Fluidstack

Driven by the escalating demands of artificial intelligence (AI) research and development, Fluidstack has emerged as a pivotal player in Europe's AI infrastructure landscape. The company's latest initiatives, including its 1 GW AI supercomputer project in France and strategic partnerships with leading AI firms, mark a significant shift in how we approach scalable GPU computing. By combining cutting-edge hardware with renewable energy sources and advanced management solutions, Fluidstack is reshaping the future of AI infrastructure.

Fluidstack's GPU Cluster Solutions

Fluidstack specializes in GPU clusters with scalable capacity options, offering on-demand instances ranging from 1 to 100+ GPUs. Their infrastructure supports up to 10,000+ GPUs through fully managed Kubernetes (K8s) and Slurm environments, with deployment options for both training and inference workloads.

The company's GPU clusters can be configured with between 1 and 100+ GPUs per instance, supporting flexible scaling requirements. On-demand instances launch in under 5 minutes, while the service offers immediate access to thousands of H200 GPUs with InfiniBand networking. Shared storage capacity reaches 1PB+, and the infrastructure is built for longevity, with plans to expand to the upcoming Nvidia GB200 GPU architecture.

As of 2024, Fluidstack's GPU infrastructure includes:

• Nvidia H100 and H200 GPUs

• Ready supply of clusters from 1 to 100+ GPUs

• Support for 8-128 GPU instances

• 1-4k GPU capacity options

The company maintains 99% uptime and offers 15-minute response times through their 24/7 support team. Fluidstack's partnerships with leading AI companies demonstrate their commitment to cost-effective GPU management, having helped firms like Poolside and Krisp optimize their AI operations. The company's scalable solutions have enabled these customers to reduce cloud costs by 70% while maintaining performance and reliability standards.

Data Center Infrastructure

Fluidstack's European data centers use 100% renewable energy from hydro and geothermal sources, with a primary facility located in Iceland to benefit from the region's cold climate and low-cost renewable energy. The company's infrastructure incorporates Dell PowerEdge XE9680 servers optimized for AI workloads with Nvidia HGX H200 GPUs and NVIDIA Quantum-2 InfiniBand networking, providing best-in-class performance and reliability for AI workloads.

The data centers feature InfiniBand networking with 3.2T of bandwidth and 1PB+ of shared storage, including zero-impact ingress/egress costs for users. The infrastructure supports scalable GPU deployment options, with facilities capable of housing up to 500,000 next-generation AI chips in Phase 1, which will be operational by summer 2025.

Through its partnership with Eclairion, a French specialist in modular, high-density data centers, Fluidstack has designed the facility for future scalability beyond 1 GW. The project, backed by a €10 billion investment from Fluidstack and the French government, will provide dedicated AI compute power and align with France's digital and energy sovereignty goals. The facility's power comes from France's decarbonized energy grid, ensuring sustainable operations.

This AI supercomputer project represents a significant investment in Europe's AI infrastructure, with Fluidstack committed to building the continent's largest GPU-based supercomputer in Bruyères-le-Château, Essonne, located 40 km south of Paris. The multi-year deployment plan calls for phased rollouts to reach the facility's full 1 GW capacity by 2028, positioning France as a leading global AI hub while staying true to the country's commitment to sustainable computing practices.

AI Supercomputer Project in France

Fluidstack's 1 GW AI supercomputer project in France represents a significant investment in Europe's AI infrastructure. The project, backed by a €10 billion investment from Fluidstack and the French government, will provide up to 1 gigawatt (GW) of dedicated AI compute power, exclusively using decarbonized energy from France's nuclear and renewable resources.

The facility will be connected to the French national electricity transmission network through RTE, allowing access to the country's abundant and controllable nuclear energy. The project aims to reach full 1 GW capacity by 2028, with the initial phase providing close to 500,000 next-generation AI chips. The supercomputer will be located in Bruyères-le-Château, Essonne, with the potential for expansion to 100 MW+ through Eclairion's modular, high-density data center design.

The facility incorporates energy-efficient technologies for sustainability, including waste heat recovery systems. The project demonstrates France's commitment to digital and energy sovereignty, with President Emmanuel Macron emphasizing the country's role in AI innovation due to its stable and decarbonized nuclear energy supply.

The supercomputer development follows Fluidstack's recent partnerships with leading AI companies and its collaboration with Macquarie Group to provide GPU financing solutions for European research labs, highlighting the growing importance of dedicated AI infrastructure in the region.

Partnerships and Funding

In a significant collaboration, Fluidstack and Macquarie Group have formed a strategic GPU financing partnership to serve premier AI research labs across Europe. This innovative partnership brings together Macquarie's expertise in asset-backed financing with Fluidstack's advanced AI cloud platform, enabling leading AI research institutions to access cutting-edge computing resources without long-term commitments.

"The financing structure allows Fluidstack to provide AI compute to customers like Mistral, Character.AI, and Poolside without the burden of multi-year contracts," explains a company representative. "This flexibility is crucial as AI labs scale their infrastructure for training and deploying next-generation models."

Key aspects of the partnership include:

• Flexible-term financing arrangements for GPU purchases

• Dedicated support for AI research labs seeking scalable computing solutions

• Tailored financing models to address specific customer needs

• A strategic focus on European market growth and AI innovation

The collaboration builds on Fluidstack's existing portfolio of AI infrastructure solutions, including GPU clusters, high-performance storage, and fully managed Kubernetes environments. The company's European data centers already leverage 100% renewable energy and feature InfiniBand networking for optimal AI workloads.

Leadership Team and Strategy

This strategic leadership reshuffle positions Fluidstack for significant growth and innovation in AI infrastructure. Rob Perdue, brought in as COO, brings proven expertise from his tenure at The Trade Desk, where he grew the company from $1 million to $700 million in revenue. His successful track record of operational scaling and enterprise growth will be crucial as Fluidstack powers the company's European expansion.

Dan Carpenter, hired as VP of Sales, brings a decade of experience leading global enterprise sales organizations at AWS and Omniva. His role will focus on global go-to-market strategy, building deeper collaborations with AI research labs, venture capital firms, and enterprise clients. Under his leadership, Fluidstack plans to expand its partner ecosystem significantly, following its recent agreements with leading AI companies including Poolside and Character.AI.

Mike McDonald, appointed as the VP of Product, brings hands-on experience building cloud platforms at both hyperscalers (Google, Microsoft) and startup unicorns (Crusoe). His track record includes scaling Crusoe's GPU cloud from $0 to over $100 million annual recurring revenue (ARR) and hands-on experience with key AI development tools like GitHub Copilot and Google Cloud Run serverless containers. McDonald's expertise will be crucial as the company designs and operates next-generation AI infrastructure solutions.

The leadership appointments complement Fluidstack's strategic partnerships and funding initiatives. The company has secured a significant €10 billion investment for its French supercomputer project, working closely with the French government and data center providers to build Europe's largest dedicated AI infrastructure facility. This strategic collaboration positions France as a leading hub for AI innovation, leveraging the country's stable and decarbonized nuclear energy supply to power cutting-edge AI research.

Through these partnerships and its focus on sustainable infrastructure, Fluidstack is well-positioned to deliver the next generation of AI computing resources while driving growth across its European customer base.