Upstream performance high growth, market capitalization ladder differentiation, in-depth analysis of capital performance across the entire industry chain
<img src="https://imageproxy.pbkrs.com/https://inews.gtimg.com/om_bt/OqYnsmix3OggLVb7lxpUAaeAecFtcd6tGs3ZKKqNQEOOYAA/641?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="Image" original-src="https://imageproxy.pbkrs.com/https://inews.gtimg.com/om_bt/OqYnsmix3OggLVb7lxpUAaeAecFtcd6tGs3ZKKqNQEOOYAA/641"/>
From 2025 to the first quarter of 2026, the A-share artificial intelligence sector has entered a structural bull market driven by policy, explosive computing power, and domestic substitution. With the release of the 2025 annual reports and the first quarter forecasts for 2026, the industry's performance realization has significantly improved, and the differentiation between upstream and downstream of the industry chain is clear: upstream computing power hardware leads with certain growth, midstream chip design welcomes a profit turning point, and downstream applications fluctuate forward in commercial exploration. This article focuses on the A-share market, using core financial and market capitalization data as the basis to sort out the performance of leading companies in the AI track and restore the true capital logic of this round of market.
This round of AI market is centered on computing power, with upstream hardware companies becoming the first choice for funds, and leading companies' market capitalization and stock price increases leading the entire sector. As of early March 2026, the total market capitalization of the A-share AI core sector has increased by over 45% compared to early 2025, with the top five leading companies contributing more than half of the sector's incremental growth, becoming stabilizers for the track.
A-share AI Leading TOP5: Market Capitalization and Stock Price Performance from Early 2025 to March 2026
<ol>
<li>
FII: Absolute leader in AI servers, with a market capitalization of approximately 480 billion yuan at the beginning of 2025, reaching 1.07 trillion yuan by March 2026, an increase of about 123%. Revenue in 2025 was 902.887 billion yuan, net profit was 35.286 billion yuan, and AI server business grew more than three times year-on-year, achieving a double hit in performance and valuation, firmly in the first tier of technology stocks.
</li>
<li>
Zhongji Innolight: Leader in high-speed optical modules, with a market capitalization of approximately 180 billion yuan at the beginning of 2025, rising to 596.1 billion yuan by March 2026, an increase of about 231.1%. Net profit in 2025 was 10.799 billion yuan, a year-on-year increase of 108.81%, with 800G and 1.6T products ramping up, driven by both overseas and domestic computing power demand, becoming an upstream benchmark.
</li>
<li>
Cambricon: Leader in domestic AI chips, with a market capitalization of approximately 90 billion yuan at the beginning of 2025, reaching 458.4 billion yuan by March 2026, an increase of about 409%. Revenue in 2025 was 6.497 billion yuan, a year-on-year increase of 453%, with net profit turning from loss to profit at 2.059 billion yuan, driven by domestic substitution and intelligent computing procurement, pushing valuation recovery.
</li>
<li>
Haiguang Information: Core enterprise in high-end computing chips, with a market capitalization of approximately 110 billion yuan at the beginning of 2025, reaching 562.3 billion yuan by March 2026, an increase of about 411.2%. Revenue in 2025 was 14.376 billion yuan, net profit was 2.542 billion yuan, with both CPU and AI acceleration chips ramping up, and government and financial orders landing, showing steady growth
</li>
<li>
Tianfu Communication: A leading integrated optical device company, with a market value of approximately 42 billion yuan at the beginning of 2025, reaching 257.3 billion yuan by March 2026, representing an increase of about 512.6%. Revenue and net profit are expected to double in 2025, with new products like optical engines being launched, binding global optical module manufacturers, and fully enjoying the dividends of computing power expansion.
</li>
</ol>
Performance across the entire industry chain: upstream leads, midstream recovers, downstream differentiates
Upstream computing hardware is the strongest main line in 2025. Orders for servers, optical modules, storage, and liquid cooling are full, with high growth in performance being realized. Leading companies like FII, Zhongji Innolight, and Tianfu Communication see significant increases in revenue and profit, with steadily improving gross margins and a notable effect of capital clustering, resulting in an average increase of over 150% in the sector, making it the most certain direction in the entire market.
Midstream chip design benefits from accelerated domestic substitution. Companies like Cambricon and Haiguang Information have reached a profitability inflection point, with significant reductions in losses, supported by policy subsidies and downstream procurement, leading to rapid valuation recovery. The sector's average increase exceeds 80%, with institutional holdings continuously rising, becoming core targets for self-controllable technology.
The downstream application layer is in a commercialization ramp-up phase, with clear differentiation. Revenue growth is rapid in areas like large models, AI office applications, and industrial AI, but profitability still needs to break through, leading to significant stock price volatility. Some companies achieve valuation premiums through scenario implementation, but overall performance is weaker than upstream, presenting a pattern of &#34;earning revenue but not profit.&#34;
Capital logic and future trends
The core logic of the AI sector in 2025 is performance realization. Upstream hardware companies will first achieve high growth in revenue and profit, gaining valuation premiums; midstream chips will shift from the investment phase to the harvest phase, opening up space for domestic substitution; while downstream is waiting for business models to mature. Capital is laid out around the transmission path of &#34;computing power - chips - applications,&#34; forming a clear tiered market.
From an industry trend perspective, the global computing power arms race continues, with accelerated construction of domestic intelligent computing centers, and upstream hardware demand is expected to remain highly prosperous. Midstream chip domestic substitution is entering deep waters, with technological breakthroughs and large-scale applications driving continuous performance growth. Downstream applications will achieve breakthroughs in scenarios such as finance, manufacturing, and healthcare with the iteration of large models, opening up long-term growth space.
At the same time, the industry also faces challenges such as valuation differentiation and intensified competition. Some upstream companies have experienced phase adjustments due to excessive increases, and the market is beginning to focus on the sustainability of performance. Midstream chip companies need to continue breaking through technical bottlenecks, while downstream applications need to accelerate commercialization to support further valuation increases.
2025 is a key year for the A-share AI sector to shift from speculative trading to value investment. The computing power main line leads industry growth, with clear differentiation in performance and market value between upstream and downstream, while leading companies dominate with technological, customer, and scale advantages. In 2026, the AI sector will continue to exhibit structural trends, with stable growth in upstream hardware, accelerated substitution in midstream chips, and gradual breakthroughs in downstream applications, collaboratively driving high-quality industry development.
For investors, it is essential to focus on performance realization capabilities and technological barriers, grasping the certainty of computing power hardware, the growth potential of domestic chips, and the breakthroughs in application implementation. Against the backdrop of accelerating industrialization of artificial intelligence, the A-share AI track will remain a core main line in the capital market, with long-term value worthy of continued attention

300308

601138

688256

From 2025 to the first quarter of 2026, the A-share artificial intelligence sector experienced a structural bull market driven by policies, an explosion in computing power, and domestic substitution. Upstream computing hardware companies performed outstandingly, with significant market value growth, and the top five companies contributed more than half of the incremental growth. Companies like FII, Zhongji Innolight, and Cambricon saw substantial increases in both market value and stock prices, demonstrating the strong growth of the industry and capital logic

The main line of computing power leads to value reassessment: A panoramic review of A-share AI leaders in the 2025 earnings season