China shipped roughly 80% of the world's PEM fuel cell capacity by 2024, and stack system costs have dropped from over RMB 30,000/kW in 2015 to below RMB 4,000/kW in 2024, making Chinese stacks the lowest-cost option in the world for buyers who can manage the supply interface correctly [S1].
Procurement, however, is no longer about price discovery: it is about matching a stack's membrane chemistry, rated output, and operating envelope to the duty cycle of the truck, drone, ship, or stationary unit you are integrating, then verifying it against a 10,000-hour or longer field benchmark [S1][S6].
What "fuel cell stack" actually means in a sourcing spec
A fuel cell stack is a series of cells with bipolar plates, a proton exchange membrane (PEM) or anion exchange membrane (AEM), gas diffusion layers, and end plates torqued to a specified compressive load. Each cell typically produces 0.6-0.8 V under load, so a 300 kW automotive stack running near 600 A will sit in the 400-600 cell range with active plate areas measured in hundreds of cm². [S3]
For sourcing purposes, the stack is only half the system: balance of plant (BOP) components, including hydrogen recirculation, air compressor, humidifier, pressure transmitter manifold, and DC-DC converter, drive the integration cost and failure rate far more than the cell count. Buyers who only price the stack and ignore BOP routinely overrun their bill of materials by 40-60%.
Choosing the right chemistry: PEM, AEM, or solid oxide
PEM (proton exchange membrane) stacks dominate the Chinese supply base because they start fast, tolerate dynamic load, and match vehicle duty cycles; a 400 kW PEM stack from Dongfeng cleared a 10,000-hour durability test in 2026, supporting its 9,200-vehicle deployment footprint across 45 Chinese cities [S6].
AEM (anion exchange membrane) stacks run on cheaper non-precious metal catalysts and tolerate CO₂ in the hydrogen feed, which simplifies upstream reformer integration; Enapter began AEM stack production in Shenzhen with partner Runqing Hydrogen on August 19, 2026, with first units ready for delivery and capacity designed to drive unit cost reductions [S4].
Solid oxide fuel cells (SOFC) target stationary power above 100 kW with electrical efficiency near 60%, but the Chinese SOFC supply base is thinner than PEM, and startup times of 30 minutes to several hours rule out mobile duty. Marine and data-center backup are the realistic SOFC use cases in 2026, not trucks.
Selection criteria: power density, durability, and pressure interface
For a long-range industrial drone, the 2026 Chinese benchmark is an air-cooled PEM stack at high power density per kilogram, with the Dalian-appraised system targeting multi-hour flight times at a fraction of equivalent battery mass [S2]. For a 49-tonne tractor unit, the benchmark is 300-400 kW rated power, 10,000+ hour stack life, and a hydrogen supply pressure interface matched to 350 or 700 bar on-vehicle tanks [S1][S6].
For a megawatt-class marine stack, the global benchmark in 2026 is the ZSW demonstrator with 1,300 cm² active plate area per cell, scalable to 500 kW per stack, with disc-spring tensioning at roughly 150 kN / 10 bar stack pressure to maintain sealing and electrical contact across the large bipolar plates [S3]. The larger the active area, the more sensitive the cooling channel design, so CFD-optimised graphite bipolar plates are now standard at this scale.
Always pin down three numbers on the datasheet before you buy: rated gross power (kW), peak power duration (seconds/minutes), and BOP parasitic load (% of rated). A 400 kW nameplate that loses 80 kW to the air compressor is not a 400 kW stack in your integration.
China's deployment clusters and what they prove
China's five national FCEV demonstration clusters (Beijing-Tianjin-Hebei, Shanghai, Guangdong, Henan, and Shanxi/Hebei) collectively account for nearly 40,000 cumulative FCEV sales and 574 hydrogen refuelling stations by the end of 2025; the Shanxi cluster alone operates more than 1,000 hydrogen trucks on real freight corridors, which is the most credible field benchmark outside Europe [S1].
Vehicle economics have improved materially: a 49-tonne FCEV tractor fell from RMB 2.0-3.0 million in 2020-2021 to RMB 1.0-1.6 million in 2025-2026, and with subsidies in demonstration regions, effective purchase prices reach RMB 350,000-500,000, putting them within striking distance of diesel parity at approximately RMB 300,000-400,000 [S1].
Production-side hydrogen costs now sit near RMB 27.5/kg, but delivered pump prices remain around RMB 45.7/kg; parity with diesel on total cost of ownership requires roughly RMB 25-30/kg at the nozzle, and that gap between production and pump price is the main barrier to scaling outside the subsidised clusters [S1].
Standards, certifications, and what to put in the contract
For automotive stacks, the dominant Chinese benchmark is the GB/T 35579-2017 series covering fuel cell engine durability test methods, with stack-level performance verified against rated power, peak power, cold-start temperature, and a 10,000-hour endurance profile; for marine and stationary, IEC 62282 series fuel cell module standards and emerging class-society type approval (DNV, Lloyd's Register) carry the contractual weight [S1][S3].
Put four contractual gates in the purchase order before you wire money: a documented 10,000-hour (or application-relevant) endurance test report, GB/T or IEC type certificate matching the rated output, a hydrogen supply purity spec appropriate to the fuel cell technology (e.g., PEM or AEM), and a BOP integration manual that includes the linear guide and fastening torque for stack compression, hydrogen injector curves, and the load cell calibration for the stack clamping force monitor.
Who this is for, and who should buy elsewhere
Chinese stacks are the right choice for high-volume automotive, drone, bus, and stationary integrations where cost per kilowatt dominates, and where your team can integrate BOP and absorb GB/T documentation. They are a poor choice for low-volume, safety-critical, or non-Chinese-certified applications that demand immediate Lloyd's Register or DNV type approval on the stack itself, or for buyers who cannot verify crossed-roller guide precision and load cell module calibration on the assembly tooling used by the supplier. [S3]
Analyst forecasts from Rystad Energy see global fuel-cell market revenue rising from $2.8 billion in 2025 to $30 billion by 2030, driven primarily by AI data-center backup power demand; that growth is mostly addressable with Chinese PEM and AEM capacity, and buyers who lock in qualified stack suppliers now will get both priority allocation and stable pricing through the ramp [S9].
Two signals to track over the next quarter: the first Dongfeng-class 400 kW AEM retrofit announcement (which would change the catalyst-cost curve), and the first ZSW-class 500 kW marine stack commercial order, which will set the price benchmark for megawatt-scale PEM in shipping.
See also our earlier report, Cycloidal Reducer Selection Rules for Cement Plants.