The world’s largest solar panel manufacturer is making an audacious wager that the future of its business lies not in the panels it sells, but in the electricity those panels generate, and the AI infrastructure that electricity can power.
Jinko has agreed with Zhongwei city in the Ningxia Hui Autonomous Region to develop a 24.5 billion yuan ($3.6 billion) AI data center complex, according to a report by Nikkei Asia and Yicai Global. The facility will span 530,000 square meters, roughly the footprint of 74 soccer fields, and is designed to deliver a total capacity of 1 gigawatt upon completion in 2028. Power will come directly from Jinko’s own solar generation assets, making the project one of the most vertically integrated AI infrastructure plays to emerge from China’s booming compute buildout.
From Panel Maker to Power Provider: Why Jinko Is Pivoting
The strategic logic behind the move is as much about survival as ambition. Jinko posted a net loss of 6.8 billion yuan in fiscal year 2025, a stark illustration of what industry observers have taken to calling solar “involution,” the destructive, margin-eroding price competition that has ravaged China’s photovoltaic sector over the past two years. Overcapacity, aggressive undercutting, and slowing export growth have left even market leaders bleeding red ink.
By contrast, demand for AI compute infrastructure in China is accelerating at a pace that is straining the country’s existing data center ecosystem. Hyperscalers, cloud platforms, and model developers are all scrambling for reliable, large-scale capacity. For Jinko, the calculus is relatively straightforward: its solar assets generate power that the market increasingly cannot absorb at profitable prices, while AI data center operators are willing to pay a premium for guaranteed, low-carbon electricity supply.
The pivot also addresses a structural problem plaguing China’s renewable energy sector. Solar curtailment, electricity generated by solar farms that cannot be fed into the grid and is simply wasted, hit 9 percent in the first quarter of 2026. By co-locating a power-hungry data center directly with its generation assets, Jinko can monetize electricity that would otherwise be curtailed, dramatically improving the economics of its solar operations while simultaneously entering the high-growth AI infrastructure market.
Zhongwei: China’s Emerging Desert Compute Hub
The choice of Zhongwei is not accidental. Situated in the arid heartland of Ningxia, the city sits at the intersection of several favorable conditions: abundant solar irradiance, low land costs, cooler ambient temperatures that reduce data center cooling loads, and proximity to China’s national power transmission infrastructure. It is, in short, an ideal location to build large-scale, solar-powered compute.
Zhongwei is also becoming something of a magnet for energy-adjacent AI infrastructure investment. China Datang Group, one of the country’s five major state-owned power generators, began operations at a wind and solar-powered data center in Zhongwei as recently as May 2026, signaling that major energy incumbents are pursuing the same convergence thesis as Jinko.
Both projects are beneficiaries of China’s national Eastern Data, Western Computing initiative (东数西算), a policy framework designed to shift data processing workloads from the energy-constrained, land-scarce eastern coastal cities to the resource-rich western interior. The initiative has effectively turned provinces like Ningxia, Inner Mongolia, and Guizhou into strategic compute corridors, with central government support for grid infrastructure, land allocation, and preferential electricity pricing providing the policy tailwind that makes projects of this scale financially viable.
The Scale of China’s AI Infrastructure Buildout
Jinko’s move is a single, highly visible data point in a much larger investment wave. Chinese technology companies and energy groups are committing capital to AI data center infrastructure at a pace that rivals, and in some metrics exceeds, the buildout underway in the United States. Alibaba Cloud’s AI revenue hit 9 billion yuan in the first quarter of 2026 alone, while Citi analysts have projected that figure could grow 90 percent annually to reach 585.5 billion yuan by fiscal year 2031. That kind of demand trajectory requires enormous quantities of physical compute infrastructure, and the race to supply it is intensifying. Meanwhile, Alibaba has launched a 10,000-card AI computing cluster powered by its domestically developed Zhenwu AI chips, underscoring that the hardware and infrastructure layers of the AI stack are developing in parallel.
ByteDance, for its part, raised its 2026 AI capital expenditure to $30 billion and is actively shifting orders toward local AI chips, decisions that will require substantially more data center real estate to house the resulting compute clusters.
Jinko’s 1 GW project, if delivered on schedule, would represent a meaningful addition to China’s national AI compute capacity at a moment when demand is outpacing supply.
Financing a $3.6 Billion Bet While Carrying Losses
The financial optics of the deal deserve scrutiny. A company that lost 6.8 billion yuan in a single fiscal year is now committing to a 24.5 billion yuan capital project, a sum that will require substantial external financing. Jinko has not publicly detailed the funding structure for the Zhongwei development, though arrangements of this kind in China typically involve a combination of local government capital, state policy bank lending, and project-level equity from strategic partners.
The involvement of Zhongwei city as a counterparty to the agreement suggests local government support will be part of the equation. Municipal governments across western China have strong incentives to attract data center investment, which brings grid revenue, tax base, and employment — making co-investment or preferential financing arrangements a standard feature of projects anchored in the Eastern Data, Western Computing framework.
What is clear is that Jinko’s management has made a deliberate strategic judgment: the solar panel business, at least in its current form, does not offer a path back to profitability fast enough to justify waiting. The AI infrastructure market, powered by the company’s own renewable generation assets, does. Whether the execution matches the ambition will become clearer as 2028 approaches and the first sections of the Zhongwei facility are scheduled to come online.
For now, Jinko’s desert pivot stands as one of the most consequential and revealing examples of how China’s AI buildout is reshaping industries far beyond the technology sector itself.