--- title: "AI-driven new industrial cycle, multiple logical resonances welcome investment opportunities" type: "News" locale: "en" url: "https://longbridge.com/en/news/279528748.md" description: "In the current market context, the semiconductor chip design sector presents good investment opportunities, as the AI-driven semiconductor industry is in the early to mid-stage of an upward cycle. The semiconductor industry exhibits significant cyclicality, influenced by product, capacity, and inventory cycles. In 2023, the chip industry cycle driven by AI has undergone significant changes, expected to last for 5 to 10 years, driving exponential growth in semiconductor sales and a sustained upward trend in industry prosperity" datetime: "2026-03-18T03:38:10.000Z" locales: - [zh-CN](https://longbridge.com/zh-CN/news/279528748.md) - [en](https://longbridge.com/en/news/279528748.md) - [zh-HK](https://longbridge.com/zh-HK/news/279528748.md) --- # AI-driven new industrial cycle, multiple logical resonances welcome investment opportunities In the current market context, why does the semiconductor chip design sector present good investment opportunities? The core reason is that we are currently in an upward cycle of the semiconductor industry driven by artificial intelligence, and we are in the middle to early part of this upward cycle. Here, we will reorganize the investment reference logic of the semiconductor industry. The semiconductor industry has obvious cyclical investment attributes, and its fluctuations are influenced by the overlapping effects of product cycles, capacity cycles, and inventory cycles. Historically, the semiconductor industry experiences a complete cycle every 4 to 5 years, with the upward cycle typically lasting 1 to 3 years. We have outlined the driving factors for each round of global semiconductor growth since 2000: the first round was the five years before 2000, driven by the internet bubble, characterized by product-driven growth, with the core investment logic being capacity expansion; then, in the second decade after 2000, around the time of the financial crisis, product growth weakened, and the pace of capacity expansion slowed, still belonging to a product-driven capacity expansion cycle; in the following second decade, it remained a product-driven capacity expansion cycle, mainly driven by the upgrade of mobile phones from 3G smartphones to 4G smartphones. After experiencing product-driven capacity expansion, the industry enters a corresponding capacity decline phase, which also confirms the significant cyclical investment attributes of the semiconductor industry. Why do we say that starting in 2023, the chip industry chain cycle driven by artificial intelligence has undergone significant changes? Because the artificial intelligence cycle that began in 2023 is different from previous cycles driven by product iterations; this time, it is a productivity revolution brought about by AI. The expansion cycle of AI infrastructure construction is expected to last 5 to 10 years. From the training of large cloud models to the gradual application of edge AI in fields such as mobile phones and intelligent driving, the sustainable growth space of this semiconductor cycle may far exceed any previous cycle. In 2023, global semiconductor sales experienced exponential growth, and there are currently no signs of this growth trend reversing, indicating that the prosperity of the entire semiconductor industry chain is continuously transmitting upward. The demand driven by AI large models is prompting cloud vendors to increase investments in data centers, which gradually transmits upstream to the semiconductor industry chain, even reaching the chip design phase that determines chip computing power and limits, serving as the core engine of technological innovation. This is also the core logic of downstream application demand gradually transmitting to the upstream chip design phase. Data source: Changjiang Securities Research Institute, Wind Next, we will elaborate on why AI can drive fundamental productivity changes and is expected to achieve medium- to long-term prosperity in the industry. According to IDC's estimates, global AI spending will reach $1.3 trillion by 2029, with spending related to AI servers and storage expected to exceed $750 billion. From 2024 to 2029, the year-on-year growth rate of global AI spending is expected to exceed 40% NVIDIA also believes that the global market size related to artificial intelligence infrastructure construction will reach 3 to 4 trillion yuan by 2030, with an annualized growth rate of about 45% from 2025 to 2030. The end of Moore's Law has shifted the computing model towards artificial intelligence infrastructure construction, and the development in related fields has transformed from a predictable growth rate to an exponential expansion rate. The core reason we emphasize the significant investment opportunities in the chip industry chain again is that the profit margin fluctuations in the chip segment are relatively large. The net profit of companies related to chip design can vary greatly; during periods of low industry prosperity, profit margins may be relatively low, but currently, during the ongoing upward phase of industry prosperity, profit margins can even exceed 70%. In the chip manufacturing field, which is highly related to chip design, TSMC forecasts an annualized growth rate of about 50% to 60% from 2024 to 2029, indicating that GPU growth is expected to continue in the coming years. This growth rate has led the market to have extremely high profit growth expectations in the AI computing field, especially in the chip design sector driven by