China Three Gorges Corporation moved its 1,000MW "CSP + PV" integrated project in Hami, Xinjiang Uyghur Autonomous Region, into commercial trial operation on June 27, 2026. The equipment breakdown consists of 900MW of solar PV and 100MW of linear Fresnel concentrated solar power (CSP) that stores heat in molten salt. Part of the solar energy concentrated during the day is shifted to nighttime, functioning as a dispatchable power source.
Here it's important not to misread the scale. What can operate after sunset is not the full 1GW of equipment, but the 100MW CSP portion. The figure of 8 hours also refers to the design-basis thermal storage duration that allows the 100MW unit to run at rated output for 8 hours. What the Hami project is testing is not "a solar plant that generates 1GW even at night," but rather how to combine cheap daytime PV with a 100MW-class steam turbine that handles the evening and beyond.
Not 1GW, But 100MW × 8 Hours
Converting the CSP side's thermal storage capacity into electrical output yields a maximum equivalent of roughly 800MWh at rated conversion. On a day with sufficiently stored heat, there's room to either run at 100MW for 8 hours or lower the output to extend runtime. However, on days with weak solar irradiance or an unfilled thermal storage tank, the same operation cannot be sustained. Eight hours is not a nightly guarantee.
Looking at the annual figures makes the equipment's role even clearer. Three Gorges Corporation projects annual generation from the CSP side, once full commercial operation begins, at over 145 million kWh. Dividing this by 100MW gives an equivalent full-load duration of 1,450 hours, or a capacity factor of about 16.6%. This should be understood not as a baseload power source delivering constant output across the year's 8,760 hours, but as a peaking/regulation asset that concentrates generation into high-value time windows.
Even the "world's largest" label comes with caveats. Hami reaches 1,000MW in combined PV-plus-CSP nameplate capacity, surpassing Dubai's Noor Energy 1 at 950MW. Noor Energy 1, however, combines 700MW of CSP with 250MW of PV, and has a thermal storage capacity of 5,907MWh with up to 15 hours of duration. Comparing nighttime-dispatchable CSP output alone, Noor Energy 1's figure far exceeds Hami's 100MW. Hami's record is the nameplate capacity of the hybrid as a whole—not a record for the size of its thermal storage or nighttime output.
260,000 Mirrors Heat Salt to 550°C
The linear Fresnel design tracks flat or gently curved mirrors arrayed on the ground along a single axis, concentrating sunlight onto fixed receiver tubes mounted above. At Hami, 260,000 reflector mirrors form an 800,000-square-meter collector area, with secondary reflectors further focusing the light onto the receiver tubes. Molten salt flowing through the receiver tubes is heated to roughly 550°C and sent to hot salt tanks.
During power generation, the hot salt transfers its heat to a steam generator. The resulting high-temperature, high-pressure steam then drives a turbine and generator, just as in a conventional thermal power plant. Because solar energy is temporarily stored as heat, the timing of power generation can be decoupled from the timing of sunlight hitting the mirrors. Whereas PV sends out electrons directly, CSP interposes a hot salt tank in between.
Because the receiver tubes can remain fixed and the mirrors can be mounted low to the ground, this design tends to simplify the structure. In exchange, optical losses tend to be larger than with trough-type curved mirrors. Hami addresses this with secondary reflectors and divides the collector field into 46 loops. Since the remaining loops can keep running even when one loop is shut down, the impact of maintenance outages can be localized even in a facility with 260,000 mirrors.
The operational challenge lies less in heating the salt than in avoiding overcooling it. The molten salt used at Hami tends to solidify around 220°C and rises to a maximum of roughly 560°C during operation. This temperature range must be continuously managed across the piping and tanks as a whole. Hami's winters bring a combination of low temperatures and weak solar irradiance, while dust reduces mirror reflectivity. Insulation of piping and tanks, temperature management of the receiver tubes, and robotic dry cleaning plus water washing by cleaning vehicles all directly affect annual generation output.
A 25MW Electric Heater Links PV and CSP
Viewing this facility as a pure solar thermal plant misses half the design. Hami is equipped with a 25MW electric heater that can use surplus power from the 900MW PV side to heat the molten salt. In this configuration, heat gathered from the mirror field and electricity that cannot be fully fed into the grid both enter the same hot salt tank, to be converted back into electricity via the steam turbine at the required time.
