OVHcloud headquarters and data centre campus in Roubaix, France

OVHcloud Unveils 55 AI and Cloud Products to Challenge US Providers on Home Turf

OVHcloud, the French cloud provider, unveiled 55 new cloud and AI products at its AI Summit in Roubaix on 19 November 2026. The launch spans infrastructure, inference and workplace software, all hosted in Europe.

The centrepiece is compute. OVHcloud is adding Nvidia RTX 6000 Pro, Blackwell B300 and Rubin R200 GPUs to its bare metal and public cloud services. Its data centres can supply up to 60,000 B300 GPUs or 35,000 R200 GPUs.

More than raw compute

The launch also rebuilds the network layer for AI workloads in production. The virtual private cloud stack gains a web application firewall, a VPN, VPC peering, endpoints and a global load balancer. Customers can manage their own public IP routing through BGP Service and keep their own IP ranges with BYOIP.

OVHcloud Connect, the private network service, now reaches 40 points of presence worldwide. Object storage gets faster with no price rise, and egress traffic stays included in the price.

Agents, search and voice

For builders, AI Fabric creates Python sandboxes for agents in partnership with Blade.sh. Serverless containers, functions and messaging cover applications that run without managed servers. AI Endpoints adds around 100 new models, including voice and image models, plus governance tools.

Speech arrives through Gladia, with transcription and voice generation in more than 100 languages, in real time and on recordings. AI Agents hosts, scales, secures and monitors agents connected to enterprise tools through nearly 480 connectors and MCP servers.

A WebAssembly sandbox spins up in two milliseconds, and the agent only costs money while it is working. AI Search lets businesses explore documents by combining semantic and lexical search across hundreds of millions of documents per collection.

A workplace platform with AI built in

OVHcloud also announced OVHai Workspace, an agentic platform for businesses. It brings together email, calendar, document editing, storage and video conferencing around a single AI, plus tools to create websites and search company data.

Octave Klaba, chief executive of OVHcloud
Octave Klaba, chief executive of OVHcloud. Photo: Wikimedia Commons

The alpha version is available free of charge from today, with 5 GB of storage per user. The platform will also be offered as a private tenant for a single customer and as an on-premises appliance with integrated GPUs.

Why Europe is the pitch

The launch is framed around sovereignty. OVHcloud says its services run in Europe, North America and Asia, and all of them are available in public cloud, single-tenant private cloud and on-premises form, including with SecNumCloud qualification.

Three OVHai research labs back the roadmap. One works on large language models, one on voice, and one on document search enriched with ontology data.

“With a single provider that masters the entire cloud and artificial intelligence chain, we offer an integrated user experience with a European offering that is open, reversible and predictable.”

Yaniv Fdida, CPTO of OVHcloud

Sovereign AI partnerships

OVHcloud also announced a partnership with Qwant, Shadow and EUSP, a joint venture between Qwant and Ecosia that builds a European search index. Together they connect the web index, models, infrastructure and use cases into sovereign AI hosted and operated in Europe.

AI-generated summaries at the top of Qwant search results now run on Olala mini 7A1B, OVHcloud’s summarisation model, executed on Blade, Shadow’s GPU platform. OVHcloud announced Olala, its first family of language models, on 17 November, with the mini model offered as open source under the Apache 2.0 licence.

The move comes as European governments and companies look more closely at where AI services run and how data is handled. OVHcloud serves 1.7 million customers in more than 140 countries from 46 data centres, and it has operated for more than 25 years.

Leave a Comment

Professor Derpy's Notes

Sixty thousand GPUs sounds like a lot until you realise I would use them all to generate one cat picture. I stand by the maths.

Join our newsletter

email subscription

Receive Latest AI Insights To Your Inbox