The 2026 Flow Batteries North America (FBNA) conference happened October 19-21 in Phoenix. I left there reflecting on the chasm between “something that is a promising battery technology” and “an energy storage technology that can be reliably produced at the GWh scale that someone can confidently procure” (with “someone” being an EPC, Utility, etc.) In only its 2nd year, FBNA’s theme was Maturing and Multiplying Markets, and I feel like it captured the fact that flow batteries are actually scaling beyond incremental or academic progress. Today there is >1 GWh of real operating projects globally, with 200 GWh targeted by 2030. Data centers, rural utilities, materials suppliers, and financiers are discussing different parts of the same challenge: delivering dependable storage on a scale, at a price, with FEOC compliance, on a schedule that customers can accept.
The energy storage market extends well beyond AI data centers, despite their dominance in the press. Consider that renewable generation overtook coal generation for global electricity in 2025, and U.S. utility-scale battery power capacity exceeded 26 GW by the end of 2024, outpacing the ~22 GW power capacity of the pumped-storage hydro fleet. We are legitimately at a societal inflection point where batteries are getting built because they are essential to enable required electrical infrastructure integration: EV charging, industrial expansion, solar and wind generation, changing utility peaks—and yes, data centers—create additional demands on an electricity system already being rebuilt.
Data centers perhaps make the urgency extra-visible to everyone, but I would argue that rural utilities in particular face analogous energy problems through long feeders, expensive upgrades, limited operating staff, and the new phenomenon of winter peaking. Their opportunity for improvement is often better service and resilience on existing distribution networks (the analog is “uptime” in data center parlance). How quickly that service can be upgraded and made resilient can determine when a business or hospital can operate (the analog is “speed to power” for data centers, which determines when they can earn revenue). Battery energy storage may help a constrained connection serve more constituents (analog: compute) by managing peaks and buffering rapid load changes. In short, serving data centers will create learnings that help serve rural consumers.
Any energy transition or decarbonization effort must respect the necessity of today’s existing energy system. It is well known that generation by fossil fuel combustion carries environmental and health costs, but gas gensets remain critical for grid reliability and are also supported by a truly extensive production, supply, and pipeline network domestically. Batteries cannot replace that entire system, and especially not overnight. Instead, storage can reduce fuel use in the near term, and then improve operation alongside generation as cleaner supply expands. The appropriate mix depends on actual deliverable manufacturing capacity paired with local demand, recharge opportunities, delivery constraints, and the consequences of a prolonged outage.
Flow Batteries Deserve Consideration for Widespread Adoption to Meet These Needs
I want to nip this one in the bud: Long duration does not mean slow response. Flow batteries are often characterized as long duration energy storage (LDES), but they can also provide fast frequency services, grid inertia, or follow automatic generation control (AGC) commands while operating because the electrolyte available in the stack electrically reacts at sub-millisecond timescales Pump speed is not a clock for electron transfer. And yes, flow batteries can store intermittent generation from renewables and shift it to a different use time to smooth the demand curve, or they can island a load for 8+ hours, and both of those are LDES services that do rely on continuous liquid flow to access the entire system energy capacity.
Most flow battery system designs allow separate sizing of power and energy capacity: The so-called stacks (electrochemical conversion cells, for the nerds) and power electronics establish the power capacity P, in MW, whereas electrolyte concentration and tank volume largely establish energy capacity E, in MWh (and the duration that the nameplate power can be provided because E = P × duration). Additional hours of duration can therefore be added with extra liquid/tank—but without proportionally increasing every component—so the marginal cost of added energy is low. There are different companies making different chemistries (e.g. Storion Energy makes high purity vanadium electrolyte, whereas BioZen Batteries, my company, makes water-soluble carbon-based Redoxolyte™ electrolytes, for example) and different system architectures (e.g. Invinity Energy Systems makes productized Endurium™ modules, whereas TerraFlow Energy makes LDUPS™ systems that are built-for-purpose with a data center), so the ultimate commercial benefit depends on chemistry, cycling profiles, footprint, efficiency, system architecture, and installed cost.
My Flow Battery Predictions and What Should Be Done About Them
Prediction 1: vertical flow battery integration will be too slow. I think the biggest winners will be component OEMs or integrators who lean into their core competencies in electrolyte production, carbon supply, electrode manufacturing, membrane production, tank fabrication, stack manufacture, power electronics, project development, creative project finance, etc. A company will expand rapidly by standardizing in their lane without owning every step, enabling others to easily incorporate their technology. Owning integration of independently produced components could also prove more valuable than the components themselves.
Prediction 2: “long duration energy storage” will be an unnecessarily restrictive flow battery identity. Yes, flow batteries fit the LDES definition. But, the value of duration changes as the grid and policy changes. Data centers need rapid response, power quality, and guaranteed uptime. Rural utilities need reliable service and resilience against outages. A stronger business case in my opinion is “how much useful work a flow battery can do” and not “how many discharge hours appear on the spec sheet.” Price arbitrage or “power as a service” / IPP type operation may also prove quite lucrative.
Prediction 3: flow battery workforce development will be a major bottleneck. Flow batteries also expand (and require) manufacturing and workforce development opportunities beyond those that one might envision from conventional Li-ion, lead acid, or alkaline battery production. Chemical processing, carbon materials, membranes, tanks, pumps, seals, and electrical assemblies all contribute to the finished system, and we need people to manufacture all of those and integrate them too. At FBNA, I asked one panel “if you had to scale up manufacturing 10X tomorrow, what breaks first?” Answer: workforce. Start training people yesterday.
