Holyvolt Funding Puts $45M Behind Lab-Tested AI Battery Discovery

Holyvolt raised $45 million to connect AI-guided battery research with physical testing and manufacturing, according to Fortune. The reported Holyvolt funding values the company at $245 million and includes backing from Zone II, Course Corrected, and other investors. The valuation comes from an unnamed source, rather than a company announcement.

The money supports a larger wager than another battery laboratory. Holyvolt argues that Western producers cannot catch China by copying the lithium iron phosphate, or LFP, supply chain China already operates at scale. It wants to find different cathode materials, validate them through automated experiments, and design production methods around those materials.

That approach puts Holyvolt against scaled incumbency, not one comparable startup. China supplied about 85% of global cathode active material in 2025 and more than 80% of battery cells. A promising chemistry therefore needs more than attractive laboratory results. It must survive qualification, production, integration, and cost competition against an established industrial system.

Holyvolt Funding Connects AI Research to a Physical Lab

The financing matters because Holyvolt is funding an integrated discovery loop, not an AI model working apart from experiments.

The Stockholm-based company is combining its manufacturing work with Wildcat Discovery Technologies, a San Diego battery research business. Holyvolt completed its acquisition of Wildcat in March 2026. The transaction joined Holyvolt’s screen-printing process with Wildcat’s high-throughput experimentation platform.

High-throughput experimentation means preparing and testing many material combinations through standardized, partially automated workflows. It replaces a narrow sequence of manually chosen experiments with a broader search across formulations and processing conditions.

The platform gives machine-learning systems structured results from physical cells. Those results include failed experiments, which are often absent from published scientific literature. Failure data can help a model avoid repeatedly proposing formulations that look plausible in theory but fail under testing.

Wildcat says its platform has supported more than 225 projects for over 85 customers. It also says it has tested more than 500,000 cells during 18 years of work. Those figures are company-reported, but they identify the asset behind the investment thesis: a large, consistently generated experimental record.

That record separates Holyvolt’s strategy from purely computational materials discovery. A model can rank candidate chemistries using known relationships and previous outcomes. The laboratory then manufactures samples, tests their behavior, and returns the results to the model.

Each cycle should narrow the search. However, the loop remains useful only when the experiments measure properties that matter outside the laboratory. Energy density, cycle life, charging behavior, temperature tolerance, safety, and manufacturability can pull a material in different directions.

Holyvolt’s Wildcat acquisition added this testing capacity to its existing manufacturing ambitions. The company says its combined platform can move from materials discovery toward production using water-based processing and screen printing.

Screen printing deposits controlled layers of material through patterned screens. Holyvolt says the method can reduce process complexity and support flexible electrode designs. It also claims the approach requires less energy and capital than conventional production, although those advantages need validation at commercial volumes.

The fresh capital gives the company more room to connect these stages. It does not eliminate the difficult transitions between them. A material that performs well in a small test cell can behave differently when electrode thickness, batch size, equipment, and quality requirements change.

The round therefore funds a sequence of linked risks. Holyvolt must identify useful materials, reproduce their performance, manufacture them consistently, and convince customers to qualify them. The first two steps fit a research platform. The final two determine whether that research becomes an industrial business.

Fortune’s funding details describe the round as entirely equity financed. Holyvolt did not publicly identify a lead investor or assign the round a conventional stage.

That ambiguity is less important than the intended use. The company is directing capital toward a pipeline that combines computation, experiments, materials development, and manufacturing design. Investors are betting that integration can shorten the path from a predicted chemistry to a qualified product.

The central question is whether Holyvolt can preserve that speed after leaving its controlled research environment. Battery customers demand stable output across thousands or millions of cells. They also require detailed safety data, reliable suppliers, and predictable production economics.

The Holyvolt funding makes those tests possible. It does not settle them.

Why AI Battery Discovery Still Needs Experiments

Holyvolt’s strongest argument is that battery chemistry cannot be optimized through computation alone.

Chief Operating Officer Steven Prast divides AI materials companies into two broad camps. One expects increasingly capable models to solve more chemistry problems through simulation. The other treats physical experimentation as an essential source of truth.

Holyvolt belongs to the second camp. Prast told Fortune that the company must return to the laboratory and test its candidates. That position reflects a practical limitation of AI battery discovery: models inherit the gaps and biases in their training data.

