Porsche is Testing AI Intelligent Soft Switching, Promising to Cut EV Inverter Losses by 95%

AI-supported intelligent soft switching can reduce switching losses in electric vehicle (EV) power transistors by up to 95 percent, and Porsche Engineering is already validating the approach through simulations. The company reported the news on January 8, 2026.

Driving range remains one of the most important purchasing criteria for electric vehicles, so maximizing powertrain efficiency is essential. The inverter plays a central role because significant losses occur in its power transistors during switching. These losses can be substantially reduced through intelligent control of transistor operation.

In an EV inverter, there are two primary loss mechanisms: conduction (line) losses and switching losses. Conduction losses result from the fact that power transistors do not behave like ideal switches; even in the “on” state they retain a residual resistance, which causes power dissipation and heat. Switching losses occur during the brief transition between on and off states, when both current and voltage are present simultaneously across the transistor.

During these transitions, the product of current and voltage produces power loss peaks that reduce vehicle range. Higher switching frequencies increase this effect, yet are desirable because they improve AC output quality. To address this inherent conflict, Porsche Engineering applies soft switching. Rather than turning transistors fully on and off abruptly (“hard switching”), soft switching dynamically adjusts switching instants so that voltage–current overlap, and therefore switching losses, are minimized. This is accomplished using either Zero Current Switching (ZCS) or Zero Voltage Switching (ZVS). In ZCS, the transistor switches when current is near zero; in ZVS, switching occurs when the voltage is near zero.

Reduced losses using ZVS

Porsche Engineering focuses on ZVS to improve inverter efficiency. ZVS significantly lowers losses in silicon-carbide and gallium-nitride devices commonly used in EVs, performs better at higher switching frequencies than ZCS, generates less electromagnetic interference, and is particularly suitable for inductive loads such as electric motors.

To implement ZVS, an additional circuit composed of transistors, inductors, and capacitors is connected around the power switches. This inverter topology is known as the auxiliary resonant commutated pole (ARCP). The innovation now underway is the introduction of artificial intelligence to control the ARCP. A pretrained AI model processes dozens of real-time vehicle parameters, including load, torque, and temperature, and determines optimal switching points within fractions of a second.

AI-driven switching optimization

Electric vehicles are constantly exposed to changing operating conditions, historically preventing broad application of soft switching. AI removes this barrier. The algorithm predicts optimal switching instants despite rapidly varying loads and operating states, enabling full soft switching with minimal losses and ultimately higher vehicle range.

Porsche Engineering is evaluating recurrent neural networks and reinforcement learning for this task. Recurrent models provide high predictive accuracy and strong performance, while reinforcement learning is advantageous for extremely demanding real-time computing requirements.

The efficiency gains are substantial. AI-based soft switching can reduce transistor switching losses by approximately 70 to 95 percent. Depending on usage profile, this can increase vehicle range by a high single-digit percentage. Lower switching losses also reduce heat generation in the inverter, easing cooling demands and allowing more compact component designs. Some filter components can be eliminated, improving PCB layout and reducing inverter volume by 20 to 50 percent. Reduced stress on power transistors also extends component service life.

The AI control algorithm is already in an advanced development stage. Once complete, Porsche Engineering intends to offer AI-based soft switching as a software solution. It can be integrated into existing control units via software libraries with only minor hardware adjustments, making it attractive for model updates or new vehicle platforms for OEMs and Tier-1 suppliers.

EVinfo.net’s Take: AI-Supported Intelligent Soft Switching: One of Many EV Breakthroughs Pushing Gas Vehicles Toward Obsolescence

The EV industry is no longer defined by proof-of-concept technology or cautious pilot programs. It is advancing through rapid, compounding innovation in batteries, power electronics, software, charging infrastructure, and manufacturing efficiency. Porsche Engineering’s work on AI-supported intelligent soft switching illustrates how quickly EV technology is evolving, and why the age of gasoline-powered vehicles is drawing to a close.

Electric vehicles are already the most environmentally friendly, cost-effective, and fastest vehicles available to consumers. They deliver zero tailpipe emissions, lower total cost of ownership through reduced fuel and maintenance expenses, and best-in-class performance thanks to instant torque. Each improvement in efficiency, range, charging speed, or power electronics further strengthens the EV value proposition and accelerates the shift away from polluting, more expensive to own gasoline vehicles.