal" performance this week. Google’s ($GOOGL$) Gemini app, driven by Nanobanana, topped app rankings for the first time over the weekend, with explosive user and data growth:

• User Data: Gained 13M new first-time users in the past 4 days (total users exceeding 23M) and added 300M new images (total images surpassing 500M), making it the second-largest AI app after OpenAI’s ChatGPT.

• Narrative Reversal: Market perception of Google’s AI has shifted from "an AI loser two years ago" (due to the botched Bard launch) to "a long-term leader in the AI space." The book Genius Makers notes that Google’s AI expertise—built through DeepMind’s acquisition, in-house TPU (Tensor Processing Unit) development, and top-level cultural alignment—far exceeds previous market expectations, even driving valuation reratings for peers like $Baidu (BIDU)$.

IV. Macro & Market Strategy: Fed Rate Cut Expectations Meet AI Theme, Seize Winner Opportunities Amid Volatility

1. Macro: Fed Rate Cut Imminent, S&P 500 Historical Performance Offers Guidance

• While the AI supercycle is the current core market theme, Fed policy remains "dominant." The first (or third) year-end rate cut is likely to arrive this week.

• Historical data shows: Since 1965, the S&P 500 has delivered solid average returns following the Fed’s first rate cut in "soft landing" scenarios. If the U.S. economy avoids a recession within 1 year of the first rate cut (6 out of 7 cases since 1950), the market may fluctuate short-term but trend upward long-term, with an average 15% return over 12 months. $iShares 20+ Year Treasury Bond ETF(TLT)$

2. TMTB Strategy View: Volatile Upward Trend, AI Winners to Lead

• Market Outlook: The market will maintain a "volatile but upward" trend over the next month, ultimately led by AI industry winners (GPU/XPU, data centers, storage, power, AI networking) for a sharp rally.

• Trading Strategy: We significantly increased net long positions near short-term (ST) lows last Monday, and added short-term hedges on Friday (with plans to add more if the market rallies early this week). The core approach is to "buy on sharp dips and avoid chasing hot spots."

• Sector Divergence: While the broader market was roughly flat over the past month, TMT sector performance diverged sharply. The top 5 gainers included $SNPS$ (+198.4%), $SNDK$ (+84.5%), $FIG$ (+61.5%), $RUN$ (+53.2%), and $CIEN$ (+43.6%); the top 5 losers were $Opendoor Technologies Inc(OPEN)$ (-13%), $MDB$ (-30.5%), $GRPN$ (-22.2%), $LULU$ (-18.1%), and $MRVL$ (-14.8%). The gap between "AI winners" and "non-theme plays" continues to widen.

V. In-Depth Analysis of Key Stocks

1. $Reddit(RDDT)$ 100% Gain, Recommend Partial Profit-Taking

• Core Logic: The stock has doubled since our June recommendation, driven by LLMs citing RDDT more frequently than any other web asset in their responses—making it an "infrastructure-level data play for LLMs." Its CEO mentioned at a Goldman Sachs conference that the company "could secure tens of multiples in pricing when renegotiating LLM agreements."

• Risk Factors: U.S. Daily Active Users (DAUs) slightly missed expectations in August; SensorTower data shows slowing U.S. mobile DAU growth; OpenAI ($OAI$) CEO Sam Altman stated that "AI Reddit content feels far more inauthentic than it did 1-2 years ago," which may impact future licensing deal pricing.

• Rating Adjustment: Downgraded from "A/A+" to "B+ (B in the short term)." Partial profit-taking is recommended, though long-term optimism remains for potential settlements and licensing deals with Anthropic.

2. $Microsoft(MSFT)$ : Reasonable Valuation, Attractive Risk-Reward

• Valuation Calculation: A $475 stock price corresponds to 28x 2026 EPS ($17)—a level the stock has rarely fallen below since 2020 (except during market pullbacks). Upside to $600 (30x 2027 EPS of $20) still represents a 25% gain.

• Catalysts: Positive progress in OAI partnerships was disclosed on Friday; Yipit data shows Azure growth is on track to exceed 40%; MSFT’s stock price is closely correlated with the semiconductor sector, which has hit new highs—supporting further upside.

• Strategy Suggestion: Buy on dips below $500; selling $475 put options to enhance returns is also advisable.

VI. Preview of Key Events This Week

1. $Meta Platforms, Inc.(META)$ Connect Conference (Wednesday, 5:00 PM ET): Expected to showcase next-gen smart glasses and Meta AI updates. The market will focus on "how hardware + AI applications support long-term investments" rather than increased Reality Labs spending. We are optimistic about the glasses form factor ahead of the holiday season and have increased META positions this week, as Instagram engagement has seen a stepwise increase.

2. Corporate Investor Days:

   • Tuesday $Workday(WDAY)$ : Focus on the split between organic growth and Paradox-driven growth, FY27 profitability (30%+ operating margin, 30%+ free cash flow margin), and the rollout timeline for AI agents.

   • Wednesday $MongoDB Inc.(MDB)$ : The market expects clarity on 20%+ Atlas growth drivers and a long-term operating margin target (low-to-mid 20s) under the new CFO.

   • Wednesday $CrowdStrike Holdings, Inc.(CRWD)$ : Fal.Con briefing will focus on reconciling fast ARR growth in Cloud/Identity/Next-Gen SIEM with revenue ambiguity from partner rebates.

   • Wednesday $Monday.com Ltd.(MNDY)$ : Needs to validate "manageable traffic impact from AI search" and the role of new AI features in driving paid conversions to ease the "AI loser" narrative.

   • Thursday $Intuit(INTU)$ : Key focus areas include updated SMB demand insights and proof that agentic AI drives ARPU growth without relying on performance marketing.

VII. Summary & Outlook

This week’s "high-density rally" in the tech market is not an end but a prologue to the AI supercycle. Four key drivers—technological breakthroughs (NVDA’s Rubin CPX), tight supply-demand (memory), application growth (Gemini), and macro easing (Fed rate cuts)—are converging, pushing the industry from "two years of dizzying growth" into an "acceleration phase." Going forward, focus on the full AI industry chain (GPU/XPU, memory, data centers, applications), buy "certain winners" on market dips, and remain cautious of volatility in non-theme plays. Industry prosperity is likely to escalate further in late 2025-2026, making the current period a critical window for positioning in long-term tech opportunities.