Sodium batteries have just arrived at the threshold of the computing power and energy storage sector, only to hit a solid wall.

In the mid-to-late June 2026, bidding for energy storage supporting facilities at "East Data, West Processing" computing power parks across multiple regions in Southwest and Northwest China officially kicked off.

Reporters from the Economic Observer learned from relevant project leaders of the Chengdu-Chongqing Computing Power Corridor, the Zhongning Intelligent Computing Cloud Base in Ningxia, and the "East Data, West Processing" Industrial Park in Ordos, Inner Mongolia, that this round of bidding rules has seen notable adjustments: both sodium-ion batteries (hereinafter referred to as SIBs) and lithium iron phosphate (LFP) energy storage systems are eligible to bid independently, with the single-station energy storage configuration scale for most projects set at 50MW/100MWh (i.e., a power output of 50 megawatts and a capacity of 100 megawatt-hours).

Over the past two years, large-scale computing power park energy storage tenders have mostly only allowed LFP batteries to participate, while SIBs have only appeared sporadically in demonstration-specific bid sections. This time, SIBs are stepping onto the main stage to undergo evaluation alongside LFP systems under the same technical criteria. The competition now extends far beyond initial procurement costs, encompassing comprehensive comparisons across dimensions such as low-temperature performance, raw material supply chain security, and levelized cost of electricity over the full lifecycle.

An unexpected phenomenon emerged at the bidding site that caught the evaluation panel off guard: most enterprises participating in SIB bids declared capacities ranging from 5MWh to 20MWh (5 megawatt-hours to 20 megawatt-hours), and few were willing to independently undertake the full 100MWh bid sections covering the entire park.

Multiple SIB equipment manufacturers have expressed similar concerns: there is insufficient effective production capacity supply of battery-grade hard carbon anodes, and without long-term material locking agreements to provide support, the delivery cycle after winning large-value orders remains uncertain. Combined with factors such as energy storage-dedicated SIB cell production lines still in the ramp-up phase and tight supply of supporting electrolytes, most enterprises at this stage have opted to test the waters only with small-capacity bid sections.

01

The "Newcomer" on the Bidding List

In mid-to-late June, the reporter reviewed the bidding status of supporting energy storage for newly built intelligent computing projects in several regions and found that this batch of tenders included a clause not seen in previous years in multiple locations: both SIB and LFP technology routes are permitted to participate in bidding.

Previously, centralized energy storage supporting facilities for large commercial computing power parks have long been dominated by LFP batteries. SIBs were primarily deployed in two-wheeled electric vehicles, low-speed vehicles, and residential energy storage systems, and had never been included in the procurement lists of high-grade computing power infrastructure projects.

The turning point came in 2026.

In April, the National Development and Reform Commission, the National Energy Administration, the Ministry of Industry and Information Technology, and the National Data Bureau jointly issued Document No. 34 [2026] of the National Energy Administration (hereinafter referred to as "Document 34"), which encourages computing power facilities to deploy grid-forming energy storage systems. After the document was publicly interpreted for the industry in May, several computing power hubs in western China immediately adjusted the energy storage supporting standards for their new intelligent computing projects.

A person involved in policy discussions told the reporter that previous energy storage policies primarily targeted wind and solar power stations, while energy storage systems on the computing power side were mostly self-initiated supporting facilities for individual projects. Document 34 is a top-level national policy document that for the first time systematically ties computing power infrastructure to grid-forming energy storage policies, which to a certain extent has prompted local state-owned platforms to revise their procurement rules.

A person in charge of energy supporting facilities at a western urban investment park told the reporter that the dual-route bidding model was not an independent decision of a single park, but a procurement regulation uniformly introduced by the municipal state-owned operation platform in conjunction with higher-level pilot policies. He stated that in previous years, energy storage tenders only set a single LFP route, but this year policies encourage the verification of new energy storage technologies. Coupled with growing concerns among computing power projects over cost fluctuations of raw materials such as copper foil, the platform has uniformly required that all centralized energy storage bid sections must reserve a bidding channel for SIB pilots.