Scientific papers usually emphasize successful experiments. Researchers have fewer incentives to publish every unproductive formulation, failed coating, unstable cell, or disappointing cycle test. An AI system trained mainly on published successes sees only part of the search space.

A proprietary experimental archive can fill some of that gap. Wildcat’s standardized workflow reportedly records both productive and unproductive trials. Consistent records make it easier to compare how changes in composition, processing, and cell construction affect results.

The approach resembles active learning. A model selects experiments expected to provide valuable information, the laboratory runs them, and the new evidence guides the next selection. This does not turn chemistry into a fully automated process. Scientists must still define targets, judge data quality, and investigate unexpected behavior.

Physical validation becomes especially important when a candidate moves beyond its initial metric. A formulation selected for high capacity might degrade quickly. Another might use abundant materials but require processing conditions that make commercial production impractical.

Holyvolt is applying this loop to disordered rocksalt, or DRX, cathodes. DRX is a family of lithium-rich materials whose metal atoms do not follow the ordered layers found in many established cathodes. That disordered structure supports a wider range of possible constituent metals.

The attraction is flexibility. Researchers can investigate DRX formulations that reduce or remove nickel and cobalt while targeting high energy storage. Holyvolt says its lead DRX material contains neither metal.

The scientific opportunity predates Holyvolt. The US Department of Energy has described DRX materials as a diverse class with potential for high energy density and lower-cost, cobalt-free cathodes. Its DRX research also identifies unresolved issues involving cycling stability, rate capability, and voltage behavior.

Those limitations explain why extensive screening matters. DRX is not one fixed chemistry with a single performance profile. It is a broad design space involving material composition, particle properties, coatings, electrolytes, and production conditions.

Holyvolt says its initial DRX product can offer performance comparable to nickel-based cathodes at a landed cost near LFP in North America. It also says the supply chain can rely on friendly countries. These remain company claims until independent testing, customer qualification, and production data establish them.

The comparison is ambitious. NMC cathodes, which combine nickel, manganese, and cobalt, support high energy density in many electric vehicles. LFP offers lower material costs, long cycle life, and strong thermal stability, but usually stores less energy for a given weight.

Matching NMC-like performance with an LFP-like cost position would address a valuable market gap. Yet every word in that comparison needs a common measurement basis. Cell design, test temperature, cycle protocol, production volume, and warranty requirements can all change the outcome.

Holyvolt’s experimental platform should help reveal those tradeoffs earlier. It cannot erase them. AI can improve which candidates researchers test, while high-throughput equipment can increase the number of trials. Neither guarantees that the selected chemistry meets every commercial requirement.

This is the mechanism behind the funding thesis. Holyvolt is not claiming that software replaces chemistry. It is betting that better software, proprietary data, and repeated physical tests can reduce wasted work before expensive manufacturing begins.

The Real Opponent Is China’s Manufacturing Scale

Holyvolt is competing against an industrial system that already combines materials, equipment, suppliers, customers, and enormous production volume.

The International Energy Agency estimates that China produced about 85% of the world’s cathode active material in 2025. It also accounted for more than 80% of battery cell production and over 90% of anode active material.

Cathode active material is especially important because it strongly influences a battery’s energy storage and cost. The IEA estimates that it represents 40% to 50% of cell production costs for NMC batteries. For LFP cells, the share is about 25% to 30%.

These numbers make Holyvolt’s strategic argument understandable. A Western startup that produces another conventional LFP material must compete with experienced Chinese suppliers operating inside a mature and closely connected supply chain.

China’s advantage extends beyond installed factory capacity. Producers benefit from nearby refiners, equipment vendors, cell manufacturers, automotive customers, trained workers, and logistics networks. Repetition also improves yield and process control over time.

The IEA’s battery supply chain analysis shows why building individual factories does not close the gap. New facilities can take more than five years after opening to approach their nominal production output.

A different chemistry offers one possible route around that position. If Holyvolt finds a material with attractive performance and inputs available outside China, Western manufacturers could build a supply chain around a new starting point.

That strategy avoids a direct contest over who can manufacture today’s dominant chemistry most efficiently. It replaces that contest with another one: who can qualify a new chemistry before established producers respond.

China’s battery companies are not standing still. CATL, BYD, and their suppliers invest in materials research, cell design, and manufacturing improvements. They can also integrate successful chemistry changes into large customer networks.

