What Sesterce does

Sesterce SAS is a French private company (société par actions simplifiée) registered in Marseille in 2017, operating as "The European AI Factory company" and building vertically integrated sovereign AI infrastructure across energy, facility, silicon, fabric, and orchestration layers. The company operates three AI factory sites in France — Bosquel (a 1 GW campus developed in a majority-owned SoftBank joint venture announced June 1, 2026), Valence (operating clusters of 1,024 H200 GPUs at 96.4% utilization), and Paris (hosted at Equinix, with 256 H100s and 512 H200s in burn-in) — plus an on-demand GPU cloud spanning 25+ regions globally. Its product portfolio consists of three layered offerings: Sesterce AI Factories (physical infrastructure, direct-to-chip liquid-cooled Tier III+ facilities with PUE 1.12 and 150 kW/rack density), Sesterce OS (a proprietary Slurm and Kubernetes-native control plane with zero-touch provisioning and rack-level DCGMI telemetry), and Sesterce Cloud (the commercial on-demand GPU rental and serverless AI inference platform, accessible via web, CLI, API, and Terraform).

The company generates revenue through three primary mechanics: usage-based GPU rental on Sesterce Cloud (pay-per-second on-demand, reserved, and spot pricing across B300, B200, H200, H100, A100, L40S, and other NVIDIA SKUs), multi-MW AI Factory capacity reservations and colocation with custom commercial terms (1-business-day sales response, NDA/DPA-ready), and a serverless AI Inference service priced hourly with autoscaling and anycast routing across 180 points of presence. Customers include EU financial institutions (Societe Generale, BNP Paribas, Credit Agricole), enterprise technology buyers (F5, NVIDIA, VAST Data, Patriarche, Flex AI), and a long-tail of self-serve developers and AI-native startups globally. Go-to-market is hybrid: enterprise field sales for multi-MW reservations via dedicated Account Executives and Solutions Architects, plus self-serve API-first and community-led motion via ML systems hackathons co-organized with PyTorch, SemiAnalysis, GPU Mode, Verda, and Prime Intellect. Sesterce is pursuing SOC 2 Type 2, GDPR compliance, and SecNumCloud readiness, and has secured 2 GW of long-term clean-energy PPAs across hydro, nuclear, and solar in Europe to anchor its sovereign positioning.

Name

Sesterce

Legal name

Sesterce SAS

Website

https://sesterce.com

Company type

Private

Founded year

2017

Operating status

Operating

Headcount range

1–10 employees

Short description

Sesterce SAS, founded in 2017 and headquartered in Marseille, operates sovereign AI factory infrastructure across three French sites (Bosquel, Valence, Paris) and an on-demand GPU cloud spanning 25+ regions, serving EU financial institutions, frontier AI labs, and global developers via its proprietary Sesterce OS control plane and Sesterce Cloud platform.

Ownership category

akta.pro rank

Product category

Sovereign AI Infrastructure / GPU Cloud Computing

NAICS

Computer Systems Design and Related Services (54151)

SIC

Services-Computer Integrated Systems Design (7373)

akta.pro primary industry

AI Server Systems & HGX/Accelerator Platforms (HDAAAAAB)

akta.pro secondary industries

AI Compiler, Runtime & Kernel Optimization Software (CUDA/ROCm/XLA, graph compilers) (HDAAAAAI), Edge AI Hardware & NPUs (SoCs, modules, accelerators) (HDAAAJAC), Compute & Acceleration ICs (CPU/GPU/AI/FPGA/ASIC) (HDAHABAC)

Keywords

• Sovereign AI infrastructure
• GPU cloud computing
• AI inference services
• AI factory colocation
• HPC cluster orchestration

Headquarters

HQ city

Marseille

HQ country

France

HQ region

Europe

Offices5 records

Markets served

GTM type

B2B

Offering type

Software

Cost components

Infrastructure, Technology or R&D, Personnel, Supply Chain, Marketing or Sales, Operations