The aforementioned person in charge of energy supporting facilities at the western urban investment park explained that the total energy storage scale for a single park in this round is mostly 50MW/100MWh, and the platform has uniformly designated approximately 10% to 20% of the capacity for SIB technology pilots, avoiding large-scale deployment all at once.

"This ratio was carefully calculated," he explained. State-owned projects have two core assessment criteria: long-term stable operation, and independent controllability of key raw material supply chains. If we rely entirely on lithium resources and copper foil, both materials have high external dependency, and fluctuations in commodity prices and production capacity changes will affect the park's operating costs for decades. The dual-route bidding model is essentially a form of risk diversification: mature lithium batteries are used as the main energy storage to guarantee the bottom line of power supply to computer rooms, while a small proportion of SIBs are deployed for technical reserves.

Low temperature is another practical consideration. In the regions where western computing power hubs are located, the minimum winter temperature generally ranges from -25°C to -35°C. The aforementioned person in charge noted that even with basic insulation structures, the discharge capacity loss of outdoor prefabricated LFP energy storage cabins in the park under low-temperature operating conditions still reaches 15% to 22%, requiring additional heating and insulation equipment that increases civil engineering and electricity operation and maintenance costs.

In the first half of 2026, they conducted outdoor field tests using small-scale SIB prefabricated cabins. Under the same scenario, the measured capacity loss of SIB prefabricated cabins in a -30°C environment could be controlled within 5% without the need for high-power insulation supporting facilities (this field test data is from park-customized SIB prefabricated cabin samples, with minor variations between cells from different manufacturers and across different testing entities).

However, SIBs have relatively low energy density, and there is insufficient compact energy storage space inside computer rooms, so indoor backup energy storage still prioritizes lithium batteries.

The urban investment platform's plan is very clear: use SIBs for outdoor centralized grid-forming energy storage pilots, and retain lithium batteries for indoor high-density backup energy storage, advancing on both fronts.

The reporter learned during interviews that after Document 34 was issued, local development and reform departments and energy authorities have simultaneously issued supporting implementation guidelines to intelligent computing parks, clarifying that newly built computing power centers must be equipped with grid-forming energy storage and encouraging pilots of diversified energy storage technologies. The entire process, from policy issuance to feasibility study adjustment of projects, grid connection plan review, and bidding plan approval, takes exactly one month to complete.

The aforementioned person in charge of energy supporting facilities at the western urban investment park revealed that during pre-bidding communication with multiple SIB enterprises, he found that although leading battery companies can provide complete 100MWh-class large-capacity solutions, they will actively advise the park to control the pilot capacity, as the delivery cycle of SIB cells cannot match the park's production launch timeline. Some small and medium-sized enterprises were even more direct, clearly stating that they can only undertake small bid sections of 10MWh or less.

"It's not that they don't want to take on the work, it's that the upstream materials can't keep up," the person in charge said.

02

The Threshold They Dare Not Cross

At the urban investment platform where the aforementioned person in charge of energy supporting facilities at the western urban investment park works, the evaluation panel also noticed when reviewing the bidding documents that most enterprises participating in SIB bids declared pilot capacities concentrated between 5MWh and 20MWh (5 megawatt-hours to 20 megawatt-hours).

Zhao Haining, a bidding leader at an energy storage integrator, explained to the reporter that 5MWh to 20MWh is the "upper limit that most SIB suppliers without long-term hard carbon locking agreements dare to sign for at this stage." For large-capacity bid sections above 30MWh, his company currently prioritizes recommending technically mature LFP energy storage solutions to project owners. The core reason is that the stable delivery capacity of hard carbon anodes cannot support a larger scale of operations.

Zhao Haining's company is simultaneously following up on intelligent computing energy storage tenders in three locations, all submitting two sets of technical solutions for lithium and sodium batteries, but actively limiting the SIB portion to within 20MWh. "If we exceed this figure, even after signing the contract, we can't guarantee delivery on time," he said.

