European BER10 chip: powerful and energy-efficient for AI

Europe is setting a milestone in digital sovereignty with the BER10, a chip designed by Openchip. This processor combines high performance with energy efficiency, specifically tailored for demanding AI workloads. With four high-performance CPU cores, vector processing, and support for Linux, the BER10 is ready for immediate deployment in data centers, critical public services, and cloud infrastructure.

A European breakthrough in chip technology

The BER10 is not an academic prototype, but industrial-grade silicon that is ready for immediate use with real-world workloads. Openchip, based in Barcelona, developed this chip in just 2.5 years, from inception to first silicon. The tape-out took place in 2025, and on July 30, 2026, the first functioning version was unveiled. The chip is designed for sectors such as healthcare, banking, government, and AI, where reliability and control are crucial. By leveraging the open RISC-V architecture, Europe avoids dependence on external chip suppliers.

Power and efficiency in a single design

The chip combines power and energy efficiency through a smart mix of hardware innovations. It features four high-performance CPU cores with out-of-order execution, meaning it doesn’t have to wait for slower operations. In addition, a vector processing unit is integrated, specifically designed for parallel computations. This is essential for AI, image processing, and scientific computing. The combination with fast memory and data interfaces such as PCIe and AXI ensures an optimal balance between performance and power consumption. The chip is manufactured using sub-2nm Gate-All-Around technology, one of the most advanced manufacturing processes in the world, which contributes to its efficiency🔗︎.

The BER10’s efficiency lies in its architecture. It is based on the open RISC-V instruction set architecture (ISA), which allows designers to customize the processor by integrating domain-specific accelerators and custom instructions. This makes it possible to add dedicated AI hardware, such as Neural Processing Units (NPUs), which are optimized for neural network calculations and can perform these tasks more efficiently and with lower power consumption than a general-purpose CPU. Specialized acceleration engines for matrix operations provide similar benefits by speeding up the calculations that form the core of many AI workloads while avoiding unnecessary processing overhead.

Impact on European autonomy and economy

The introduction of the BER10 has a direct impact on European digital sovereignty. By running AI workloads locally and independently, Europe reduces its dependence on foreign technologies. This is crucial for sectors such as healthcare, finance, and government, where control over sensitive data and AI applications is essential.