Revenue model

1. Sesterce Cloud GPU rental (on-demand, reserved, spot): Pay-per-second on-demand GPU instances on cloud.sesterce.com across multiple global regions; reserved and spot tiers for frontier training runs. Public pricing disclosed per GPU/hour for B200, H200, H100, A100, L40S, A5000, A6000, RTX6000Ada, V100, L4, L40, A10, A16, A4000, RTXPro6000, B300.
2. AI Factory capacity reservations and co-location: Multi-MW capacity reservations (from 1 GPU on-demand to multi-megawatt training pods and co-location) with custom commercial terms and dedicated solutions architects; 1-business-day response time.
3. AI Inference service (pre-charged and custom models): Hourly-priced managed inference endpoints with autoscaling, edge routing, and pre-charged public model catalog (Llama, Mistral, Pixtral, Whisper, Stable Diffusion variants) as well as custom Docker registry deployments.
4. AI Factory infrastructure development joint venture (SoftBank JV): Majority-owned SoftBank JV developing the 1 GW Bosquel campus; combines Sesterce's data center capabilities with SoftBank's global AI infrastructure platform, supported by a €10 million community endowment fund.
5. Long-term Power Purchase Agreements (PPAs): Revenue tied to long-term clean energy PPAs for AI factory sites; the Energy Procurement Manager role explicitly owns PPA sourcing, supplier negotiations, and site expansion planning.

Pricing tiers

Go-to-market motion5 records

Distribution channels5 records

Marketing channels8 records

Core offering

Sesterce operates vertically integrated sovereign AI factories in France — purpose-built, direct-to-chip liquid-cooled data centers (PUE 1.12, 150 kW/rack) running the latest NVIDIA GB300/B200/H200 silicon on 800 Gb/s InfiniBand fabric — and exposes this capacity commercially through Sesterce Cloud, an on-demand GPU compute and serverless AI inference platform (cloud.sesterce.com) controlled by the proprietary Sesterce OS control plane (Slurm + Kubernetes native, zero-touch provisioning). The portfolio spans self-serve pay-per-second GPU rental, multi-MW enterprise reservations, managed AI inference endpoints, and strategic JV capacity, all backed by 2 GW of long-term clean energy PPAs in Europe.

Product overview

Sesterce offers a vertically integrated sovereign AI infrastructure platform sold under a platform-plus-modules architecture, marketed as 'The European AI Factory company.' The top-level product is the Sesterce AI Factory physical infrastructure (three French sites: Bosquel with SoftBank, Valence, and Paris at Equinix), exposed to customers through three layered offerings. Sesterce OS is the unified control plane (Slurm + Kubernetes, DCGMI telemetry, zero-touch provisioning) that runs across all factory sites. Sesterce Cloud is the on-demand commercial product built on top, with a Compute module (VMs and bare-metal servers spanning B300, B200, H200, H100, A100, L40S, L4, RTXPro6000, etc., across 10+ countries), an AI Inference module (serverless LLM/diffusion/Whisper/ResNet endpoints with anycast routing), persistent Storage, and Cluster orchestration. Customers access Sesterce Cloud through the Sesterce CLI, official Terraform providers, and the Sesterce Cloud API. The portfolio ties every layer (energy, facility, silicon, fabric, orchestration) into a single system, as described on the Sesterce homepage.

Differentiator

Problem solved

Functional benefit

Brands

• Sesterce Cloud: Cloud GPU rental and on-demand AI inference platform (cloud.sesterce.com) offering compute instances, serverless inference, clusters, and storage across multiple global regions.
• Sesterce OS
• Sesterce Datacenters