This assessment largely aligns with the project owner's calculations. The aforementioned person in charge of energy supporting facilities at the western urban investment park provided another perspective: this round sets the SIB pilot quota at a ratio of 10% to 20%, which translates to 10 to 20MWh. The upper limit of the pilot capacity planned by the project owner happens to be the upper limit that the bidders dare to quote.

Why is 20MWh a critical threshold?

Zhao Haining explained that computing power energy storage differs from conventional wind and solar energy storage. Wind and solar energy storage focuses on long-term static energy storage, with relatively lenient requirements for cell consistency and system response speed. Computing power energy storage is different: AI (artificial intelligence) computing centers have large load fluctuations with high frequency, so energy storage systems need to perform frequent and fast frequency regulation, significantly raising requirements for cell consistency and BMS (Battery Management System) dynamic regulation algorithms. The company's R&D team is urgently iterating on grid-forming SIB energy storage PACK (battery modules) dedicated to computing power applications, but the development and debugging cycle of the entire control system is longer than expected, making it difficult to deliver mature large-capacity solutions in the short term.

Zhao Haining noted that a greater constraint lies upstream. In the second quarter of 2026, downstream energy storage and low-speed two-wheeled vehicles simultaneously drive SIB demand, and the overall hard carbon production capacity is already in a tight balance. If a gigawatt-hour (GWh, i.e., millions of kilowatt-hours, referring to the total capacity of large-scale energy storage projects) level large order is undertaken at once, it will be very difficult to obtain sufficient spot goods.

Another bidding detail confirms this supply crunch. The aforementioned person in charge of energy supporting facilities at the western urban investment park noticed during the technical clarification session that multiple bidders actively indicated that hard carbon spot supplies are tight, and the cost of locking in prices via long-term agreements has risen. Affected by the simultaneous price increases of multiple raw materials such as hard carbon and sodium-based electrolytes, the cost quotation advantage of SIB energy storage systems of the same capacity over LFP systems has significantly narrowed compared to the beginning of the year. At the same time, SIB enterprises have universally requested to extend the delivery cycle by 30 to 60 days. For urban investment platforms, the production launch timeline of computing power rooms is a hard constraint, and power supply supporting facilities must be in place synchronously.

Zhao Haining compared that at the beginning of 2026, SIB systems of the same capacity still had a clear cost advantage over LFP systems, but after two rounds of hard carbon price hikes, the price gap has continued to narrow. Except for scenarios such as low-temperature outdoor energy storage in western China, where SIBs have natural performance advantages, SIB quotations for energy storage projects under normal operating conditions no longer have much competitiveness.

Zhao Haining further explained that the company's internal capacity allocation strategy is to "guarantee mainstay orders with lithium batteries, and dedicate SIBs exclusively to pilot projects" without crowding out mature lithium battery delivery capacity. The quotation also adopts a dual-track strategy: appropriately compress profits for SIB pilot bid sections in exchange for scenario verification opportunities; promote LFP for large-capacity mainstay bid sections to ensure cash flow and delivery stability.

This explains why enterprises at the bidding site dare not quote for large orders: it's not that they don't want the orders, but that neither the material side nor the delivery side has the conditions to undertake large orders.

The advantages of SIBs have been repeatedly emphasized in laboratories and promotional materials: no copper foil required, excellent low-temperature resistance, and low cost.

However, at the bidding site, a different set of logic determines whether a contract can be signed: whether delivery can be completed on time, whether cell consistency can be guaranteed, and whether the system can pass grid connection tests.

Zhao Haining analyzed that "hard carbon" is one of the bottlenecks: "We made it very clear to the project owner that SIBs are not recommended for capacities above 20MWh at this stage."

The urban investment evaluation panel also heard this statement. The aforementioned person in charge of energy supporting facilities at the western urban investment park told the reporter that after comprehensive assessment, the evaluation panel will not break through the 20% pilot upper limit in this round. "It's not that we are being conservative, it's that this is the only way to go at this stage. SIBs have been included in the bidding documents, but they are not yet ready on the delivery side," he said.

