VERDA raises €165 million in a bid to become Europe’s AI hyperscaler

Formerly known as DataCrunch, Verda has announced around €165 million in new financing to accelerate the development of its AI cloud, expand its computing capacity and invest in inference. Having reached an annualised revenue run rate of approximately €144 million in July, the company is gradually moving beyond GPU rental and into the race to become a hyperscaler. In Europe, that ambition puts it up against Nebius and Nscale. In the United States, CoreWeave already illustrates the scale of capital, power and infrastructure that such a trajectory demands.

Verda has announced approximately €165 million in new financing, in a transaction that includes a Series B led by Emergence Capital, with participation from MUFG Innovation Partners, Supermicro, Varma Mutual Pension Insurance Company, Lifeline Ventures, 6 Degrees Capital, byFounders and Tesi. The company says it has now raised more than €390 million since its inception, reaching a valuation approaching €875 million.

Verda also reports an annualised revenue run rate of around €144 million in July, up from approximately €87 million just a few weeks earlier.

From DataCrunch to a full-stack AI cloud

Founded in Helsinki in 2020, DataCrunch initially built its business around a straightforward proposition: making powerful GPUs available on demand, without requiring developers and businesses to build their own infrastructure. The shortage of Nvidia accelerators and the surge in demand driven by generative AI made this model particularly attractive. But that advantage is becoming less sufficient as the market matures.

The transition from DataCrunch to Verda reflects this shift. The company now wants to control more layers of the technology stack, from physical infrastructure and servers to networking, orchestration, the cloud platform and services for teams training or running their models. Verda no longer wants merely to sell GPU hours. It wants to become the environment in which AI applications are developed, deployed and brought into production.

This shift brings its model closer to that of the established hyperscalers, with one major difference: AI cloud services rely on far more capital-intensive infrastructure. Each stage of growth requires securing additional GPUs and adding servers, networking, cooling systems, buildings and, above all, electricity. The more the business grows, the more financing it needs, giving this new generation of cloud providers a very different profile from traditional software companies.

Ruben Bryon: from a Helsinki garage to hyperscaler ambitions

This trajectory remains closely tied to Ruben Bryon, Verda’s founder and CEO. In 2020, he launched an initial self-service GPU cloud from his garage in Helsinki, before being joined by Miłosz and Tamir. Within weeks, the business was cash-flow positive. Bryon, who already had experience in cloud software and GPU virtualisation, says he created DataCrunch in response to the shortcomings he saw among traditional hyperscalers: high prices, complexity and an experience insufficiently tailored to AI developers.

That starting point partly distinguishes Verda from its main European competitors. Nebius emerged from Yandex’s former international operations, while Nscale was built more directly around large-scale AI infrastructure projects. Verda retains more of the DNA of a developer-led startup that has gradually moved up the value chain, from basic GPU access to cloud services, orchestration and AI services. The rebranding from DataCrunch to Verda in 2025 was intended to support precisely this ambition: to be seen not simply as a compute provider, but as a future European hyperscaler.

Bryon now takes that ambition further, presenting AI infrastructure as the starting point for a project to build what he describes as “the first true tech company in Europe.”

But moving up the technology stack comes with an immediate trade-off: the broader Verda’s scope becomes, the more capital it must tie up in servers, GPUs, data centres and energy.

Growth that consumes ever more capital

Supermicro’s investment illustrates this new reality. The US manufacturer is a central player in the AI server supply chain, and the ability to secure equipment incorporating the latest GPU generations quickly is becoming a decisive operational advantage. In this market, financing alone is no longer enough. Companies must also secure suppliers, manufacturing capacity and the megawatts needed to operate their infrastructure.

This dynamic also explains why neoclouds are raising capital at such a rapid pace. The more demand Verda converts into revenue, the more it must reinvest to make that capacity available. Commercial growth almost immediately creates a need for further investment.

Inference as the next growth driver

One priority for the new capital will be inference, which has become a major growth driver for the market. Following an initial phase largely driven by the training of large models, another market is developing around their everyday use. Assistants, agents, image and video generation, business applications and development tools are multiplying production computing requirements, gradually turning AI compute into permanent infrastructure.

