• Home
• Technology

Two government agencies in the Netherlands and Germany are teaming up to develop new technologies for designing custom artificial intelligence (AI) computing chips. The collaboration is part of Europe's broader push to strengthen its technological capabilities and reduce its reliance on American expertise.

The newly established Dutch National Agency for Disruptive Innovation (NADI), founded to support strategic breakthroughs, is partnering with its German counterpart, SPRIND. Together, they will execute a project focused on leveraging AI to accelerate the design of chips customized for training and running AI models.

According to Reuters, the two agencies will allocate €40 million over 20 months to fund small research teams and startups. The goal is to achieve a massive leap in the speed of chip design, a highly complex process that currently takes years.

Betting on the ASML Ecosystem

Giel Prins, co-founder of NADI, stated that collaborating with Germany is a natural step, given the strength of the semiconductor ecosystem surrounding ASML in the Netherlands, combined with Germany's prowess in scientific research and manufacturing.

ASML remains a pivotal player in the global semiconductor industry, developing the advanced lithography equipment used to manufacture the world's most sophisticated microchips.

Through NADI, the Netherlands seeks to build strategic capabilities in deep tech, modeled after the US Defense Advanced Research Projects Agency (DARPA). The Dutch agency operates with an allocated government capital of €500 million, according to its announced structure.

Challenging US Chip Dominance

The joint Dutch-German project specifically focuses on developing chips that can more efficiently handle 'inference'—the phase where AI models are deployed to generate answers or execute tasks in real-time, rather than the initial training phase.

Prins noted that while Nvidia’s chips boast immense computing power, they can be less efficient for certain inference tasks, likening their use in such scenarios to 'going grocery shopping in a truck instead of a small cart.'

Ultimately, the partners hope to use AI itself to accelerate the chip design lifecycle, enabling the rapid testing of a larger volume of designs and drastically slashing the time required to reach viable commercial solutions.