News
DealsCorvex agrees USD 103 million take-or-pay deal to perform AI inference with Mid-Atlantic GPUsSovereignNeysa scales AI infrastructure in India with AMD Instinct MI350X GPUs on TytanInferenceMicrosoft Foundry adds Baseten, DeepSeek AI inference (MSFT:NASDAQ)ClustersDataOne and Rowan University sign engineering co-op program agreement for DataOne's Vineland AI factoryClustersLook inside Cal Poly's $3 million AI Factory (Video)DealsOpenAI Taps AMD Chips for 457 MW of AI Capacity in Six-Year TensorWave DealClustersCal Poly launches an AI data center on campus. Take a look insideInferenceArista takes aim at the next AI network battleground: distributed inferenceDealsOpenAI Signs 457 MW AI Data Center Deal as AMD Expands Its Challenge to NvidiaClustersBehind Dell AI Factory’s NVIDIA & AMD Enterprise ExpansionDealsCorvex agrees USD 103 million take-or-pay deal to perform AI inference with Mid-Atlantic GPUsSovereignNeysa scales AI infrastructure in India with AMD Instinct MI350X GPUs on TytanInferenceMicrosoft Foundry adds Baseten, DeepSeek AI inference (MSFT:NASDAQ)ClustersDataOne and Rowan University sign engineering co-op program agreement for DataOne's Vineland AI factoryClustersLook inside Cal Poly's $3 million AI Factory (Video)DealsOpenAI Taps AMD Chips for 457 MW of AI Capacity in Six-Year TensorWave DealClustersCal Poly launches an AI data center on campus. Take a look insideInferenceArista takes aim at the next AI network battleground: distributed inferenceDealsOpenAI Signs 457 MW AI Data Center Deal as AMD Expands Its Challenge to NvidiaClustersBehind Dell AI Factory’s NVIDIA & AMD Enterprise Expansion
EveryNeocloud
Verda

Verda

European full-stack AI cloud offering GPU clusters, instances and model APIs

Verda, previously named DataCrunch, is a European AI cloud company that rents out GPU clusters, virtual machines and storage, and serves open-weight models through one API. The company says it began as a self-serve GPU cloud started by Ruben in Helsinki, and has since opened offices in London and San Francisco.

Share on WhatsApp

  • Began as a self-serve GPU cloud run from a Helsinki garage
  • Renamed from DataCrunch while raising a $64M Series A
  • Holds NVIDIA Preferred Partner status
  • Raised $13M seed, $64M Series A and a further $189M
  • Serves Kimi K3.1 Pro and DeepSeek V4.1 Flash through one API

What it is

Verda is a European company that sells computing power for AI work. Customers can hire whole GPU clusters linked with InfiniBand, a fast network normally used inside supercomputers, or single GPU virtual machines, or run containers that scale up and down on their own. Storage comes as NVMe block disks, a shared POSIX filesystem, S3-style object buckets and an image registry. The firm also runs an inference service, Verda Inference, which serves open-weight models through one API that speaks the same protocol as OpenAI's. The company describes itself as built in Europe and used in more than 50 markets, with an in-house AI Lab doing research alongside the platform.

How it came about

According to the company, Ruben began offering self-service GPU capacity from a garage in Helsinki and was soon joined by Miłosz and Tamir. It says the business covered its own costs within weeks. A $13M seed round followed, funding the AI Lab and platform work. The firm then changed its name from DataCrunch to Verda, raised $64M in a Series A and became an NVIDIA Preferred Partner. Offices opened in London and San Francisco, headcount passed 200, and a further $189M was raised to add compute capacity.

The compute

Hardware on offer spans NVIDIA B300 and B200 SXM6 boards, RTX PRO 6000, H200 and H100 SXM5, and A100 SXM4, each paired with a stated number of CPU cores and memory. Nodes scale from a single tray up to two racks or more, joined by NVLink v5. The network layer uses InfiniBand and RoCE. Storage covers NVMe disks, a shared filesystem, object storage in beta, and container images. Serverless containers handle inference and batch work, and confidential computing gives hardware-attested inference and fine-tuning.

Where the GPUs are

The company states that its locations are in the EU and that it owns its stack end to end, from data centres through to managed services. It reports running on 100% renewable energy. Beyond the EU footprint it names offices in London and San Francisco. Specific data centre sites, power figures and hosting partners are not given in what the company publishes.

Customers and deals

Epsilon Health, described as an AI-native radiology practice, was given a dedicated NVIDIA Blackwell Ultra B300 cluster with managed support to train on full-resolution medical images. Verda says it works with 1X on multi-GPU inference for the 1XWM generative video model. It sponsors the SGLang project with compute and infrastructure, and gave vLLM early use of a GB300 cluster. Moonshot AI is a partner for the Kimi K3.1 Pro model served on the platform.

Money and ownership

Funding disclosed by the company totals $13M in seed capital, $64M in a Series A, and another $189M raised afterwards. It reports reaching a $100M revenue run rate and says the business was cashflow positive within weeks of starting. The company is not listed on any exchange.

Its place in AI

Verda positions itself as a European provider with EU ownership and EU locations, and says it is trusted across more than 50 markets for data protection and digital sovereignty. It holds NVIDIA Preferred Partner status. Its AI Lab contributes to open-source projects, and the firm reports work with SGLang on reinforcement learning training using FP8 and INT4 on Hopper and Blackwell, plus vLLM results on a GB300 cluster.

Questions and criticism

The company publishes security certifications including SOC 2 Type II and ISO 27001, 27017, 27018 and 27701, and states GDPR compliance. It reports historical uptime above 99.9% and, elsewhere on its site, above 99.95%. No figures for debt, energy draw or export-control exposure appear in the company's own pages.

Questions people ask about Verda

What does Verda sell?

It rents GPU clusters with InfiniBand links, GPU and CPU virtual machines, and auto-scaling containers. Storage covers NVMe disks, a shared filesystem and object buckets. A separate service, Verda Inference, serves open-weight models through an API that works with OpenAI client code.

Where did the company come from?

The firm says Ruben started a self-service GPU cloud from a garage in Helsinki, with Miłosz and Tamir joining soon after. It was called DataCrunch before taking the name Verda. The company reports being cashflow positive within weeks of launch.

Which chips can customers use?

The published line-up covers NVIDIA B300 and B200 SXM6, RTX PRO 6000, H200 and H100 SXM5, and A100 SXM4. Each configuration lists CPU cores, system RAM and GPU memory. Nodes can be joined from one tray up to two racks using NVLink v5.

How much funding has it taken?

The company reports a $13M seed round, a $64M Series A, and an extra $189M raised later. It also states a $100M revenue run rate and a team of more than 200 people. Investors behind those rounds are not named on its own pages.

What does the AI Lab do?

The in-house AI Lab works on research and open-source projects, and turns that work into platform features. Areas listed include customer co-research, large-scale training, reinforcement learning frameworks, inference optimisation, AI compilers and kernel engineering.

Good reading and sources

    Where this page comes from

    There is no English Wikipedia article on Verda yet, so the facts on this page come from the sources listed above: verda.com/company, verda.com/. Some are the subject's own pages, so its claims about itself are reported as its claims. EveryNeocloud wrote the text in its own words, with AI assistance, and compares it with the sources to make sure no passage is copied.

    It may contain mistakes, and figures change over time. Nothing here is investment or buying advice. If something is wrong, tell us. Last refreshed 2026-11-14.

    The weekly digest

    Every Monday: the provider of the day, new pages on EveryNeocloud, and the conferences and anniversaries of the week ahead.