Products and services

• Sesterce AI Factories Purpose-built, sovereign AI factory sites across France combining energy procurement, direct-to-chip liquid-cooled facilities (Tier III+, PUE 1.12, 150 kW/rack), NVIDIA GB300/H200/B200 silicon, and 800 Gb/s non-blocking InfiniBand Quantum fabric. Three live sites: Bosquel (Hauts-de-France, 1 GW JV with SoftBank), Valence (France, 1,024 H200s at 96.4% utilization), and Paris (Equinix, 256 H100s). Sold via multi-MW capacity reservations, on-demand/reserved/co-location agreements with custom commercial terms and a 1-business-day response SLA.
• Sesterce OS Proprietary control plane and operating system for sovereign AI compute. Slurm and Kubernetes native, multi-tenant, with zero-touch provisioning (signed contract to live cluster in under 2 hours), rack-level DCGMI telemetry (P99 scrape under 11 ms), and single-pane fleet observability across every Sesterce site. Powers provisioning, orchestration, metering, and billing for Sesterce AI Factories and Sesterce Cloud.
• Sesterce Cloud On-demand GPU cloud platform offering Compute instances (VMs and Bare Metal across many global regions), AI Inference instances (serverless model deployment with autoscaling and anycast endpoints), Clusters, and Storage volumes, billed on-demand, reserved, or spot. Accessible via Sesterce CLI, official Terraform providers, and the public Sesterce Cloud REST API. Powers the self-serve motion for AI developers, startups, and enterprise platform teams worldwide.
• Sesterce Cloud AI Inference Serverless AI inference service for deploying custom or pre-charged public models (LLMs, diffusion, Whisper, ResNet-50, distilbert-base) close to end-users via anycast endpoints with smart routing across 180 points of presence worldwide. Supports autoscaling, pod lifetime, custom Docker registries, environment variables, and OpenAI-compatible APIs. Targeted at application developers and product teams needing low-latency global AI model serving.

Quantifiable outcome

• 2,432 GPUs (+128 deploying) across 38.4 MW with fleet PUE 1.11
• +10 more outcomes

Named customers10 records

Segments6 records

Ideal customer profiles5 records

Technology focussed Yes

API detail

Has API

Yes

API docs

API detail

Core technology

AI maturity

App detail

Integration8 records

AI capability11 records

Feature7 records

Partnerships

Ten partnerships are on record, tiered flagship, core and minor.

• SoftBank Group Corp.flagshipStrategic or Co-development Partner · 1 June 2026Majority-owned joint venture between SoftBank Group and Sesterce to develop and operate a 1 GW AI data center campus in Bosquel, Hauts-de-France, France, announced May 30, 2026. The campus will create 400 long-term skilled jobs, establish a €10 million endowment fund with the Hauts-de-France region, and serve major European markets with advanced AI workloads at low latencies. Combines SoftBank's global AI infrastructure platform with Sesterce's data center capabilities; positions France as a major hub for AI infrastructure in Europe as part of SoftBank's 5 GW commitment from the 2026 Choose France summit.
• PyTorchcoreStrategic or Co-development Partner · 9 April 2026Co-organizer of the 'No Gradient Descent, Only Ascent' ML systems hackathon in Paris on April 9, 2026, held as a PyTorch Conference Europe side event. Will Feng from PyTorch's Helion team delivered a technical talk on high-level DSLs for kernel authoring.
• SemiAnalysiscoreStrategic or Co-development Partner · 9 April 2026Co-organizer of the 'No Gradient Descent, Only Ascent' hackathon in Paris. Jeremie Eliahou Ontiveros from SemiAnalysis participated in the 'Power is the new compute' fireside chat with Sesterce CEO Youssef El Manssouri.
• GPU ModecoreStrategic or Co-development Partner · 9 April 2026Co-organizer of the 'No Gradient Descent, Only Ascent' ML systems hackathon in Paris.
• VerdaminorStrategic or Co-development Partner · 9 April 2026Co-organizer of the 'No Gradient Descent, Only Ascent' ML systems hackathon in Paris, bringing community participants to push frontier AI systems performance.
• Prime IntellectminorStrategic or Co-development Partner · 9 April 2026Co-organizer of the 'No Gradient Descent, Only Ascent' ML systems hackathon in Paris. Matej Sirovatka from Prime Intellect delivered a technical talk on reinforcement learning at scale.
• EquinixcoreChannel Partner/ Reseller/ DistributorHosts the Sesterce Cloud Paris site in Equinix's Paris data center facility. Equinix is featured prominently on the Sesterce site as a Paris site partner.
• NVIDIAcoreTechnology or IntegrationListed as a 'Trusted by' partner on the Sesterce homepage. Sesterce deploys the latest NVIDIA silicon (GB300, B200, H200) in its AI factories; uses NVIDIA HGX/DGX-grade platforms, InfiniBand fabric, and DCGMI telemetry.
• PatriarcheminorChannel Partner/ Reseller/ DistributorListed as a 'Trusted by' partner and customer on Sesterce's homepage; President/Architect Damien Patriarche provided a video testimonial stating 'Sesterce builds infrastructure ready for long-term AI evolution.'
• Flex AIminorChannel Partner/ Reseller/ DistributorListed as a customer on Sesterce's homepage; EMEA Sales Director Sebastien Urios provided a video testimonial stating 'Sesterce gives us European capacity for critical AI workloads.'