Holyvolt therefore needs more than scientific novelty. Its material must create enough value to justify qualification costs and the risk of switching suppliers. A battery producer will not redesign a cell simply because a material emerged from an AI-guided workflow.

The first target market could make that transition easier. Holyvolt plans to begin commercializing DRX in 2027, initially for unmanned aerial vehicles. Drones place a premium on weight, power, and supply security, particularly in defense applications.

The market also faces a policy deadline. Starting October 1, 2027, US defense procurement restrictions cover batteries produced by six named Chinese companies, including CATL and BYD. The measure creates demand for alternatives that satisfy both technical and sourcing requirements.

The restriction does not guarantee a contract for Holyvolt. Defense buyers still need safe, dependable cells, while battery integrators need materials they can process consistently. The law changes purchasing boundaries, but qualification determines which suppliers benefit.

The company is also developing formulations for automakers and data centers, according to Fortune. Data centers sometimes need batteries that can deliver rapid bursts of power when computing loads change. Automakers present a larger opportunity, but their qualification cycles and manufacturing requirements are demanding.

Holyvolt’s near-term route reflects those differences. A specialized drone application can accept lower volume while rewarding energy density and secure sourcing. Automotive adoption requires high output, aggressive cost control, extensive validation, and long operating histories.

The company also has a more conventional path into production. Wildcat and EnergyX announced plans for an LFP cathode material facility in Texas with annual capacity targeted at 15,000 metric tons. That project could give the group practical manufacturing experience before DRX reaches scale.

However, LFP production creates a strategic tension. It can establish customer relationships and manufacturing processes, but it also places Holyvolt closer to the competition Prast says the West cannot win through imitation.

The distinction lies in purpose. Holyvolt can use LFP as a manufacturing bridge while directing its differentiated research toward DRX. The plan works only if the bridge generates operational knowledge without consuming the resources needed for the newer chemistry.

The Holyvolt funding is therefore a bet on sequencing. The company wants laboratory data first, application-specific commercialization next, and broader markets later. That order reduces some risk, but every transition exposes the company to China’s advantages in speed, scale, and cost.

What the $245M Valuation Does Not Prove

A reported valuation signals investor confidence, but it says little about cell performance, factory yield, or customer acceptance.

Fortune attributes Holyvolt’s $245 million valuation to a person familiar with the company. Holyvolt has confirmed the $45 million financing, but it has not publicly confirmed that valuation.

That difference matters. Reported financing terms can establish how investors valued a company during a transaction. They do not independently verify technical claims or predict commercial success.

Holyvolt’s research platform has credible operating history through Wildcat. Its reported project count, customer relationships, experimental volume, and years of activity show that it is more than a newly assembled software demonstration.

Yet research services and materials commercialization are different businesses. A laboratory can help customers screen formulations without owning the risk of producing those materials at scale. A materials supplier must deliver consistent batches and support long qualification programs.

DRX also remains a developing category. Researchers have identified attractive theoretical and experimental properties, but challenges include voltage decline, cycling stability, and rate performance. Improvements in one dimension can introduce costs elsewhere.

Manufacturing can expose further problems. A process that works with small powder batches may produce different particle distributions at larger scale. Coating uniformity, moisture sensitivity, contamination, heat treatment, and material handling can influence the final cell.

Holyvolt’s screen-printing technology introduces another layer of validation. The company says it can produce battery components with lower energy use and simpler equipment. Customers will need evidence that the method delivers competitive throughput, yield, quality, and durability.

The water-based process also sounds attractive because conventional electrode production can rely on solvents and energy-intensive drying. Still, environmental advantages must be measured across the full process, including material preparation, equipment utilization, scrap, and recycling.

None of these concerns invalidates Holyvolt’s approach. They define the evidence needed to judge it.

The most useful skeptical position is not that AI cannot contribute to battery research. Machine learning can prioritize experiments, identify patterns across many variables, and help researchers allocate laboratory capacity.

The stronger question is whether faster discovery changes the slowest part of commercialization. Battery materials still pass through synthesis development, cell integration, safety testing, pilot production, customer qualification, and supply-chain approval.

If those later steps dominate the schedule, reducing discovery time produces only a partial benefit. Holyvolt’s integrated structure attempts to address that problem by considering manufacturing earlier. Its execution will show whether those connections remove delays or merely identify them sooner.

