Verda (Finland) raises $189M in Series B

Verda became Europe’s latest unicorn today, raising $189 million in new funding. The oversubscribed Series B was led by Emergence Capital, with MUFG Innovation Partners, Supermicro, Varma Mutual Pension Insurance Company, Lifeline Ventures, 6 Degrees Capital (6DC), byFounders, Tesi (Finnish Industry Investment Ltd), and a group of angel investors including Ola Tørudbakken and Mark Saroufim participating. That brings our total funding to date to over $450 million across equity and debt. "There's a window right now to build one of the defining compute companies of this generation, and to do so from Europe. It won't be open for long." — Ruben Bryon, founder and CEO Click here to accept cookies and watch the video on this website or watch on YouTube. You can watch this video on YouTube . Building the AI Cloud of tomorrow For the last few years, access to capacity has been a procurement problem. Teams negotiated for months, took delivery of a fixed environment, and organized their workload around whatever they'd been allocated. That worked when capacity was scarce and workloads were predictable. It doesn't anymore. Training runs that once took months of advance planning now need to start on infrastructure that's already there. Production inference, where latency and availability aren't negotiable, needs capacity that shows up when the workload demands it. Agentic workloads raise the bar even more, with context that builds across many turns and demand that comes in bursts, spiking the moment a tool call returns. Deciding how much capacity to get and when, increasingly sits with a handful of engineers, not a procurement team working an annual cycle. Verda is building the full stack to meet that need. We design and build our own data centers, the infrastructure inside them, and the platform that runs on top, enabling customers to get the capacity how, when, and where they need it, while operating it at a lower carbon footprint than the industry average.This raise deepens all three parts of that stack, with more data center capacity coming online and deeper investment in the platform, the AI Lab, and the features our customers are already using. Teams already run their production AI on Verda Verda now powers AI workloads for organizations across 50+ countries, from early-stage startups to enterprises. We reached a $165 million annualized revenue run rate in July 2026, with a team of over 250 people across Helsinki, London, Taipei, and San Francisco. Our GPU clusters power Aleph Alpha’s R&D infrastructure, with engineers-in-residence collaborating on the underlying software stack. Magnific serves media generation at millions of requests per day on our infrastructure. Epsilon Health trains its radiology AI models on a dedicated Verda cluster, at native image resolution. Our AI Lab lives on the platform our customers use We build our own compilers and serving software, and do our own performance and reliability engineering. Our AI Lab is a dedicated team that runs real research workloads on the platform, working on solving hard engineering problems like improving GPU utilization, inference optimization, and kernel engineering. That work means customers get more out of their infrastructure, and it’s what shapes our product roadmap, as the Lab runs into what needs building before a customer has to ask for it. "What we didn't expect was a team of engineers who could help us design our own data streaming and cluster management, not just hand us a cluster and walk away." - Arjun Karpur, Head of Machine Learning, Epsilon Health What’s next More capacity, and the power to run it: We will have more than 250 MW of operations in 2027, with data centre capacity live in Finland today and more coming online in Europe, the UK, the US and Asia. That also means getting the latest infrastructure to our customers as it’s available. We anticipate early deployments of NVIDIA VR200 NVL72 in the coming months. Inference, at scale: The workloads customers run today already look different from a few years ago, longer sessions, context accumulating over many turns, and agentic use spiking demand. For the AI teams, the challenge isn't just serving one of those sessions well, it's serving thousands of them at once without one team's burst becoming another's problem. We are investing in providing access to models for inference, tuned for how they're used. Faster provisioning: We're focused on improving provisioning and setup times so they don’t become bottlenecks for AI workloads. That means faster spin-up for instances and clusters, quicker storage attachment, and infrastructure that’s fast to respond once it's running. Platform updates: We continue to broaden our product portfolio to offer AI teams every capability they need across the AI lifecycle.Following the recent release of Container registry, we are on path to release S3-compatible Object storage. Our Instant clusters just got Kubernetes with Kueue or Slurm via Slinky pre-configured at deployment, which we will follow with the long-awaited release of Managed Kubernetes. Enterprise readiness: As AI moves from pilots to production, enterprises need to trust it the way they trust anything else running their business. This includes who has access to it, how it’s protected, and whether that holds up to an audit. We are building for all three, continuously improving and shipping new capabilities. Recently, we added Audit Logs and SSO for IAM, so enterprises can see who did what and control who can log in. We enhanced our Confidential computing offering with the industry-first support of 8× NVIDIA HGX™ B300 and B200, to keep data protected even while it’s being used. And now offer SOC Type II and C5 certifications alongside our existing ones. Engineering depth: We’re funding the AI Lab to work closely with other AI Labs, open-source projects, and developer tool companies, and to go deeper on co-research with the teams already building on us. That’s how we understand what compute gets used for, the model architectures and the optimization techniques shaping real workloads today. That understanding feeds into decisions at every level from data center design and hardware selection to provisioning and software. Developer experience: None of this matters if the people building on Verda can’t work the way they already do. That’s why, as we ship new features and capabilities, we build for every way a team works, whether that’s through our console, the CLI or via API and everything in between. Many of these capabilities are already available in the Verda cloud platform: Try them out in the UI by logging in Or learn more about our CLI, API, and other provisioning methods on our docs