China produces more than 80% of the world's battery cells, while Europe struggles to scale up manufacturing in a booming market. As industrial ambitions face mounting challenges, digital twins and virtual testing offer a promising path to cut battery development times from years to hours.

It will probably end like Achilles and the Tortoise, Zeno of Elea’s famous paradox in which the hero never quite manages to catch up with the animal, despite running much faster. In this case, the ultimate prize is electric battery production. Europe is trying to accelerate in order to close, at least partially, a huge gap in the strategic industrial sector of electric batteries. But China, which enjoys a significant head start because it began the race much earlier, may ultimately remain out of reach.

Europe and the challenge of battery autonomy

Today, China accounts for more than 80% of global battery cell production, while Europe's share remains marginal in a rapidly expanding market. According to the International Energy Agency (IEA), global electric vehicle battery deployment is expected to more than triple by 2035. As the agency notes, “to put this in perspective, on average, a single month of EV sales in 2035 would exceed the entire annual deployment of 2021”.

Europe risks missing a train that is accelerating fast. Companies operating in the European battery market are struggling to survive, as illustrated by the recent difficulties and failures of players such as Northvolt. Research is advancing, but industrial production is lagging behind.

But what if Europe could test electric batteries while dramatically reducing physical testing on real batteries and shifting much of the validation process to digital models? Development cycles, which currently can take up to five years from design to industrial deployment, could be significantly shortened, while testing costs could be significantly reduced. Europe’s pursuit of greater autonomy would gain momentum, even if full independence in this strategic sector remained a distant goal.

The numbers are stark. In 2025, China had approximately 4 TWh of lithium-ion cell manufacturing capacity, compared with around 210 GWh in Europe. In other words, Chinese production capacity was roughly 19 times greater than Europe's.

“Compared with China, Europe faces a fifteen-year gap in the battery sector – says Christophe Pillot, Director of Avicenne Energy – China started earlier and is moving much faster. Take factory construction as an example: in China it takes around two years to build a battery plant, while in Europe it takes four years on average, and up to five years in Germany”. One of the first bottlenecks in Europe’s still-young and relatively fragmented battery industry is the scale-up phase: the transition from laboratory research to industrial production. Other challenges follow, including the limited capacity to recycle used batteries and recover critical raw materials as part of a broader strategy for industrial autonomy. This is crucial while 1.2 million electric vehicle batteries could reach the end of their lives in 2030 and 14 million in 2040 (IEA’s data), and today’s supply chains for battery minerals and components are highly concentrated. In this context “some European gigafactories have shipped brand-new batteries, produced only for testing purposes and never actually used, to South Korea for recycling instead of recycling in Europe”, Pillot states.

Europe bets on research

According to Silvia Delbono, Funding Project Manager at Flash Battery, Europe's competitiveness in the battery sector can be strengthened through an approach increasingly centred on methodological innovation in development processes.

“In a global context characterised by rapid change and strong industrial pressure – Delbono explains – competitive advantage depends not only on access to resources or manufacturing capacity, but also on the quality and efficiency of design and industrialisation processes.”

Against this backdrop, the European THOR project aims to foster closer integration between modelling, advanced simulation and experimental validation. The focus is on predicting battery behaviour, identifying and solving critical issues at the earliest stages of development in order to reduce time-to-market, limit physical testing and lowering the costs.
From years of testing to hours of simulation.

How? By using a digital twin: a virtual avatar of the battery that relies on artificial intelligence to process data and simulate battery behaviour under a wide range of operating conditions.
It is a kind of “Back-to-the-Future scenario”: an ambitious challenge that could make it possible to simulate “thousands of situations in a matter of hours”, whereas testing them on physical batteries could take years, experts say.

The concept of a digital twin is not new. It is already widely used in several industrial sectors, including aerospace, automotive and energy, particularly for the simulation and monitoring of complex systems.

“In the specific case of lithium-ion batteries – Delbono adds – large-scale adoption is still in a consolidation phase. This is precisely why several European projects, including THOR, are working to demonstrate the value of the digital twin.” Flash Battery contributes to the project through its expertise in advanced battery management systems (BMS) and in data collection and exploitation.

Data is at the heart of the research effort. At this stage, THOR's digital twin is being trained using experimental and simulated datasets. In practice, this means feeding the virtual model with large amounts of data so that it can progressively learn to mirror how real batteries age, degrade and perform under different conditions. Once fully operational, it could perform thousands of simulations almost instantly.

The ultimate goal is to create a digital replica of the battery capable of predicting its behaviour over time by estimating key parameters such as state of charge (SoC), state of health (SoH) and remaining useful life (RUL). The system will operate at different levels, from individual cells to battery modules and complete battery packs.

The THOR project focuses on cylindrical lithium-ion batteries based on NMC (nickel-manganese-cobalt) and LFP (lithium iron phosphate) chemistries, both using graphite and widely employed across industry. Project partners carry out physical testing on battery performance, ageing and safety. The experimental results are then used to develop and validate simulation models.

Ioannis Boutopoulos, AI, System Simulations & CFD Engineer at FEAC, which is responsible for developing the digital twin, explains that "What the digital twin will do is learn, from all this data, the patterns of the battery. From parameters such as voltage, current and temperature, we may be able to predict the state of charge, the state of health and the remaining useful life of the battery."

Predicting the unpredictable: the reliability challenge

The undertaking is not without obstacles. One of the main challenges in battery modelling is that battery behaviour is highly non-linear. “The challenge to create a deep learning model based on this data is not straightforward because the internal processes inside batteries are not linear and it is very difficult to capture these phenomena – Boutopoulos adds – Even with neural networks, a great deal of tuning is required, and in many cases we need additional data and more experiments.”

Another major challenge concerns certification. At present, test results obtained through digital twins cannot simply replace conventional validation procedures.

“The industry is still cautious about fully adopting AI-based tools, mainly due to safety and certification requirements. Since these batteries are used in critical applications, including industrial machinery, ensuring reliability and trust in AI-driven predictions is a major concern”, Boutopoulos explains.

If an electric truck battery catches fire, manufacturers and regulators cannot simply respond that the algorithm failed to predict the event. Researchers are therefore working to standardise and validate virtual tests based on digital twins, so that the results can eventually be recognised by certification authorities and accepted by industry.

The broader issue of industrial adoption extends well beyond the battery sector. The use of digital twins is still at a relatively early stage. The technology is promising, but its widespread integration into industrial processes is likely to take time.

The road ahead is marked by numerous obstacles and a huge gap with China that Europe is still struggling to bridge. All this comes at a time of growing geopolitical uncertainty. So what should Europe do to strengthen its strategic autonomy? For Christophe Pillot, the answer is clear: public aid. “Europe currently put around €10 billion of public subsidies in the sector, but it would need closer to €100 billion – he says – and these funds should directly support industrial operations such as machinery, personnel, operating costs and salaries. In the United States, producing an electric battery costs about 30% more than in China, and Washington bridges that gap through direct subsidies to manufacturing sites. Europe should do the same”.

Contacts:
Project coordinator:
Lise Daniel, Commissariat à l’Energie Atomique et Aux Energies Alternatives, lise.daniel@cea.fr
Communication Manager:
Roberta De Michele, Fondazione ICONS roberta.demichele@icons.it

Project website: www.thorbatteries.eu
LinkedIn: THOR EU PROJECT
Bluesky: THOR EU PROJECT

By Gioia Salvatori