The OVH group has taken advantage of the celebration of its OVHcloud AI Summit to mark the beginning of its acceleration in the field of artificial intelligence. It has done so in such a way that it has announced 55 new products related to cloud and AI, ranging from infrastructure to applications, including orchestration, inference and PaaS services powered by AI.

Yaniv Fdida, CPTO at OVHcloud, said: “In Europe, most AI projects never make it beyond the prototype phase, as moving into production means having to manage multiple products, multiple vendors, fees, and data transfers. By having a single provider that dominates the entire cloud and artificial intelligence chain, we offer an integrated user experience with an open, reversible and predictable European offer.

NVIDIA Blackwell and Rubin GPUs: computing power changes scale

As the only European cloud services provider capable of offering its services in Europe, North America and Asia, the OVH group adds the NVIDIA RTX 6000 Pro, Blackwell B300 and Rubin R200 GPUs to its Bare Metal and Public Cloud offering. The capacity of its data centers allows it to supply up to 60,000 B300 GPUs or 35,000 R200 GPUs, allowing it to meet all its customers’ needs.

New VPC-based network services

An AI in production is, above all, a controlled network. The VPC stack is extended with a WAF, VPN, VPC peering, VPC endpoints, ALB, Global Load Balancer, and private DNS. With the BGP service, customers manage the advertising and routing of their public IP addresses themselves, while BYOIP allows them to maintain their own IP ranges. Customers connect to the private network through OVHcloud Connect, now available in 40 points of presence worldwide. And in terms of storage, object storage gains speed without incurring any additional cost, since outbound traffic is still included.

AI Fabric and Serverless: AI inference expands and incorporates voice with Gladia

Developed with Blade.sh, AI Fabric makes it easy to launch test environments for agents in Python. To run applications without managing servers, Public Cloud now offers serverless containers, serverless functions, as well as messaging and events. Regarding inference, AI Endpoints is incorporating around a hundred new models, including voice and image models, as well as governance tools. With Gladia, the offering is expanded to transcription and speech generation, both in real time and from recordings, in more than 100 languages.

AI Agents and AI Search to power enterprise agents

AI Agents hosts, scales, secures and monitors agents connected to enterprise tools using nearly 480 connectors and MCP servers. It only takes two milliseconds to activate a WebAssembly sandbox. The agent only generates costs when it is operational. AI Search allows companies to explore their documents by combining semantic and lexical searches across hundreds of millions of documents per collection.

Application Security with HSM and Confidential Computing

Anti-DDoS Customer Profiles adapts protection against DDoS attacks to the traffic profile of each application. Cloud HSM stores encryption keys on a specific hardware module, under the control of the client, while Confidential Computing keeps data encrypted during processing. Backup Center, Certificate Manager and PKI complete the system to protect data, identities and certificates.

3 deployment models and 3 AI research laboratories

All services are available in Public Cloud, single-tenant Private Cloud, and On-Prem, even with SecNumCloud certification. The AI ​​roadmap is supported by three OVHai labs, respectively dedicated to large-scale language models (LLM), speech, and ontology-enriched document search.