In a significant strategic move, OVHcloud, a prominent European cloud services provider, is embarking on an ambitious journey to develop cutting-edge artificial intelligence models. This initiative positions the company to become a formidable challenger in Europe's large language model (LLM) arena, aiming to stand alongside current leader Mistral.

OVHcloud's AI Ambition: Reshaping Europe's LLM Landscape

On a recent Wednesday, June 17, 2026, during the VivaTech conference in Paris, OVHcloud CEO Octave Klaba unveiled the company's plans. Klaba articulated that OVHcloud intends to train "frontier AI models" – sophisticated, large-scale systems meticulously constructed from extensive data and substantial computational resources. This direction signifies a pivotal change for the company, as European nations and enterprises increasingly seek independent alternatives to artificial intelligence systems predominantly offered by American and Chinese tech giants. This urgency has been amplified by recent disruptions, such as the unexpected discontinuation of Anthropic's premium models.

Klaba emphasized the critical importance of mastering this advanced technology for OVHcloud's future viability. He pointed out that the economic landscape for developing such sophisticated models has dramatically evolved. Significant progress in chip technology, training methodologies, and synthetic data generation has substantially reduced development costs. What might have previously demanded an investment of approximately 1 billion euros (around 1.2 billion U.S. dollars) can now potentially be achieved with an estimated 150 million to 200 million euros.

The CEO characterized the current phase as a "second wave" in the industry, where new entrants are building upon the foundational work established by pioneers like OpenAI, Anthropic, and Mistral. Klaba reassured that OVHcloud's models would be trained without utilizing client data. The company plans to release a suite of models, rather than a singular system, recognizing that different models excel in specific applications. As Klaba noted, "There's no one model that does all the magic alone." He cited DragonLLM, a recently acquired startup, as a key part of this strategy, revealing that a model has already undergone pre-training using Jupiter, Europe's fastest supercomputer. While optimistic, Klaba maintained that it is premature to disclose detailed performance metrics. OVHcloud also expressed its commitment to open-sourcing these models once they achieve satisfactory performance benchmarks, aligning with its long-term vision.

OVHcloud's strategic pivot into frontier AI model development highlights a growing trend among European tech companies to foster technological independence and innovation within the continent. This move not only promises to diversify the LLM landscape but also addresses the rising demand for sovereign AI solutions that prioritize data privacy and regional control. By investing in its own advanced AI capabilities and committing to an open-source approach, OVHcloud is not merely competing but actively shaping the future of artificial intelligence in Europe, potentially empowering a new era of digital autonomy and collaboration.