About a-team Marketing Services
The knowledge platform for the financial technology industry
The knowledge platform for the financial technology industry

A-Team Insight Blogs

BSO Unveils DataOne: Europe’s First Giga-Scale AI Hosting Data Centre

Subscribe to our newsletter

BSO, the global infrastructure and connectivity provider, has announced the creation of DataOne, Europe’s first giga-scale AI hosting infrastructure, following months of confidential development. Located in France, the project is backed by Ardian Debt Fund alongside leading commercial banks and represents a significant step forward for AI-compatible data centre technology in Europe.

The site’s infrastructure will be optimised for next-generation GPU technologies, featuring liquid-cooled Facilities Distribution Units (FDUs) capable of managing power densities between 60–250KW per rack.

GPUs on Demand

“This is a game-changer for financial institutions and the markets overall,” declares BSO’s CEO & Co-Founder Michael Ourabah, in conversation with TradingTech Insight. “Quantitative traders rely heavily on GPUs to model strategies and anticipate market trends, and we’ve already received requests from customers for GPU-as-a-Service solutions. For trading firms and financial institutions, the biggest priority is access to large volumes of GPUs on demand, with rapid turnaround times.”

He continues: “We see demand from two main types of clients in the financial sector. The first group consists of firms with hundreds of trading teams that need their own dedicated platforms. These clients are often willing to take 2–5 megawatts of capacity, bringing their own GPUs to operate independently. The second group includes firms that don’t need dedicated platforms but want access to a GPU-as-a-service offering. Through our platform, these clients will be able to order GPU capacity directly via a portal or API, with a bare-metal solution that allows them to push their workloads to the GPUs quickly and efficiently.”

High Density Infrastructure

The new DataOne facility, developed on a Tier-3 data centre campus acquired by BSO, spans 50,000 m² across Grenoble (Eybens) and Lyon (Villefontaine). With an initial 15MW IT load, the site is ISO 27001, 14001, 50001, and SOC 1 & 2 certified. Its carrier-neutral infrastructure is interconnected via dark fibre and 100G+ DWDM wavelengths, supported by seven connectivity providers, with plans to expand further. Its advanced design achieves a Power Usage Effectiveness (PUE) rating of 1.06–1.15, exceeding European energy efficiency standards.

“We’re building on an existing low-density data centre campus and repurposing additional spaces, including large office buildings with five- to six-metre-high ceilings. These concrete buildings, which are 50–60 metres wide and 100 meters long, are ideal for conversion into high-performance data centres,” explains Charles-Antoine Beyney, Co-Founder of BSO and CEO of DataOne.

“This data centre is specifically optimised for GPUs, designed to handle extremely high densities of up to 250 kilowatts per rack,” he continues. “To put that into perspective, a standard data centre typically doesn’t exceed 15 kilowatts per rack. Here, we’re talking about cutting-edge liquid cooling systems, featuring dry coolers and active rear door heat exchangers. Chilled water is used to cool the GPU cores, CPUs, motherboards, and power supply units before being recirculated as hot water, which is then managed by the dry coolers.”

The facility’s roadmap outlines a phased expansion, with the first 80MW of capacity set to launch in April 2025, establishing DataOne as Europe’s largest AI-focused data centre. By late 2028, the site will reach its full 400MW capacity, placing it on course to be Europe’s largest data centre.

“Currently, densities of around 60–70 kilowatts per rack are required for technologies like the H100 and H200 GPUs powering systems such as Elon Musk’s Colossus supercomputer,” says Beyney. “Our design anticipates future advancements, including Nvidia’s GB200 chips, which will likely need 140–150 kilowatts per rack, and further generations that could push requirements to 250 kilowatts within four years. This infrastructure is built to be future-proof, ensuring it can accommodate the demands of next-generation GPUs—or potentially even entirely new silicon technologies that may emerge to replace GPUs.”

Sustainability

DataOne aligns with BSO’s zero-carbon vision by incorporating several key sustainability measures. The facility will operate entirely on renewable energy, primarily sourced from EDF’s hydroelectric power, with nuclear energy providing backup to ensure reliability. It will also repurpose waste heat, producing 50°C water that can be reused by local industrial and commercial communities. Additionally, customers will benefit from real-time energy transparency through an integrated portal, offering detailed access to the centre’s performance metrics.

“Sustainability is at the core of this project,” states Beyney. “We’re using hydroelectric power, and the heat generated by the data centre will be fully recycled. The city will connect to our heat exchangers, allowing the waste heat to be repurposed for nearby industrial buildings, exhibition centres, schools, and other public facilities. This approach not only reduces our carbon footprint but also provides significant energy savings for the community. It’s a win-win, as no one else is currently recycling server heat at this scale.”

The two founders envision DataOne redefining Europe’s data centre landscape, by delivering scalable, efficient, and environmentally conscious infrastructure tailored for the next generation of AI technologies. “We plan to launch the first 80 megawatts of capacity by April 2025, with GPU-as-a-service offerings available for at least 15 megawatts at that time, thanks to our customers hosted infrastructure,” says Ourabah, with Beyney adding: “This will immediately make the cluster hosted by DataOne the largest in Europe, with 16,000 GPUs coming online at launch. By the end of 2025, we aim to scale our customer’s hosting capacity up to 100,000 GPUs, which would be equivalent to the largest in the world. That’s our target, and we’re well on track to achieve it.”

Subscribe to our newsletter

Related content

WEBINAR

Recorded Webinar: Trade the Middle East & North Africa: Connectivity, Data Systems & Processes

In Partnership With As key states across the region seek alternatives to the fossil fuel industries that have driven their economies for decades, pioneering financial centres are emerging in Egypt, United Arab Emirates (UAE), Saudi Arabia and beyond. Exchanges and market intermediaries trading in these centres are adopting cutting-edge technologies to cater to the growing...

BLOG

Avelacom Enhances Network Latency Between Europe and East Asia

Ultra-low latency connectivity provider Avelacom has significantly reduced network latencies between European markets and key East Asian cities, including Tokyo, Shanghai, and Hong Kong. The company claims that it has set new speed records, with the round-trip latency for its London to Shanghai route now under 125 milliseconds, surpassing the previous industry benchmark of approximately...

EVENT

TradingTech Summit MENA

The inaugural TradingTech Summit MENA takes place in November and examines the latest changes and innovations in trading technology and explores how technology is being deployed to create an edge in sell side and buy side capital markets financial institutions in the region.

GUIDE

Institutional Digital Assets Handbook 2024

Despite the setback of the FTX collapse, institutional interest in digital assets has grown markedly in the past 12 months, with firms of all sizes now acknowledging participation in some form. While as recently as a year ago, institutional trading firms were taking a cautious stance toward their use, the acceptance of tokenisation, stablecoins, and...