Scale indicators13 records

Recent moves7 records

Expansion highlights6 records

Direct peers

• CoreWeave: US-based hyperscale GPU cloud built on NVIDIA silicon (H100/H200/GB200), serving frontier AI labs and enterprise with on-demand, reserved, and bare-metal offerings. Closest direct competitor to Sesterce Cloud on product shape (GPU cloud + dedicated clusters) but without Sesterce's European sovereign positioning or owned energy/facility stack.
• OVHcloud: Largest European cloud provider with explicit sovereign-cloud positioning, data centers across France and the EU, and a GPU cloud product (H100/A100 instances). Directly comparable as an EU-resident alternative for regulated buyers — the incumbent Sesterce must displace.
• Scaleway: French/European cloud provider (subsidiary of Iliad) offering GPU instances (H100, L40S) with strong sovereignty and environmental positioning. Competes with Sesterce Cloud for the same French and EU regulated-customer base, though without AI-factory-scale buildouts.
• Lambda Labs: US GPU cloud and cluster provider offering on-demand NVIDIA H100/H200/B200/B300 instances and reserved capacity to AI labs and enterprises. Direct peer to Sesterce Cloud on product and pricing model, though without EU sovereign positioning.
• Crusoe Energy: AI-focused data center operator that pairs GPU compute with stranded/renewable energy and has built large-scale clusters for AI training. Comparable to Sesterce on the energy-first, vertically integrated philosophy, though US-based and not focused on EU sovereignty.

Emerging players

• Flex AI: AI infrastructure startup providing European AI compute capacity, listed as a Sesterce customer/testimonial partner. Operates in the same EU sovereign-AI space as an emerging challenger that could also become a channel or partner.
• Verda: GPU cloud provider offering NVIDIA H100/B200 clusters for AI training and inference, co-organizer of Sesterce's ML systems hackathon in Paris. Direct overlap on GPU cloud product, with smaller community- and Europe-leaning footprint.

Broad incumbents

• Equinix: Global colocation and interconnection incumbent that hosts Sesterce's Paris site and is also an indirect competitor for sovereign workloads. Comparable as a data-center operator in EU regulated markets, though it does not offer GPU cloud or AI-factory services natively.

Regional players

• Gcore: European-headquartered cloud and edge provider with GPU instances and AI inference services across global PoPs. Comparable as a European sovereign-leaning GPU/AI cloud, primarily competing in EMEA rather than Sesterce's French-AI-factory niche.
• Hetzner: German-headquartered hosting and cloud provider with strong European footprint, recently adding GPU servers. Competes for price-sensitive European compute buyers but lacks the sovereign-AI-factory positioning and NVIDIA Blackwell focus.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat6 records

Key risks6 records

Key highlights7 records

Customer concentration

• Xx.com/sestercegroup

Compliance3 records

Revenue estimate

Valuation estimate

Number of profiles

Profiles4 records

Subsidiaries1 record

Funding overview

Funding rounds

Investors

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments1 record

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Sesterce