[Red Envelope] Big Beauty and Persian Cat go to war. Where will the big A go?

[Taoguba]
The blogger has been a full-time trader for 6 years with 11 years of trading experience. Skilled at perceiving subtle changes in the market, with a forward-looking sense of market patterns. Good at analyzing the rhythm of indices, themes, and market sentiment; low-entry buy patterns are becoming more refined. Among Taoxian’s low-entry bloggers, I hold my own place! Stable monthly profit of 20%, compounded annually is 8 times. Stable monthly profit of 15%, compounded annually is 4 times. Stable monthly profit of 10%, compounded annually is 2 times. The blogger’s desire is not high; around 10% monthly stability is enough. Anything more is just gratitude for the market’s feedback! Maintaining a good mindset, treating trading as a joyful activity—why not enjoy it?

The best offense is defense: control emotional rhythm, flexibly manage positions, strictly divide positions, and control drawdown risks. Slow is fast; compound interest will expand your imagination! The so-called leading stock strategy of relay trading—how many leaders can there be in a year? That’s a road to no return. Nine out of ten traders fail; surviving one is a genius!

The blogger successfully transitioned from a trader to a leader in the market. No need for quick trades here—relying solely on market understanding! This will open your new horizons. Trading is fundamentally a contest of market cognition. Only by standing shoulder to shoulder with high-awareness traders can you see the market’s essence, avoid risks, and grasp the flow of wealth.

Our connection is fate. There are countless ways of thinking, but only through your frequent likes, tips, and encouragement can you stay on track. Giving generously—only by giving can you receive. Your support is also recognition of Brother Ding’s ideas. Brother Ding’s pre-market insights are truly golden, waiting for you to discover! I sincerely wish all old and new friends a red-hot account in 2026, climbing higher and higher!

Back to the main topic

1. Weekly case review summary:

On the first trading day, the fermentation of SD 2.0 and token overseas listing, driven by holiday effects, was cashed out by funds. Quantitative focus shifted to precious metals, non-ferrous metals, and chemicals. Before the holiday, AI applications and computing power were under pressure. Happy Century’s bid was canceled; on February 12, Capital Online unfollowed. The last trading day before the holiday, macroeconomic tech was unfollowed—Hongjing Technology. The market lacked a clear main line; focus was on various branches, with intraday attention on Jiangbolong and Wangbian Electric. For AI computing power, focus on Dawan Technology at the limit down, and on Wangshu Technology in deep water.

On the second trading day, the market fermented around phosphorus chemicals and non-ferrous metals, NVIDIA’s industry chain including optical communications, PCB, and electronic glass fibers, with some movement in commercial aerospace. Domestic computing power stocks opened with some volatility and were cashed out. The focus was on two bidding signals in phosphorus chemicals, arbitraging Chuanjinnuo. In commercial aerospace, lurking in Western Materials and Xinwei Communications. Domestic computing power: Dawan Technology and Wangshu Technology unfollowed. Wangbian Electric maintained attention. The first two days, the rhythm was not well caught.

On the third trading day, phosphorus chemicals were cashed out, small metals continued to rise, PCB strengthened, domestic computing power weakly recovered, and sentiment was stronger than indices. Focus was on spreading small metals and domestic computing power. New attention on Yunnan Germanium and Chi Hong Zinc Germanium. For domestic computing power, focus on Runze Technology and Huasheng Tiancai, with Capital Online in intraday attention. Chuanjinnuo, Western Materials, and Xinwei Communications unfollowed; Wangbian Electric maintained attention. The third day’s rhythm was regained.

On the fourth trading day, power and non-ferrous metals continued to rise, NVIDIA’s industry chain was cashed out, and domestic computing power surged then fell back. The focus was on pre-market expectations that domestic computing power would further strengthen, with new attention on Huawei Ascend, Taijia Shares, Capital Online, Huasheng Tiancai, and Runze Technology. After opening, computing power retreated, with three targets rolling T+0. Small metals like Tungsten of Xianglu, attention on Yunnan Germanium. Unfollowed Wangbian Electric and Chi Hong Zinc Germanium. Early in the session, full positions; in the afternoon, Wanshu Technology plunged, dragging the sector down. Huasheng Tiancai opened with a limit down; to avoid risks, reduced positions and unfollowed all holdings. With no clear main line, the approach remained spreading and rolling, flexible for attack or defense.