03

What's Causing the Bottleneck?

The aforementioned person in charge of energy supporting facilities at the western urban investment park stated that they have signed annual long-term agreements with leading hard carbon enterprises, but these long-term agreements only lock in basic production capacity, with "very little reserved incremental space." If they undertake large-capacity SIB bid sections of dozens of MWh or more, the long-term agreement quota will be far from sufficient to cover the demand, forcing them to split deliveries into multiple batches and postpone completion.

Zhou Ming, a sales professional at a hard carbon material enterprise, explained the relevant production capacity logic to the reporter.

Zhou Ming stated that the total publicly announced planned production capacity in the industry appears large, but a large number of newly built production lines have not yet completed the ramp-up phase, and the actual effective shipment capacity that can stably supply goods and meet the long-cycle standards of energy storage is very limited. The centralized bidding for western computing power energy storage in June has brought new demand for cells, and spot hard carbon in the market has already experienced a phased shortage. Downstream battery companies need to lock in goods 1 to 3 months in advance, and it is difficult to achieve rapid and full delivery for temporary large-volume purchases. This supply-demand imbalance began in April-May 2026. After the release of Document 34, the energy storage supporting standards of computing power parks across the country were successively adjusted, and SIB pilot channels were opened in a concentrated manner. The originally scattered small-batch demand converged into concentrated procurement expectations in the short term, but the upstream hard carbon production line construction cycle is long and cannot respond synchronously, leading to phased supply tension in the spot market even before the arrival of the June bidding season.

Zhou Ming also stated that for industry-standard 10,000-ton-level hard carbon production lines, the complete implementation cycle from project initiation and EIA (Environmental Impact Assessment), civil engineering construction, equipment commissioning to stable mass production ramp-up is generally 18 to 24 months. Within 2026, only a small number of production lines that started construction in the first half of last year can release incremental capacity on a small scale, and new expansion projects launched this year cannot form effective supply within the year. "The short-term supply increment is very limited."

The upstream raw material segment is also creating additional pressure.

Zhou Ming introduced that there are two mainstream raw material routes for hard carbon: the first is coconut shell-based hard carbon, which has stable performance but relies on overseas imports, with production areas concentrated in Southeast Asia. Policy fluctuations in producing regions and changes in maritime transportation will directly affect supply stability; the second is coal-based hard carbon, which is being promoted domestically as the main direction for domestic substitution, but the large-scale stable mass production capacity at this stage is still insufficient. The current stable mass production capacity of domestic coal-based hard carbon is limited, and it can only divert about 30% of the total market demand on the whole, making it difficult to fully fill the new procurement gap brought by computing power energy storage in the short term.

He confirmed a trend to the reporter: the import price of coconut shell-based precursors has continued to rise due to fluctuations in overseas production capacity and maritime transportation. Although domestic coal-based hard carbon is expanding production, there is still a gap in process maturity. At the beginning of the year, SIB systems of the same capacity had a clear cost advantage over LFP systems, but after two rounds of hard carbon price hikes, the price gap has continued to narrow. Except for scenarios such as low-temperature outdoor energy storage for western computing power, where SIBs have natural performance advantages, SIB quotations for energy storage projects under normal operating conditions can hardly undercut LFP prices.

Zhao Haining said that when submitting SIB solutions at the beginning of the year, project owners would pay extra attention due to the cost advantage. By the time of this round of bidding in June, hard carbon price hikes have eaten up most of the price gap. "When we submit SIB proposals now, we can only basically highlight low-temperature performance and supply chain security; cost is no longer a plus point," he said.

Zhou Ming added, however, that leading battery manufacturers with locked-in hard carbon long-term agreements already have the delivery capacity to undertake SIB pilot bid sections of 50MWh or more, while most small and medium-sized manufacturers in the industry are still constrained by material restrictions on capacity.

The aforementioned person involved in policy discussions believes that developing the top-level SIB industry can optimize the overall energy storage supply chain structure and hedge against the periodic supply risks of lithium and copper resources. However, industrial development has phased characteristics: the sudden surge