As China’s technology giants and well-funded start-ups engage in a fierce battle for supremacy in the foundation model space, a different group of companies is quietly reaping the most immediate financial rewards. The physical infrastructure suppliers, the companies manufacturing the printed circuit boards, optical modules, and power systems that make artificial intelligence possible, have emerged as the undisputed, near-term winners of the AI revolution, posting triple-digit profit growth and dominating domestic stock market returns.
While high-profile AI developers like DeepSeek and Moonshot AI capture the headlines with massive funding rounds, the “pick-and-shovel” providers of the AI gold rush are generating transparent, straightforward revenue. According to a recent analysis by The Wire China, the capital expenditure cycle driven by hyperscalers is flowing directly into the physical infrastructure supply chain, creating a boom for specialized hardware manufacturers.
“The more obvious near-term AI trade is often the enabling layer—power, semiconductors, cloud, telecom infrastructure, data centers and hardware suppliers,” Philip Reschke, a Hong Kong-based investor, told The Wire China. “Those businesses can benefit from broad AI adoption regardless of which individual model or application developer ultimately wins.”
The Infrastructure Boom
The financial performance of these hardware suppliers stands in stark contrast to the heavy losses often sustained by companies developing large language models (LLMs). The infrastructure companies are highly profitable and model-agnostic, selling into a market where demand vastly outstrips supply.
Victory Giant Technology, a Huizhou-based manufacturer of printed circuit boards for AI servers, exemplifies this trend. The company, which supplies circuit boards used in Nvidia’s servers, saw its revenue nearly double to 19.3 billion yuan ($2.8 billion) last year. Its Shenzhen-listed shares have risen twenty times over the last three years. In April 2026, Victory Giant completed a massive initial public offering on the Hong Kong Stock Exchange, raising HK$20.1 billion ($2.6 billion) in what was the city’s largest listing of the year to date.
Similarly, Kingboard Laminates, a rival printed circuit board manufacturer, reported that its profit jumped 84 percent last year to reach HK$2.4 billion ($300 million), driving triple-digit gains in its stock price this year.
The optical module sector, critical for the high-speed data transfer required in AI data centers, is experiencing an even more dramatic surge. Zhongji Innolight, a Shenzhen-listed optical module maker, reported that its first-quarter 2026 revenue nearly doubled year-on-year to 19.5 billion yuan ($2.8 billion). More impressively, its profit increased 262 percent to 5.7 billion yuan ($840 million). The company’s shares have seen a nearly tenfold jump over the past year.
According to Bank of America Global Research, optical module companies are expected to be the most profitable segment in the AI supply chain this year, with projected returns on equity of around 40 percent.
“For hardware companies, what investors like is the earning growth and that monetization is very transparent and straightforward,” noted Winnie Wu, head of Asia Pacific equity strategy at Bank of America Global Research.
Pivoting to Power the AI Revolution
The AI hardware boom is also prompting established industrial companies to pivot their business models. Weichai Power, a Shandong-based company traditionally known for manufacturing car and truck engines, has successfully transitioned into providing power solutions for AI data centers.
As the energy demands of advanced computing facilities skyrocket, a trend driving the integration of data centers into virtual power plants, Weichai’s new segment has seen explosive growth. The company reported three-digit growth in sales of its data center power products in the first quarter of 2026, and its shares have doubled over the past six months.
“The money that the hyperscalers are spending is flowing into the physical infrastructure supply chain,” Eric Wong, founder of the U.S.-based hedge fund Stillpoint Investments, told The Wire China. “It is a very comprehensive supply chain that’s benefiting from this capital expenditure cycle.”
The Contrast with Big Tech and Start-ups
The soaring valuations of hardware suppliers highlight a divergence in the market’s assessment of the AI ecosystem. While infrastructure providers post tangible profits, the major technology platforms and pure-play AI start-ups face a more complex financial reality.
Despite their massive investments in AI, traditional tech giants have seen their stock prices struggle. Alibaba’s shares have fallen by around a quarter from their peak in October, and Tencent’s shares have dropped by 25 percent since the start of the year, as investors weigh the massive capital expenditures required to compete in the AI space against the uncertain timeline for monetization.
Meanwhile, the start-up ecosystem remains heavily reliant on venture capital to fund the exorbitant costs of model training. Moonshot AI recently raised $2 billion in a funding round led by the venture capital arm of Meituan, and StepFun secured $2.5 billion from industrial investors as it prepares for an IPO. DeepSeek is reportedly in talks with a national AI fund as a lead investor, potentially valuing the company at up to $50 billion.
However, as Dermot McGrath, a tech investor in Shanghai, pointed out, the hardware suppliers avoid the intense competition and high cash burn of the model layer. “The profitable infrastructure companies—storage, packaging, equipment—are model-agnostic,” he said. “They sell into demand regardless of which lab is on top this month.”
As China continues to build out its national computing power network and domestic AI chip ecosystem, the demand for domestic hardware infrastructure will only intensify. For investors seeking clear, near-term returns in the AI sector, the quiet winners manufacturing the physical building blocks of the revolution have proven to be the most lucrative bet.