This week’s XI class pause due to limited energy and high pressure. Focused on personal trading; remaining energy will continue to support your long-term support. I will carefully prepare pre-market ideas and share thoughts during the day. Wishing all old and new friends a red-hot account next week!

2. Weekly pre-market and intraday idea review

3. Next week’s index and theme projection

4. Shanghai Composite Index outlook

Big gains and the Persia Cat event are only short-term impacts; China A-shares have their own rhythm. Last Friday, NVIDIA saw its largest single-day drop since April last year. For China A-shares, it only affected the bidding. The Shanghai Index opened down 0.43%, with an initial low of 4128 points, then gradually recovered to close up 0.39% at 4162 points, with a volume of 2.49 trillion yuan, slightly reduced by 504 million. The index exceeded expectations. Even without external factors, the previous high near 4190 remains a significant resistance. Without 3 trillion yuan in volume support, a divergence near the previous high may cause a correction. External influences over the weekend may cause short-term pressure; the Big Gains and Persia Cat events are not surprises—just the landing of the shoe. 4100 points will serve as a defense line, with major financials expected to support the market. Even if it falls below 4100, it won’t affect the overall rhythm; at most, it tests the third leg.

2. Key theme projection outlook

(1) Oil & Gas, Precious Metals, Small Metals.

The Big Gains and Persia Cat event are expected; they directly benefit oil & gas and precious metals. However, many funds are lurking, so caution is advised. Observe how many one-word bids appear and the strength of the order book. Without the first mover, the second may cash out quickly—be cautious. The logic of strategic resources like germanium, tungsten, antimony, gallium remains unchanged, with prices rising and becoming strategic resources. The start of the year sees the strongest continuous rise in these sectors, especially tungsten with huge gains. Focus on relatively low positions in small metals like germanium, antimony, gallium, and indium.

Using indium phosphide (InP) and indium as examples: InP is a second-generation III-V compound semiconductor, with advantages like direct bandgap, high electron mobility, emission at 1.3–1.6μm, and radiation resistance—considered a “golden substrate” for optical communications and high-frequency RF.

Optical Communications (largest application, AI computing needs)

• Core high-speed optical modules: the only practical substrate for 800G/1.6T/3.2T modules
• Transmit: DFB/EML lasers (1310/1550nm)
• Receive: PIN/APD photodetectors
• Components: Modulators, optical amplifiers, silicon photonics
• Scenarios: AI data centers, 5G/6G base stations, backbone networks, submarine cables

High-end RF (6G, satellites, radar)

• Ultra-high frequency devices: 100GHz+ millimeter wave/THz, outperforming GaAs/GaN
• HBT/HEMT transistors, MMIC microwave integrated circuits
• Applications:
  • 6G/millimeter wave base stations, satellite communications (Starlink/Beidou)
  • Phased array radar, electronic warfare, missile guidance
  • Radio astronomy, deep space exploration

Intelligent driving and sensors

• LiDAR: 1550nm solid-state LiDAR core light source/detector supporting over 250m high-precision detection, suitable for L3+/L4 autonomous driving
• Millimeter-wave radar: 77/79GHz vehicle-mounted front-end

Aerospace and National Defense

• Radiation-resistant solar cells: for space/satellites, efficiency 15%+ higher than silicon, used in Jilin-1 and NASA Mars exploration
• Space photodetectors: infrared/visible sensors for extreme environments

Quantum Technology

• Quantum dot lasers, single-photon sources: core hardware for quantum computing/communication
• Quantum sensing: nanometer displacement/magnetic field detection, used in bio-imaging and geological surveys

In the US stock market, indium phosphide (InP) has increased 8 times in 4 months, with comparable benchmarks. Key stocks include:

Yunnan Germanium Industry (002428)