innovative chip design breakthroughs, which has further spurred the increase in net profits along the industry chain and the further enhancement of valuations, becoming an important factor attracting capital allocation. We continue to analyze that the high prosperity of the artificial intelligence industry is essentially driven by demand. The high prosperity of the artificial intelligence industry drives cloud vendors to further increase capital expenditures for expansion based on market demand. According to Morgan Stanley's forecast, global capital expenditures for data centers will exceed $734 billion by 2026, with an annual growth rate of about 60%. In other words, the expansion of demand for artificial intelligence has further spurred high capital expenditures from overseas vendors, laying the foundation for the continued improvement of the prosperity of the artificial intelligence industry. For the prosperity of artificial intelligence and the cutting-edge field of AI computing, the chip design sector, as the core driving factor of innovation, is expected to achieve relatively certain growth through the demand transmission of the industry chain. All of our analyses above are aimed at demonstrating the positive transmission and promotion effect of the high prosperity of the artificial intelligence industry on the chip design sector. How can we know, or is there an intuitive data point to understand the proportion of chip design in the entire artificial intelligence field? We can establish a more intuitive concept of data proportion through a targeted breakdown of capital expenditures for a computing power platform. This model from NVIDIA represents the leading level of computing power clusters that NVIDIA can currently offer. This cabinet can operate as an independent computing power platform, equipped with 72 GPU chips and 36 CPU chips. By breaking down its capital expenditures, we can provide a more intuitive understanding: creating a high-end artificial intelligence computing power device shows a prominent gross profit proportion for chip design. We have listed the relevant data on the right side, showing that the gross profit proportion of GPU design reaches 31.5%. The capital expenditure proportion of chip design in leading computing power clusters is quite high Overall, chip design accounts for about 40% of the capital expenditure in the entire cutting-edge computing platform, making it the largest portion of capital expenditure in cutting-edge computing devices. This expenditure ratio also confirms that in the AI era, where computing power is king, chip design is an extremely important value high ground. Data source: Bernstein (Note: The value proportion of storage and HBM reaches 7.6% because, in addition to storage servers, GPUs and CPUs also contain a certain amount of storage.) After discussing the overseas market, let's further break down the domestic market. In fact, the development of domestic large models closely follows the global pace and possesses high competitiveness in the global AI large model field, even leading in certain areas. Currently, the resonance of AI hardware and software in North America is accelerating the formation of a closed loop in AI development. Looking ahead, the intelligent level of AI large models on the model side continues to improve, which is expected to drive higher demand for computing power. This undoubtedly creates targeted benefits for domestic chip design companies. Under the favorable supporting effect, we also know that after the relevant foreign enterprises' technology is strictly regulated, it further forces domestic chip design companies to carry out independent research and development, achieving a leap from "usable" to "well usable," and even at some critical points, they have the opportunity to compete with foreign enterprises. This is also the core reason why the science and technology innovation chip design ETF can seize investment opportunities in domestic substitution amid the wave of domestic replacement. Next, let's look at the end of AI applications, namely the edge side. The market generally believes that 2026 is expected to become the first year of commercialization for agents. AI applications are gradually extending from the cloud to the edge, which, by definition, is a further expansion of the boundaries of AI applications. The rapid development of downstream applications will continue to drive the iterative upgrade of midstream chip design and manufacturing, constantly exploring the capability boundaries of cutting-edge technologies, while also helping chip design companies further release production capacity, forming a virtuous cycle of "downstream demand driving midstream design and manufacturing upgrades." The chip design industry is gradually entering an important growth cycle. AI empowerment is not only reflected in the well-known cloud field but is also further penetrating into edge intelligent terminals. Thus, it can be judged that: in the cloud, the large-scale release of domestic computing power drives rapid growth in chip demand, especially as the demand for computing power and storage continues to rise; on the edge, large models are expected to redefine the value of intelligent terminals, and edge AI is no longer limited to fields such as automobiles, with various terminals including smartphones expected to become important carriers of computing power. Next, the core investment logic of science and technology innovation chip design can be summarized as a resonance of four factors: cycle, empowerment, elasticity, and domestic substitution: First, from the perspective of the cycle, the semiconductor industry is