In many articles from "Value Institute," the director consistently emphasizes that we must, without a doubt, pay close attention to opportunities related to AI, as it is certain to bring earth-shattering changes to our world. Taking the director as an example, honestly, writing articles now is inseparable from AI. Whether it's writing the in-depth corporate and industry analysis for "Value Institute" or the usual casual essays for "Value Institute Max," AI significantly boosts efficiency and improves the quality of the content. Furthermore, AI is advancing rapidly; the director can clearly feel it becoming more useful, playing an increasingly important role in both work and daily life. A couple of years ago, the public's understanding of AI was largely limited to large language models like ChatGPT that assist with thinking. However, in just these past two years, practical AI applications have noticeably multiplied. For instance, the recently launched Doubao AI phone that caused a stir across the market was immediately touted online as an AI application capable of颠覆 all mobile apps, becoming the ultimate portal on a phone. The speed of AI progress is indeed astonishingly fast, even alarmingly so. Two years ago, we might have considered models like Wenxin Yiyan rudimentary, but today, their mature practical applications are already emerging. Isn't that somewhat terrifying? Yet, from another perspective, this frightening pace of AI evolution precisely signifies massive investment opportunities. Products that bet on AI's long-term potential, such as the CSI SSE STAR Market and ChiNext AI Index ETF Invesco (159142), are highly值得关注.
Previously, when the director visited a retail company for research, they explained how they use AI for predictive inventory management. They have AI forecast which products will sell well in the coming period based on sales trends, seasonal factors, and even trending topics on social media. This allows for proactive inventory optimization, ensuring sufficient stock is available when customers need it, thereby avoiding losses from stockouts and guaranteeing a good customer experience. Because this system delivered substantial economic benefits, the company proudly elaborated on their AI-powered inventory management during the research visit. You see, the practical application of AI is happening this quickly and delivering immediate results. No wonder Cathie Wood previously predicted that by around 2030, AI could drive the global economic growth rate to more than double, reaching 7.3%. In contrast, over the last 100 years, which we consider the period of fastest human technological progress (1900-2024), the global economy's compound annual growth rate was only about 3%.
We are currently at a historic inflection point where AI is颠覆ing the world, a feeling most people likely share. Consequently, leading companies and even nations are continuously making strategic moves in this area. In capital markets, the primary focus for top tech firms is no longer just their current earnings, but rather their capital expenditures related to AI. Both domestically and internationally, it's evident that capital spending on AI by major tech giants has surged in recent years. In Q2 2025 alone, the combined capital expenditure of Microsoft, Amazon, Meta, and Google increased by 69% year-over-year to $87.4 billion. According to FactSet consensus estimates from July 30, the total capital expenditure of these four cloud providers for 2025 is projected to grow 50% year-over-year to $333.8 billion. Therefore, today we discuss another company closely tied to AI – Zhongji Innolight Co.,Ltd.. Zhongji Innolight manufactures high-speed optical communication transceiver modules,简称光模块. What exactly do they do? Simply put, they act as converters between optical and electrical signals, primarily enabling high-speed data transmission through fiber optics while ensuring the data can be recognized and processed by equipment like servers and routers. For example, everyday WiFi often relies on fiber optic internet. When we browse the web or stream videos, data uploads and downloads essentially travel as light signals. However, devices like computers, phones, and servers can only interpret electrical signals, such as voltage levels representing 0s and 1s; they cannot understand light signals directly. Thus, optical modules function as translators and converters, facilitating the mutual conversion between optical and electrical signals. So, you understand now how crucial optical modules are in the era of big data. Amid the AI wave, optical modules represent one of the most directly受益 upstream "picks and shovels" segments, with their downstream customers being telecom companies or cloud computing giants like Amazon and Google.
