AI chipmaker Groq expands to Europe with new Finland hub

Artificial intelligence chipmaker Groq has launched its first European data centre in Helsinki, Finland, in partnership with global infrastructure provider Equinix.

The $2.8 billion US-based startup, backed by investors including Samsung and Cisco, is expanding its international footprint as demand for AI inferencing surges across Europe.

The move positions Groq to capitalise on the continent's rising interest in sovereign AI infrastructure while taking aim at Nvidia's dominance in the sector.

Unlike Nvidia, Groq focuses on chips tailored for inferencing tasks—interpreting live data using pre-trained AI models—rather than model training.

Helsinki site to serve traffic within days

Groq’s Helsinki facility will go live this week, just four weeks after the company made the decision to build it.

The company said it is already unloading server racks at the location.

Equinix, known for linking major cloud providers including AWS and Google Cloud, will host Groq’s LPUs (language processing units), giving customers immediate access to Groq’s AI inferencing technology.

The Nordic region has become a preferred location for AI data infrastructure due to its cooler climate, renewable energy access, and political support for sovereign AI strategies.

Finland, Sweden, and Norway are increasingly seen as attractive hubs for data centre investments, and Groq’s entrance further validates the trend.

In June, Nvidia CEO Jensen Huang visited Europe and signed several infrastructure deals of his own, highlighting the intensifying competition in the space.

AI inference chips see surging demand

Groq is one of several startups seeking to disrupt Nvidia’s stronghold over AI semiconductors.

While Nvidia dominates the training segment of AI with its high-performance GPUs, companies like Groq, Cerebras, SambaNova, Ampere (soon to be acquired by SoftBank), and Fractile are betting on the growing need for inference chips.

"As demand for AI inference continues at an ever-increasing pace, we know that those building fast need more – more capacity, more efficiency, and with a cost that scales," said Jonathan Ross, CEO and Founder of Groq.

Groq’s chips are built differently from Nvidia’s. Its LPUs don’t rely on high-bandwidth memory (HBM), a component that is expensive and in short supply globally.

Instead, Groq uses a supply chain that is mostly based in North America, which helps ensure consistent availability.

That strategy supports Groq’s goal of delivering high-volume, lower-margin inference computing at scale—precisely the area where Nvidia is less focused.

Sovereign AI drives localised expansion

Europe's political push for sovereign AI—where local infrastructure hosts and processes data within regional borders—is shaping how foreign AI companies enter the market.

By establishing a physical presence in Helsinki, Groq is aligning with EU regulatory preferences and improving the latency of its services for European users.

Groq’s data centre in Finland will also benefit from the country’s commitment to green energy and its well-developed digital infrastructure.

Helsinki’s connectivity to other parts of Europe makes it a strategic choice for delivering low-latency inference services to businesses across the region.

Groq already operates data centres in the US, Canada, and Saudi Arabia, and this Helsinki facility marks a significant step in its global growth strategy.