• The only domestic mass producer of InP substrates, subsidiary XinYao Semiconductor (Huawei Hubble invested)
• Capacity: 150,000 wafers/year now, expanding to 300,000–400,000 by 2026; 6-inch yield over 70%
• Clients: Huawei, Guangxun, Sanan; domestic market share expected to exceed 50% by 2026
• Logic: AI optical modules + 6G RF demand, global shortage over 70%, prices and volume rising

Tin Industry Co. (000960)

• Global indium leader: indium reserves 4,821 tons (world’s largest), annual capacity 127 tons (over 11%)
• Controls Hualian Zinc Indium (largest native indium base), full industry chain self-sufficient (mining→smelting→high-purity indium→targets)
• Core logic: indium price elasticity highest, driven by AI optical modules, panels, and photovoltaics

HuaXi Nonferrous (600301)

• Large Guangxi mine, indium reserves 3,500 tons (29% nationwide), annual capacity 50 tons
• Technical barrier: 7N ultra-high purity indium mass production, supplies semiconductors, optical modules, high-end targets
• Core logic: scarcity of high-end indium materials, benefiting from AI computing and phosphorus chemical expansion

ZhuYe Group (600961)

• Zinc smelting by-product recovery, annual capacity 60 tons, recovery rate over 95%, purity 5N–6N
• Suppliers for BOE, TCL, with strong cost advantage in recycled indium
• Core logic: indium price increases almost entirely profit-driven, short-term elastic leader

Zinc Industry Co. (000751)

• Key supplier of high-purity indium in Asia, annual capacity 30 tons, purity up to 8N
• Market share for ITO targets over 30%, costs 15% lower than industry average
• Core logic: combined demand from panels, photovoltaics, and semiconductors

Yuguang Gold & Lead (600531)

• Recovered from lead-zinc by-products, small batch production of 7N ultra-high purity indium, suitable for semiconductors and optical modules
• Leading technology, a hidden champion in high-end indium materials, small but highly elastic

Longhua Technology (300263)

• Annual capacity of 50 tons of high-purity indium, 15% market share of ITO targets, gross margin over 60%
• Full industry chain: recycled indium → high-purity indium → targets, partnering with CATL and Longi
• Core logic: shifting from raw material sales to high-value products, with strong performance certainty

(2) Token overseas expansion (directly ask Doubao)

Token overseas = packaging China’s AI large models’ inference computing power into tokens (AI computation units), sold via API/cloud/open source globally, enabling “electricity not leaving the country, value crossing borders” digital service export.

 Essentially: digital “power + computing power export”

• Pack China’s low-cost green electricity + scaled computing power + optimized models into measurable tokens, sold worldwide
• Zero tariffs, zero logistics, zero inventory—digital service trade

Two main models (2026 mainstream)

1️⃣ TaaS (Inference as a Service, main profit model)

• Mode: overseas users connect directly to Chinese API/cloud nodes, pay per token
• Computing power location: domestic/overseas Chinese AI centers handle inference
• Examples: MiniMax, Kimi, Zhipu GLM, DeepSeek open global API
• Advantages: earn foreign cash directly, control quality, rapid scaling
• Challenges: cross-border latency, compliance, payments, SLA guarantees

2️⃣ SaaS (Standard as Ecosystem, light asset expansion)

• Mode: overseas download Chinese open-source models (e.g., GLM-5), local/overseas deployment and fine-tuning
• Tokens: produced overseas, China exports model standards and ecosystem
• Examples: Tongyi Qianwen, Llama 2 China version, Kunlun Wanyuan open ecosystem
• Advantages: light assets, no latency, broad penetration, global tech discourse
• Challenges: weak profitability, high ecosystem maintenance costs

Why can Chinese tokens go overseas (core advantages)?