entering a new upward cycle driven by AI Second, AI is empowering the entire industrial chain, with GPUs and ASICs serving as the computing power infrastructure. Benefiting from the high growth in capital expenditures by overseas cloud vendors, related fields are expected to achieve approximately 60% high growth by 2026. The demand for core storage chips on the storage side remains tight, driving storage prices upward. It is expected that the storage demand driven by computing chips will reach a historical high by 2026. Third, the chip design industry itself has significant elasticity characteristics, with prominent light asset and high gross margin attributes. Under normal circumstances, the gross margin can reach 60%; during the upward phase of industry prosperity, the gross margin can even exceed 70%, significantly higher than other core segments of the semiconductor industry. This also lays the foundation for the tech innovation chip sector to have higher performance elasticity during the AI prosperity cycle. Fourth, against the backdrop of domestic substitution and facing external technological restrictions, the domestic industry continues to promote technological self-control, with independent innovation capabilities continuously improving. Domestic large model research and development, represented by Deepseek, Alibaba, and some Hong Kong-listed companies, continues to advance, continuously driving the demand for domestic computing power. Core domestic chip design companies are accelerating their transition from "usable" to "user-friendly," with market share expected to continue to rise. Coupled with the 20% price fluctuation mechanism of the STAR Market, this further amplifies the trading elasticity of the sector. Overall, companies in the chip design sector possess light asset characteristics, having higher performance elasticity and valuation elasticity during the upward phase of industry prosperity. Combined with the trading mechanism of the STAR Market, the elasticity advantage is even more pronounced. The Tech Innovation Chip Design ETF (subscription code: 589263) selects high-quality targets in chip design on the STAR Market, providing investors with an efficient tool to position themselves for the upward cycle of AI and capture the high elasticity direction of chip design empowered by AI. The Tech Innovation Chip Design ETF (subscription code: 589263) benefits from multiple logics such as the AI cycle, domestic substitution, and high elasticity, accurately capturing the growth dividends in the chip design field, making it a high-potential allocation target in the technology sector for 2026. Risk Warning: Investors should fully understand the differences between regular fixed investment and zero-sum savings methods. Regular fixed investment is a simple and practical investment method that guides investors to make long-term investments and average investment costs. However, regular fixed investment cannot avoid the inherent risks of fund investments, cannot guarantee returns for investors, and is not an equivalent financial management method to replace savings. Both stock ETFs/LOF funds belong to securities investment fund varieties with higher expected risks and expected returns, with expected returns and risk levels higher than those of mixed funds, bond funds, and money market funds. Fund assets invested in STAR Market and ChiNext stocks will face unique risks arising from differences in investment targets, market systems, and trading rules, which investors should be aware of. The short-term price fluctuations of the sector/fund listed are only for reference as auxiliary materials for the article's analysis and do not constitute a guarantee of fund performance. The short-term performance of individual stocks mentioned in the text is for reference only and does not constitute stock recommendations or predictions and guarantees of fund performance Daily Economic News ### Related Stocks - [SMH.US](https://longbridge.com/en/quote/SMH.US.md) - [159546.CN](https://longbridge.com/en/quote/159546.CN.md) - [588780.CN](https://longbridge.com/en/quote/588780.CN.md) - [562590.CN](https://longbridge.com/en/quote/562590.CN.md) - [588170.CN](https://longbridge.com/en/quote/588170.CN.md) - [SOXL.US](https://longbridge.com/en/quote/SOXL.US.md) - [159582.CN](https://longbridge.com/en/quote/159582.CN.md) - [561980.CN](https://longbridge.com/en/quote/561980.CN.md) - [XSD.US](https://longbridge.com/en/quote/XSD.US.md) - [512480.CN](https://longbridge.com/en/quote/512480.CN.md) - [159516.CN](https://longbridge.com/en/quote/159516.CN.md) - [PSI.US](https://longbridge.com/en/quote/PSI.US.md) - [159558.CN](https://longbridge.com/en/quote/159558.CN.md) - [159995.CN](https://longbridge.com/en/quote/159995.CN.md) - [562820.CN](https://longbridge.com/en/quote/562820.CN.md) - [560780.CN](https://longbridge.com/en/quote/560780.CN.md) - [588200.CN](https://longbridge.com/en/quote/588200.CN.md) - [512760.CN](https://longbridge.com/en/quote/512760.CN.md) - [SOXX.US](https://longbridge.com/en/quote/SOXX.US.md) ## Related News & Research - [Exaforce Raises US$125M Series B to Expand AI-Native Security Operations Platform](https://longbridge.com/en/news/286193730.md) - [Analog Devices in talks to buy Empower Semiconductor for $1.5 billion, Bloomberg News reports](https://longbridge.com/en/news/286846060.md) - [AI boom puts SK Hynix on the cusp $1 trillion market value](https://longbridge.com/en/news/286347462.md) - [KLA approves 10-for-1 stock split and dividend hike](https://longbridge.com/en/news/286340025.md) - [China's CXMT expects revenue to surge as memory chip demand soars](https://longbridge.com/en/news/286706506.md)