Of course, converting electricity to heat and back to electricity again incurs conversion losses. Purely in terms of short-duration charge-discharge efficiency, lithium-ion batteries tend to have the advantage. Even so, if the market value of electricity is nearly lost during hours when PV curtailment is the only option, accepting the loss to shift that power to nighttime starts to make sense. Through the 25MW heater, the 100MW CSP generator and the 900MW PV array are actually linked together in real operation.
This is also where the value of retaining a steam turbine lies. Unlike inverter-connected PV, a rotating synchronous generator has inertia and can participate in frequency and voltage regulation. China's National Energy Administration has indicated a policy direction of valuing this kind of output regulation within electricity markets. Black-start capability—the ability to restart independently after an outage—and inertia support are also covered. Measuring profitability solely by generated kWh would fail to count part of what CSP actually sells to the grid.
The "Work Beyond Generation" That Has Value in Hami
Hami's power grid has the conditions in place to serve as a testing ground for this kind of regulation capability. By the end of 2025, total installed capacity exceeded 50 million kW, with new energy sources centered on wind and solar exceeding 37 million kW—a share of 73.5%. As daytime PV output grows, more power sources are needed to fill the evening decline and stabilize the power flowing into transmission lines.
The region's power is also linked to long-distance transmission. The Hami–Chongqing ±800kV UHV DC transmission line, which began operation in June 2025, spans 2,260km with a rated capacity of 8 million kW and is planned to transmit over 36 billion kWh annually. Of the 14.2 million kW of paired generation capacity, new energy sources such as wind, PV, and CSP account for over 70%. While it has not been confirmed that Hami's integrated project feeds directly into this transmission line, across the region as a whole, the challenge of matching a vast fleet of weather-dependent power sources to distant demand has already become a reality.
At 100MW, the CSP facility is not on a scale that can single-handedly transform Hami's power mix of over 50 million kW. Even so, placing a 100MW-class linear Fresnel system under the same control regime as 900MW of PV—storing energy from both heat and electricity, and dispatching output at night via a synchronous generator—yields a valuable body of operational data. The value of the commercial trial has shifted from merely demonstrating that the mirrors and salt work, to verifying just how reliably the grid can be served at the moments it actually needs power.
After the Commercial Trial, the Numbers That Will Measure Economics
Whether the technology works and whether it can win in competition are two different questions. According to the National Energy Administration, construction costs for CSP plants in China have roughly halved over the past decade, from about 30,000 yuan/kW to 15,000 yuan/kW, with generation costs falling to around 0.6 yuan/kWh. Even so, government documents cite the large scale of initial investment, weak market competitiveness, and the lack of institutional mechanisms to convert regulation value into revenue as remaining challenges. The target of deploying 15 million kW of CSP capacity by 2030 while bringing generation costs down to parity with coal-fired power is itself a reflection of the fact that current profitability still depends on policy support and market design.
The price decline of competing technologies is rapid. According to 2025 data from the International Renewable Energy Agency (IRENA), the global weighted-average cost of electricity was $115/MWh for CSP and $44/MWh for PV. Moreover, at sites with even better solar conditions, PV-plus-battery storage assuming 95% reliability came in below $85/MWh. Installation costs for 4-hour grid-scale battery storage have fallen to about $140/kWh globally on average, and below $70/kWh in China.
This comparison is not the result of putting Hami's 8-hour CSP head-to-head in a bid against 4-hour battery storage. CSP costs include the mirror field, thermal storage, and steam turbine, while battery storage installation costs do not include the generation equipment needed to charge them. Even so, now that PV-plus-battery is starting to move into the "long-duration, high-reliability" territory, molten salt can no longer take its cost advantage for granted.
Judging commercial viability requires operational track record, not planned figures. Can the facility exceed 145 million kWh per year? On how many days per year can it actually deliver 100MW for 8 hours? How many kWh of PV power that would otherwise have been curtailed has the 25MW electric heater actually absorbed? How much power is consumed internally to prevent the salt from freezing, and how much does maintaining the mirrors and receiver tubes cost? And how much revenue is earned from frequency regulation and inertia support? Once these figures are made public, the assessment of Hami will shift from a photograph of a giant power plant to commercial data that allows CSP and battery storage to be compared on equal terms.