OVHcloud’s AI pivot proves Europe can compete—but not by hardware alone

What does it take for Europe to build a credible alternative to the American cloud giants? Not another promise of sovereignty, and certainly not a warehouse full of GPUs with no clear customer attached. OVHcloud’s latest results suggest the answer could be a disciplined combination of infrastructure, software and specialized services—but the company’s AI pivot is about to test every part of that formula.

The numbers are encouraging. For FY2026, OVHcloud reported revenue of €1.145 billion, adjusted EBITDA of €480.5 million and a 42% adjusted EBITDA margin. Levered free cash flow turned positive for the first time since its IPO, reaching €24.7 million, while capital expenditure rose to €372.9 million. The company’s results, reported by ABC Bourse, pushed its shares up 9.72% on October 20, 2026.

That is a meaningful financial reset. But it is not yet proof that AI is profitable for OVHcloud. The more important question is whether its existing assets can become differentiated products rather than simply a European-flavored version of rented compute.

OVHcloud’s advantage is industrial discipline, not just European branding

OVHcloud has a stronger starting position than many sovereignty narratives imply. The company says it owns the majority of its data centers, designs its servers, operates its own network and runs two server-production sites. It also says it has 3,200 employees and more than one million customers and partners. Those capabilities matter because cloud economics are unforgiving: owning more of the stack can improve control over costs, supply and deployment speed.

The financial results suggest that this model is beginning to work. The adjusted EBITDA margin is up, cash flow is positive and the company says its 2027 target is like-for-like revenue growth above 10%, an adjusted EBITDA margin above 42.5% and continued positive levered free cash flow. Those targets are forecasts, not guarantees, but they show a company trying to make growth answer to capital discipline rather than treating infrastructure spending as an achievement in itself.

That distinction is crucial for Europe. A sovereign cloud cannot merely promise that data stays within a preferred jurisdiction. It must also be competitive, reliable and broad enough to avoid forcing customers back to a hyperscaler for the next stage of their workload. OVHcloud’s portfolio now spans bare metal, public cloud, SecNumCloud-qualified services and an on-premises cloud platform. The company also says it has developed more than 80 Public Cloud products over five years and is now integrating and regionalizing them.

In our view, that product layer is more important than the rhetoric around independence. Infrastructure is the foundation. The customer buys the outcome.

AI will reward the companies that sell workflows, not GPU hours

OVHcloud already offers cloud GPU instances for training, inference, machine learning, video and graphics workloads. Its catalog includes NVIDIA A10, H100, L4 and L40S options, among others, with pricing described as transparent and predictable. That gives the company a credible entry point, but compute capacity is increasingly a difficult place to stand out.

The better move is to climb the stack. OVHcloud says it is developing OVHai Workspace for businesses, plans to use AI to assist customer support and organizes small multidisciplinary teams around new projects. Its stated internal objective is for each initiative to reach €1 million in monthly revenue as quickly as possible. That is a revealing ambition: it treats AI not as a branding exercise, but as a portfolio of products that must earn their keep.

The company’s acquisitions point in the same direction. It completed the acquisition of Dragon LLM on March 25, 2026, creating an internal laboratory for training and fine-tuning sovereign large language models. It also acquired Seald for €2 million, adding end-to-end encryption for data in transit and during processing. These moves could help OVHcloud sell something more valuable than raw acceleration: controlled, specialized AI deployments for organizations that cannot simply send sensitive data to a general-purpose service.

That is where the European opportunity looks most credible. Government, defense and regulated industries may value data location, reversibility, encryption and deployment flexibility enough to choose a provider on more than price. OVHcloud says it is building a defense vertical around SecNumCloud-certified products, with use cases including AI-assisted command, drone orchestration and communications interoperability. Those are company-reported plans, not evidence that the contracts or revenues already exist.

Sovereignty does not eliminate dependence; it changes where the dependency sits

Here is the uncomfortable part: OVHcloud’s independence has limits, especially in AI. Its GPU offering is built around NVIDIA hardware, and the company’s own results mention advance purchasing to manage inflation in memory and disk components. Capital expenditure reached €372.9 million for FY2026. Even with server design, data centers and liquid-cooling expertise in-house, the company remains exposed to a supply chain it does not control completely.

This does not invalidate the sovereignty proposition. It makes the proposition more precise. Europe may not need to manufacture every component to create a meaningful alternative. It does need to control enough of the service layer—where data runs, how models are deployed, how customers can move them and who supports the workload—to prevent infrastructure dependence from becoming total platform dependence.

OVHcloud’s emphasis on open standards, open-source technologies and open-weight models is therefore strategically sensible. The company says customers can retain control over data location, deployed models and the ability to change providers. If that flexibility is real in practice, it could become a stronger differentiator than simply offering another GPU instance.

But the market should resist treating every sovereignty claim as a moat. Gartner positioned OVHcloud as a Challenger in its July 2026 Magic Quadrant for cloud AI infrastructure, which is recognition of progress, not a verdict that the company has matched the largest providers. OVHcloud still has to prove that its services are easier to buy, integrate and operate for the customers it wants to win.

The next phase should be judged by recurring AI revenue

The stock’s reaction shows that investors welcomed the improved accounts, but even favorable analysts are asking for more detail on the AI strategy. That is the right standard. The company has earned credit for returning to positive cash flow and maintaining a high margin while investing. It has not yet earned a blank check for AI expansion.

The winning version of OVHcloud is not a smaller hyperscaler chasing every benchmark. It is a European infrastructure company that turns sovereignty into practical advantages: predictable bills, secure deployment, model portability, specialized support and services that solve identifiable industry problems. Its own plan to deploy the AI strategy in 2027 makes the next year particularly important.

Our verdict is cautiously positive. OVHcloud is one of the more credible tests of whether Europe can build cloud independence without abandoning commercial discipline. But customers and investors should watch product adoption, AI-related recurring revenue and returns on the new capacity—not GPU inventories or ambitious language about the future. If OVHcloud can make its infrastructure useful at the workflow level, AI may strengthen its independence. If not, it will simply make the company a well-run tenant in someone else’s hardware ecosystem.