Researchers at Chalmers University of Technology in Gothenburg developed an AI-based fast-charging strategy that extends lithium-ion battery life by approximately 23% — without meaningfully increasing charging time. The method addresses a real problem in current practice: fast charging degrades batteries differently depending on their age, but today's charging systems don't account for this. The study was published in IEEE Transactions on Transportation Electrification.
The result, in precise terms: battery lifespan extended to 703 equivalent full cycles, representing a 22.9% improvement over the standard baseline — while maintaining comparable charging speeds.
The key mechanism is reinforcement learning — a machine learning technique in which an algorithm improves its decisions through continuous feedback from each charging cycle. The AI varies current based on charge level and overall battery health, keeping charging time within a few seconds of today's rates.
As Professor Changfu Zou of Chalmers' Department of Electrical Engineering explained: "Smart adaptation of the current during charging, taking into account the changing electrochemical state of the battery, can maximise both the performance and the life of the battery."
The underlying problem isn't charging speed itself. It's what repeated high-power charging does to battery chemistry over time.
The most damaging consequence is lithium plating — metallic lithium depositing on the electrode rather than intercalating correctly into the anode structure. This reduces capacity and can, in severe cases, cause a short circuit. The risk grows as the battery ages, yet standard protocols apply the same current and voltage to a new pack and a five-year-old one alike.
Traditional charging systems use fixed parameters regardless of the battery's current state — its age, temperature, charge level, or chemical condition. The Chalmers system discards that one-size-fits-all approach entirely.
The system monitors multiple parameters including temperature, voltage and charge state to make real-time decisions. This dynamic approach differs from conventional charging protocols that apply fixed parameters regardless of individual battery condition or environmental factors.
The 23% figure becomes more meaningful when mapped against real-world driving data.
According to a 2024 study by Geotab, the average annual degradation of an EV battery is only around 1.8% per year — suggesting batteries could last at least 20 years or 200,000 miles, with many exceeding this range. Some estimates place Tesla battery longevity between 300,000 and 500,000 miles, depending on usage and charging patterns.
A 23% improvement over that baseline translates to nearly 70,000 additional miles at lower usage levels, and more than 100,000 miles at higher ones. According to the U.S. Federal Highway Administration, Americans drive an average of approximately 13,476 miles per year — meaning this AI improvement could represent five to seven additional years of useful battery life for many drivers.
For frequent fast-charging users, that's a material difference. Not just for resale value, but for the total cost of EV ownership over a vehicle's lifetime.
A Software Fix for a Hardware Problem
One of the most significant aspects of this research is its delivery mechanism. The method could be deployed through a software update to battery management systems — meaning it doesn't require new hardware, new battery chemistry, or new manufacturing processes. It's an algorithmic improvement that could, in principle, reach existing vehicles through an over-the-air update.
This positions the technology alongside other software-defined EV improvements — the same category of update that has allowed automakers to push range improvements, charging speed adjustments, and performance tuning to vehicles already in owners' driveways.
For taxis or heavy vehicles in industry, access to fast charging means a lot — but this is also true for passenger cars. Although private motorists usually charge their electric cars at home, the availability of fast charging outside the home is a crucial factor, as it facilitates commuting and driving over longer distances, noted Professor Zou.
The Chalmers results are significant — and they come with an important caveat. Results so far are simulation-based. Physical battery validation is the next step. The charging experiments were conducted in a laboratory environment using a simulation of a common EV battery, not on physical cells in real vehicles under real-world conditions.
Before this technology can influence battery warranties, used EV valuations, or OEM charging strategies at scale, it will need to demonstrate equivalent results across different battery chemistries, temperature ranges, and usage profiles in the field.
That's a standard threshold for any promising battery research — and it doesn't diminish what the Chalmers team has demonstrated. It simply marks where the work goes next.
Battery longevity has always been one of the quieter anxieties around EV ownership — less visible than range anxiety, but more consequential for long-term cost calculations and resale confidence. Research like this signals that the industry is beginning to address degradation not just through better cell chemistry, but through smarter software controlling how energy flows into and out of those cells.
The implications extend well beyond individual vehicles. The environmental implications extend beyond individual vehicle ownership — longer-lasting batteries mean fewer battery replacements, less raw material demand, and a stronger sustainability case for EVs as a long-term transportation solution.
As Electrive reported, the Chalmers approach treats battery health not as a fixed variable to be managed around, but as a dynamic condition to be actively optimized — a shift in thinking that is likely to shape how battery management systems are designed across vehicle categories for years to come.
The AI is still learning. So is the industry. Both are improving faster than most expected.