Prediction 4: data centers will accelerate flow battery deployment even if they never become the largest market. And we should embrace this—with an eye towards standardization. Data centers are well financed and put an unusual adaptive pressure on their surroundings: rapid load swings that shear conventional generator shafts and need better power electronics, ride-through requirements that force redundancy and control algorithm improvements so a huge load doesn’t suddenly disappear from the grid, and speed-to-power requirements that force suppliers to expand consistent production and commissioning. Data centers also have a nebulous commercial trajectory, so in my opinion the best solutions will standardize integration that makes future utility and industrial installations easier to deliver. The value of that may exceed direct revenue from data center sales, and it also alleviates the risk of wasting engineering on ephemeral or bespoke data center demonstration.
Prediction 5: proprietary flow battery integration will kill sales. Flow batteries are competing against lithium ion, sodium ion, thermal, and other storage technologies writ large. Flow battery companies must collaborate, or risk going out of business. This is key, too, because buyers are considering infrastructure with a long service life, and suppliers whose own longevity is uncertain can undermine project finance. What if the battery outlives its manufacturer? Vanadium apparently doesn’t degrade, after all. In my opinion, easily-serviceable system designs, accessible operating documentation, replaceable controls, and qualified alternative suppliers will be required bankability features. Buyers may accept a less-impressive spec in exchange for confidence that someone else can maintain the plant. Excessive dependence on proprietary OEMs will make the project harder to finance. Independent service businesses may be lucrative to operate.
Prediction 6: aggregated GWh of demand will be a bad indicator. A manufacturer’s ability to deliver “more batteries” depends heavily on what customers order. Consider that different applications might require high-power, high-frequency, AGC, long-duration (dis)charge, etc. Those applications therefore create different demand for electrolyte, electrodes, stacks, membranes, tanks, power electronics, etc.—and each of those have different scaling problems. This also means that a company could simultaneously have surplus capacity in one part of the system and shortages in another. Aggregate GWh demand is too coarse a metric and poorly matched to the actual order book. Justifying, say, a factory expansion from that number could be a company-ending decision. (Announced factory capacity is equally unhelpful if customers cannot finance the resulting product.) A better metric might at the very least separate power and duration, and ideally answer what mix of customer projects the supply chain can deliver.
Prediction 7: there is not enough vanadium, and we will need alternative chemistries to scale. Vanadium is the most commercially-mature active storage material used in flow batteries, and it represents 40-70% of the system cost. World resources of vanadium exceed 63 million tons, so it appears sufficiently abundant at first shake. However, it must be mined, extracted, and purified, and vanadium pricing and availability is largely dependent on the market characteristics of the steel industry (dominated by China, with a central role in global steel production and the top vanadium producer in 2025). A generous estimate of global annual mined vanadium is 200,000 mt. Approximately 10,000 mt are needed per 1 GWh of vanadium flow batteries. This means that the prior 2030 forecast of 200 GWh represents 2,000,000 mt of vanadium: 10 years of the entire global production capacity! By several accounts, individual companies have that much demand for their data center applications alone, so this supply chain simply does not pencil—especially considering that over 90% of vanadium production is used as an alloying agent in steel. Flow batteries must go to market with vanadium electrolyte because it works, but in my opinion it is critical that alternative formulations of electrolyte with massively-scalable domestic supply be established. One example is aqueous-soluble organic semiconductor active materials (such as BioZen’s Redoxolyte™ electrolytes) because they are manufactured through well-established petrochemical processes analogous to those used in polymers, pharmaceuticals, and other commodity chemicals.
My Conclusion: Flow Batteries Are Becoming an Industrial Coordination Business
If I took one thing away from FBNA, it is that flow battery manufacturing and deployment in the short term increasingly depends on organizing and synchronizing capabilities that already exist, while making the complete system reliable and financeable. Chemistry and system architecture still determines what is possible, but the competitive advantage over, e.g., Li-ion emerges when we get independent suppliers, installed equipment, deployment contracts, and service organizations to operate like one coherent, dependable product. The next stage of flow battery growth is going to redistribute power within the industry, rewarding the companies that make flow batteries straightforward to buy and operate. In the longer term, companies that enable massive domestic supply chain scaling across the entire tech stack will win.
References
[1] Flow Batteries North America 2026: The Flow Battery Markets are Maturing and Multiplying
[2a] Battery Council International, Global Flow Battery Project Map; [2b] Flow Batteries Europe, Flow Battery Targets: The road to 20 GW and 200 GWh by 2030
[3] IRS, Notice 2026-15, interim prohibited-foreign-entity guidance
[4] Ember, Global Electricity Review 2026
[5] EIA, U.S. battery capacity increased 66% in 2024 (12 March 2025)
[7] NYISO, Fuel Security: A Growing Winter Reliability Concern
[8] EPA, Human Health and Environmental Impacts of the Electric Power Sector
[9] EIA, Natural Gas Explained: Delivery and Storage
[10a] Storion, White Paper: Fast Response Validation of Storion Flow Battery Stacks for LDUPS; [10b] Invinity, Flow Battery Response Time and the Business Case for Ancillary Services
[11] PNNL, Analytical Modeling for Redox Flow Battery Design, Journal of Power Sources (2020)
[12] USGS, Mineral Commodity Summaries 2026: Vanadium
[13] Storion Energy, Vanadium Redox Flow Battery Fast Facts
[15] BioZen Batteries, Redoxolyte™ electrolytes for flow batteries