1️⃣ Cost advantage (most critical logic)

• Western green electricity (0.2 yuan/kWh) + scaled AI computing, token unit cost only 1/20 of the US
• Inference costs only 1/3–1/5 of OpenAI/Gemini, performance close to top tier

2️⃣ Scale surpasses the US (inflection point in Feb 2026)

• OpenRouter data: China models’ weekly calls hit 5.16 trillion tokens, first surpassing the US
• Global share: 61%, top three are all Chinese models (MiniMax, Kimi, Zhipu)

3️⃣ Compliance + duty-free channels

• Digital service export, WTO electronic transmission zero tariffs
• Cross-border data, foreign exchange, cybersecurity compliance only—no physical trade barriers

4️⃣ Engineering + ecosystem advantages

• Weekly iterations, strong agent/code capabilities, suitable for global developers
• Open-source ecosystems rapidly expanding on Hugging Face, GitHub

Complete industry chain for token overseas (benefiting A-shares)

1️⃣ Model layer (“money printing machine”)

• Domestic large models: MiniMax, Kimi, Zhipu, DeepSeek, Kunlun Wanyuan (Tian Gong)
• A-share related: Kunlun Wanyuan (300418), Zhejiang Wenhulian (Kimi), Daily Interactive (DeepSeek)

2️⃣ Computing power leasing / IDC (core necessity)

• Provide GPU clusters, intelligent computing centers, support token inference
• A-shares: Hongbo Co. (exclusive for MiniMax), Capital Online (Zhipu), Runze Tech (ByteDance/Tencent), Xiechuang Data (overseas Wanka)

3️⃣ Cross-border network / CDN (“traffic toll station”)

• Solve cross-sea latency, ensure stable calls
• A-shares: Wangsu Technology (over 2800 nodes globally), 263 (cross-border private network + submarine cables)

4️⃣ Hardware base (domestic substitution)

• AI servers, chips, optical modules, liquid cooling, power supplies
• A-shares: Tuowei Info (Ascend), Haiguang Info, Cambrian, Inspur, Zhongji Xuchuang (optical modules)

Core impact on A-shares (latest 2026)

1️⃣ Mainline established: computing power + overseas expansion + domestic substitution threefold resonance

• Token overseas = explosive demand for computing power + accelerated domestic substitution + global pricing power, becoming the strongest mainline in 2026

2️⃣ Clear performance realization path

• Model makers → computing power leasing → IDC → hardware → power, full industry chain upward trend
• Companies with high overseas revenue share (e.g., Hongbo, Xiechuang, Wangsu) will be the first to realize profits

3️⃣ Valuation restructuring

• From “domestic AI” to “global AI service provider”, valuation benchmarked against overseas SaaS/computing companies
• Computing leasing and cross-border network have the greatest valuation elasticity

4️⃣ Sector rotation order

• Short-term: computing leasing, cross-border network (direct beneficiaries)
• Mid-term: AI hardware, optical modules, liquid cooling (order fulfillment)
• Long-term: green power, power equipment (expanding computing infrastructure)

Token overseas, surged on Friday then retreated. My personal understanding is that the sector is not finished; it will fluctuate repeatedly. Especially on Friday, Big Gains, NVIDIA’s stock plunged another 4%. Correspondingly, the NVIDIA industry chain in China was only affected in bidding. Tianfu Communications resisted stubbornly in deep water and turned red. So, on Monday, it might not be so lucky; the industry chain will face pressure again, benefiting domestic computing power. After repeated fluctuations, token overseas expansion has a high probability of becoming a mainline. The strongest fermentation on Friday was Huawei Ascend, with strong sentiment targeting Taijia Shares. Even if external negative factors impact Monday, the sector’s rebound may be limited. Low divergence in core recognition remains a valid approach.

How to respond on Monday? See the pre-market idea projection around 8 am Monday morning. Brother Ding’s pre-market insights are truly golden—waiting for your discovery! I hope everyone, please click a free like with your little finger, and your support with 100 points or a fuel coupon makes me even more grateful. A little motivation will push me to keep reviewing and giving back to everyone.

Disclaimer:
Case review is only for personal review and idea recording!
Investing involves risks; trade cautiously. Plans are never faster than the market; follow the market movements!
This article reflects my understanding of the market and is for personal sharing only!
It does not constitute any investment advice; buy and sell at your own risk!

Thanks to the last article’s supporters—your encouragement and support are appreciated. Wishing you daily limit-ups and accounts as red as koi! No losses!