Customer evidence will carry more weight than broad performance comparisons. Named evaluations, sample shipments, repeat orders, and qualification milestones would show that external engineering teams can reproduce the company’s results.

Production evidence will matter equally. Batch consistency, manufacturing yield, throughput, energy use, and delivered cost will determine whether Holyvolt’s process has an advantage outside its own models.

Investors are funding a plausible mechanism. Wildcat contributes physical infrastructure, a historical dataset, and experienced battery researchers. Holyvolt contributes a manufacturing thesis and a strategy built around Western supply chains.

However, vertical integration increases the number of capabilities the company must manage. Research, materials production, electrode processing, application engineering, and customer qualification each require different expertise.

The company also spans multiple target markets. Drones, vehicles, stationary storage, and data centers value different combinations of energy, power, life, safety, and cost. Serving them with specialized formulations can expand the opportunity, but it can also divide development resources.

Holyvolt must prove that its platform creates repeatable products, not just more candidates. Until then, the valuation reflects expectations about the path rather than evidence that the company has completed it.

Three Signals That Will Test the Holyvolt Funding Thesis

The next evidence should come from product validation, manufacturing execution, and customer adoption, in that order.

The first signal is independently comparable DRX performance data. Holyvolt says the material can approach nickel-based performance without nickel or cobalt, while reaching a cost position near LFP in North America.

Readers should look for full test conditions rather than a single headline metric. Relevant disclosures include energy density, cycle retention, rate performance, operating temperature, safety behavior, and the cell format used.

Third-party replication would strengthen Holyvolt’s argument. Results produced by customers, research partners, or independent laboratories would show that the chemistry performs beyond Wildcat’s own equipment.

This evidence would also clarify where the material fits. A high-energy drone cell has different requirements from an automotive cell expected to operate across seasons for many years. A formulation can succeed in one application without being ready for another.

The second signal is progress from samples to repeatable pilot manufacturing. Holyvolt’s platform is designed to connect discovery with production, so manufacturing milestones test its central claim more directly than another model improvement.

Investors and customers should watch for qualified production equipment, increasing batch sizes, stable yields, and consistent material delivered across multiple runs. The Texas LFP project can provide relevant operational evidence, even though it uses an established chemistry.

Holyvolt’s planned 2027 DRX commercialization creates a clear timeline. Sample delivery, partner testing, and initial production readiness must advance before meaningful sales can follow.

The company does not need automotive volume immediately. It does need a credible progression from laboratory cells to customer-relevant formats. Each stage should preserve the performance that justified moving forward.

Manufacturing data would either strengthen or weaken the Holyvolt funding thesis. Stable production at growing volume would support the claim that discovery and process design can advance together. Repeated delays would suggest the integrated platform has not removed the traditional scaling bottleneck.

The third signal is a named customer moving into qualification or deployment. The 2027 defense procurement deadline can create urgency, especially for drone makers seeking battery components outside the restricted Chinese supply base.

The underlying procurement restriction identifies CATL, BYD, Envision Energy, EVE Energy, Gotion High-Tech, and Hithium. A Western supplier with qualified material could benefit, but only after meeting program requirements.

A customer announcement should include enough detail to distinguish evaluation from adoption. A research collaboration shows interest. A qualification program indicates deeper technical engagement. A production order demonstrates that a product has crossed a commercial threshold.

Automotive qualification would be the most consequential signal, but defense or specialized industrial use could arrive first. Those markets can support lower volumes while providing operating data that informs later versions.

Data-center applications deserve similar scrutiny. Holyvolt says it is working on a formulation for high-power demand when GPUs ramp up. The relevant proof would involve power delivery, response time, cycle behavior, cooling needs, and deployment economics.

These three signals create a straightforward test for the story. Independent performance data evaluates the material. Pilot manufacturing evaluates the process. Customer qualification evaluates the market.

The $45 million round gives Holyvolt resources to pursue all three. It also raises expectations that the company will publish more than research counts and theoretical comparisons.

For developers, enterprise buyers, and AI product teams, the story offers a broader lesson. AI creates the most value in physical industries when it connects directly to reliable measurements, domain expertise, and operational feedback.

The next year should show whether Holyvolt has built that connection deeply enough. Watch for reproducible DRX data, stable pilot output, and named customer qualification. Those results, rather than the valuation alone, will determine whether Holyvolt funding accelerated a new battery platform or simply financed a longer experiment.