For Verda, this shift could also strengthen its business model. Inference opens the door to more integrated services spanning deployment, orchestration, serverless computing, storage and workload optimisation. Differentiation is therefore gradually moving away from the number of available GPUs towards the software layer that makes them easy to use, high-performing and cost-efficient.

Nebius and Nscale have already reached a different scale

The challenge is that Verda is not alone in pursuing this trajectory. Across Europe, several companies are already building large-scale AI clouds, albeit with different approaches.

Nebius currently stands out as the most advanced benchmark on the platform side. The group generated approximately €508 million in revenue in the second quarter of 2026, while its annualised revenue run rate reached around €2.6 billion at the end of June. At that scale, the company is no longer simply operating GPU infrastructure: it is seeking to provide a complete platform for developers and enterprise customers.

Nscale is following a more industrial approach. The British company raised approximately €1.75 billion in a Series C in March 2026, at a valuation approaching €12.7 billion, and is seeking to integrate more components across data centres, networking, energy and software. Its model nevertheless gives a clear indication of the cost of this strategy: in the first half of 2026, Nscale generated approximately €123 million in revenue, while recording a net loss of nearly €900 million.

Verda occupies a different position. Its origins in self-service GPU cloud give it a stronger focus on developers and on-demand consumption. But its ambition is now taking it onto the same ground: building enough of the technology stack to move beyond commoditised compute and become a platform.

These companies are not pursuing exactly the same model. But they are responding to the same question: can Europe build a meaningful share of its AI infrastructure without relying exclusively on US hyperscalers?

CoreWeave reveals the scale of the gap with the United States

The US comparison nevertheless shows how much ground remains to be covered. CoreWeave generated approximately €2.25 billion in revenue in just three months during the second quarter of 2026, with a revenue backlog approaching €91 billion. The company expects capital expenditure of between €30.5 billion and €34 billion in 2026 alone—amounts that move the sector beyond venture capital and into industrial financing.

At this level, building an AI cloud means financing tens of thousands of GPUs, data centres, cooling equipment, grid connections and ultra-high-speed networks. The business sits at the intersection of technology, telecommunications, infrastructure and energy. That is precisely the dimension that could become the main obstacle for European players.

In Finland, the advantage is also measured in megawatts

Verda’s Finnish base nevertheless gives it an advantage that could become increasingly important. The Nordic countries offer temperatures conducive to cooling, greater land availability and largely low-carbon electricity generation. In a sector where the main bottleneck is gradually shifting from GPUs to megawatts, geography can become a genuine industrial asset.

European sovereignty remains partial

Geography does not, however, resolve the issue of technological sovereignty. Verda can operate European infrastructure, host data on the continent and offer a local alternative to AWS, Microsoft Azure or Google Cloud. But its stack remains heavily dependent on US suppliers. Nvidia provides most of the GPUs, Supermicro operates at the server layer, and Emergence Capital is leading the current financing. Much of the same dependency exists among its European competitors.

Sovereignty must therefore be assessed at several levels: control over data, control over operations and control over the underlying technologies. European companies can make substantial progress on the first two, while the third remains largely dominated by US semiconductor and infrastructure suppliers.

Beyond equity: infrastructure financing

Verda’s next step will also be financial. As its cloud grows, equity funding alone will no longer be sufficient to meet its infrastructure needs. The sector’s largest companies are beginning to use their GPUs, data centres and multiyear contracts as assets against which they can raise debt. Verda will need to follow the same path to move from a few hundred million euros of financed capacity to several billion.

This shift towards infrastructure financing is already visible elsewhere in Europe, notably in Mistral AI’s use of debt to finance its data-centre infrastructure.

This is where the nature of the project truly changes. A future European hyperscaler will need to master software, hardware, energy and infrastructure financing simultaneously. Simply owning GPUs will not provide a durable advantage. As supply expands, differentiation will have to come from orchestration, performance, workload execution costs, inference, data location and the developer experience.

The real moat will sit above the GPUs

Verda now has the capital to accelerate. But its next challenge will be less about proving that demand exists for a European AI cloud than demonstrating that it can sustain the investment pace this industry imposes. As GPU capacity becomes increasingly commoditised, access to hardware alone will be less sufficient to build a defensible position. Value will shift towards the platform, execution quality and the ability to finance expansion without undermining the economics of the business.