Zhongji Innolight describes itself in its financial reports as providing "high-speed optical modules such as 200G, 400G, 800G, and 1.6T for cloud data center clients, offering 5G fronthaul, midhaul, and backhaul optical modules for telecom equipment manufacturers, and supplying optical components for metropolitan area networks, backbone networks, core network transmission, as well as FTTx fiber access." Driven by strong demand for high-speed optical modules from AI infrastructure construction and other factors, LightCounting predicts that the global Ethernet optical module market will continue rapid growth. It is forecast to grow 35% year-over-year to $18.9 billion by 2026, with growth rates remaining in the double digits from 2027 to 2030, potentially exceeding $35 billion by 2030. Frankly, the director believes LightCounting's forecasts are conservative, because the AI construction wave is far more powerful than imagined, and the demand growth from related companies is not linear but exponential! Those unfamiliar with this concept should look at the data in the following chart. Taking this year as an example, Google's average monthly token processing volume increased by 480 trillion in May, and just two months later, in July, it grew by 980 trillion. Look at this token growth rate – it's clearly exponential. Therefore, the capital expenditures of these cloud providers are also growing exponentially. Alibaba recently stated in investor communications that "the pace of deploying AI servers simply cannot keep up with the growth rate of customer orders, and the backlog of orders on hand continues to expand."
Downstream AI demand is growing exponentially, so upstream demand for optical modules should also be exponential, with a particularly urgent need for high-end modules. Zhongji Innolight recently stated in its financial report: "Key customers further increased capital expenditures to invest heavily in computing infrastructure, leading to significant growth in demand for high-end optical modules like 800G and accelerating the technological iteration towards 1.6T and higher speeds. During the reporting period,得益于 the rapid increase in the shipment proportion of high-end products like 800G and continuous cost reduction and efficiency improvements, the company's operating revenue, gross profit margin, and net profit further improved." Higher-end products naturally come with higher profit margins, which is why Zhongji Innolight's performance has been soaring recently. However, judging by future trends, even after this surge, Zhongji Innolight's future can be said to be just beginning, as the vigorous development momentum of AI itself is also just starting. Those who don't understand should look at the following chart (HTI's forecast of Zhongji Innolight's optical module business revenue composition) – understanding it will clarify the director's point! Furthermore, in the optical module field, Zhongji Innolight possesses considerable strength. As early as 2021, Zhongji Innolight ascended to become the world's largest optical module manufacturer and has maintained its leading position since (the light blue company marked in the chart below is Zhongji Innolight). Zhongji Innolight was the first manufacturer globally to release 400G/800G optical modules, 1-2 years ahead of peers; it released 1.6T optical modules in 2023 and currently has the fastest progress in the industry. In other words, Zhongji Innolight can be likened to TSMC in the chip manufacturing sector. So, you tell me, isn't Zhongji Innolight poised to win big? Isn't it a company worth significant attention in the AI era?
However, having said that, we must return to a familiar adage: while Zhongji Innolight is likely to secure a large slice of the AI pie, risks remain. After all, any company carries uncertainty, especially in the high-tech sector. Therefore, for such rapidly iterating high-tech companies, a more universally applicable investment logic involves focusing on index-based products that track the core sector. This approach allows for broader coverage to capture the growth红利 of the entire industry while better分散风险. If we're discussing what might be better than Zhongji Innolight and更能受益 from the massive AI innovation wave, it would be targets like the CSI SSE STAR Market and ChiNext AI Index ETF Invesco (159142). The CSI SSE STAR Market and ChiNext AI Index ETF Invesco (159142) tracks the CSI STAR Market and ChiNext AI Index, launched in May 2025. Its purpose is self-evident from the name, requiring little further explanation from the director. Its best feature is spanning both the STAR Market and ChiNext Board "双创" sectors, integrating the STAR Market's strength in core technologies with the ChiNext Board's high growth potential and application vitality. Compared to the STAR Market AI Index, it overweight optical modules by over 30%, heavily weighting industry leaders, allowing it to firmly grasp the确定性需求 from the AI computing explosion. Compared to the ChiNext AI Index, it overweight AI chips by over 20%, also heavily weighting leading companies, locking in the core of the intelligent computing industry from the source. The aforementioned Zhongji Innolight is its top holding with a weight exceeding 20%. Let the results speak for themselves: since its base date (2019/12/31), the CSI STAR Market and ChiNext AI Index tracked by this ETF has accumulated a gain of approximately 200%, significantly outperforming major indices like the CSI 300. Year-to-date, the index has also doubled, arguably performing the best among similar AI indices. Therefore, compared to individual companies, well, the director reiterates a preference for ETFs covering the relevant sectors and recommends everyone keep a close watch on them.
Comments