🚨 Before we dive into this month's updates, we've got something exciting to share:
On April 9, we’re hosting a hackathon alongside PyTorch Conference Europe in Paris and you’re invited.
Build, experiment, and push your ideas to the limit for a chance to win 48 hours on NVIDIA GB300 NVL72, one of the most powerful AI systems available today.
March was a strong month for Verda as we continued to build across platform, infrastructure, and ecosystem, moving closer to our goal of becoming Europe's AI cloud. We introduced new platform capabilities to improve developer workflows, expanded our presence in Europe, deepened integrations across the AI ecosystem, and connected with the community through key events and collaborations
Platform updates
Container Registry
We are rolling out our container registry in beta, an early release for users who want to test and provide feedback ahead of general availability.
During the beta period, all users receive 100GB of free storage, allowing teams to experiment, validate workflows, and help us refine the product before wider release.
Get started today and help us shape the product
Status page
We have launched a dedicated Status page to give users real-time insight into the health of the Verda platform. You can track incidents, monitor ongoing issues, and stay informed about scheduled maintenance in one place.
Bank transfer available
We have enabled bank transfer as a payment method for all users. Transfers are currently supported in EUR only and receipts are automatically generated once funds are received, similar to card payments, so no invoice request is needed.
If you would like to switch your account currency to EUR, you can contact our team for assistance. When making a transfer, please ensure the recipient name is set to Verda Cloud Oy.
Company News
New AI data center
Last month, we announced our plans to establish a new AI data center in Sundsvall, Sweden. We have now shared more details on the project and our vision for building next-generation AI infrastructure in the region.
The Sundsvall site will combine high-performance compute, renewable energy, and strong local partnerships to support large-scale AI training and inference. It also aims to contribute to a broader ecosystem of research, innovation, and talent development in the Nordics.
This expansion is a key step toward building sovereign AI infrastructure in Europe and meeting the rapidly growing demand for advanced AI compute.
London office
We are expanding our footprint with a new office in London, strengthening our presence in the UK and across Europe. This move allows us to work more closely with customers and partners while continuing to build and scale sovereign AI infrastructure in the region.
More official updates will follow soon.
Ecosystem
SkyPilot Integration
Verda is now supported in SkyPilot, making it easier for teams to run and manage workloads on Verda using SkyPilot’s multi-cloud orchestration workflows. This integration gives users more flexibility in how they provision and scale AI infrastructure across environments.
Efficient LLM training with FP4
Researchers at the Institute of Science and Technology Austria demonstrated that a ~4B parameter model trained for under $10K can approach the performance of GPT-3. The work was powered by Verda GPUs, showcasing how high-quality data, thoughtful engineering, and native FP4 precision can dramatically improve training efficiency.
Enabled by Blackwell GPUs, massive FP4 throughput, and increased HBM capacity, this research highlights how full-stack optimizations are translating into real-world gains. Verda is proud to support teams like ISTA pushing the boundaries of efficient and accessible AI.
Job Openings
We’re growing fast and expanding our team. If you’re passionate about AI infrastructure and want to help build Europe’s next hyperscaler, we’d love to hear from you.
Explore our open roles and join us on the journey to shape the future of cloud computing:
- Senior/Principal Site Reliability Engineer (EU)
- Global Data Center Expansion & Operations Manager
Blog
Understanding NVFP4 on NVIDIA Blackwell
Our latest post explores NVFP4, a 4-bit floating point format introduced with NVIDIA Blackwell GPUs, and why it marks a major step forward for low-precision AI. NVFP4 combines microscaling techniques with hardware-level support to overcome the traditional limitations of ultra-low precision, making it possible to run large-scale models more efficiently.
By reducing memory usage and improving throughput while maintaining accuracy, NVFP4 enables more efficient training and inference pipelines, especially for large language models and frontier AI workloads.
Events
NVIDIA GTC San Jose 2026
We joined NVIDIA GTC San Jose last month. The event highlighted rapid progress across the latest GPU systems, low-precision formats like FP4, and the growing importance of infrastructure designed for large-scale AI workloads.
It was a valuable opportunity to connect with researchers, engineers, and partners across the ecosystem, and to see how quickly the industry is moving toward more efficient, scalable AI. We are excited to bring these insights back into what we are building at Verda.
HelsinkiJS March meetup
We recently took part in the HelsinkiJS March meetup, hosted at Verda, where we gave a short talk and connected with developers across the local ecosystem.
Events like these are a great way to stay close to the community, share what we are building, and exchange ideas with engineers working across different parts of the stack. We are excited to continue supporting and engaging with the developer community in Helsinki.
PyTorch Paris and Hackathon
On April 7-9, we’re going all in at PyTorch Conference Europe in Paris with a Verda booth, a hackathon, and our entire AI team on-site.
If you’re building, training, or scaling AI systems, this is where you want to be. Come meet the team, dive into real workloads, and explore what’s possible with next-generation AI infrastructure.
We’re excited to connect, exchange ideas, and talk about what’s next for AI at scale.
Communities
Developer Community
Our developer community is a space where AI builders share benchmarks, workflow tips, and real-world lessons from training and inference at scale. It is the best place to ask questions, learn from other users, and stay close to what we are shipping next.
Join the conversation on forum.verda.com and build with the community.
Discord
You can also join our Discord, where the Verda community connects in real time. It is a great place to get quick help, exchange ideas, share experiments, and keep up with new features as they launch. Come hang out with AI builders from around the world and be part of the conversation.