What Happened

On September 22, 2026, Helsinki-based Verda Cloud Oy — formerly operating as DataCrunch — announced a $189 million Series B funding round led by Emergence Capital. The round drew participation from MUFG Innovation Partners, Supermicro, Varma Mutual Pension Insurance Co., Lifeline Ventures, 6 Degrees Capital, byFounders, and Tesi. Notable angel investors include Ola Tørudbakken, director of AI systems at Meta, and Mark Saroufim, co-founder of Core Automation.

The raise brings Verda's total funding to over $450 million in equity and debt. The company currently operates data center capacity in Finland and plans to expand across Europe, the UK, the US, and Asia, targeting more than 250 megawatts of operational capacity by 2027. Verda also confirmed it is preparing early deployments of Nvidia's VR200 NVL72 rack-scale supercomputers in the coming months.

Verda differentiates itself from commodity GPU renters by building its own compilers and serving software, with an internal AI lab focused on GPU utilization, inference optimization, and kernel engineering. Existing customers include German sovereign AI firm Aleph Alpha, AI visual content company Magnific (processing millions of inference requests daily), and Epsilon Health, which trains bespoke radiology AI models on dedicated Verda clusters.

Why It Matters

The neocloud market has been one of the most aggressively funded segments in AI infrastructure over the past 18 months. Runpod raised $100 million in June 2026 for its developer-focused cloud platform. Starcloud secured $250 million in August for orbital data centers. Aranya raised $11 million in September for rapid bare-metal GPU cluster provisioning. Verda's $189 million Series B fits squarely in this pattern — but with a distinct European sovereign AI angle and a compiler-level optimization thesis.

What sets Verda apart is its claim of building custom compilers and serving software rather than reselling raw GPU capacity. If the company can genuinely improve inference performance per dollar through kernel engineering and GPU utilization optimization, it addresses a real pain point: enterprises running production AI workloads are increasingly cost-sensitive at the inference layer, not just the training layer. The partnership with Aleph Alpha — a prominent European sovereign AI player — validates the data residency use case that US-based providers struggle to serve.

However, Verda remains early-stage. Scaling from Finland-based capacity to 250+ MW across four continents by 2027 is an ambitious target, and the company has yet to prove it can operate at the reliability and scale of incumbents like CoreWeave or Lambda. The VR200 NVL72 deployments will be an early test of whether Verda can handle rack-scale infrastructure, not just individual GPU nodes.

Who Is Affected

European AI startups and enterprises now have another well-funded managed inference provider to evaluate, particularly those with data sovereignty requirements that rule out US-based hyperscalers. GPU cloud customers running high-volume production inference should benchmark Verda's compiler-optimized serving against their current providers — the price-per-token differential could be meaningful if the optimization claims hold. Sovereign AI initiatives across Europe gain additional domestic infrastructure capacity, which matters for policy-driven AI deployment in healthcare, government, and defense.

Strategic Implications

For AI startup founders

If you're building in Europe and need managed inference with data residency guarantees, Verda is now a credible Series B-funded option worth benchmarking against CoreWeave, Lambda, and Runpod. Their custom compiler stack could translate to better price-per-token — but validate with your own benchmarks on your specific model architecture before committing. The 250 MW by 2027 target means capacity should be available, but track whether deployments actually come online on schedule.

For developers/operators building with AI APIs

Verda's focus on inference optimization and kernel engineering suggests they're targeting the cost-sensitive production inference market, not training workloads. If you're running high-volume inference (millions of requests daily, like their customer Magnific), their platform may offer better GPU utilization than commodity providers. The upcoming VR200 NVL72 deployments will be a signal of whether they can handle rack-scale infrastructure — watch for performance benchmarks from those systems.

For non-technical business owners evaluating AI tools

This funding round signals continued investor confidence in AI infrastructure, which should keep compute costs competitive across providers. For European businesses specifically, Verda's expansion adds a viable domestic alternative to US-based cloud providers — relevant if you have GDPR-driven data residency requirements or are working with government and healthcare partners that mandate sovereign infrastructure.

What to Watch Next

Monitor whether Verda's VR200 NVL72 deployments come online on schedule in the coming months — that will be the first concrete signal of whether the company can execute at rack scale. Also watch for any benchmark data comparing Verda's inference performance against CoreWeave or Lambda on standard models. If Verda announces US or Asia data center locations coming online, that would signal faster-than-expected geographic expansion.

Frequently Asked Questions

Q: What is Verda Cloud and how is it different from other GPU cloud providers?

Verda Cloud (formerly DataCrunch) is a Helsinki-based AI cloud provider that builds its own compilers and serving software rather than reselling raw GPU capacity. Its internal AI lab focuses on GPU utilization, inference optimization, and kernel engineering — aiming to deliver better performance per dollar on inference workloads compared to commodity GPU renters.

Q: How much funding has Verda Cloud raised and what will it be used for?

Verda has raised over $450 million in total equity and debt, including a $189 million Series B announced on September 22, 2026. The funding will be used to expand data center capacity to 250+ megawatts by 2027 across Europe, the UK, the US, and Asia, deploy Nvidia VR200 NVL72 systems, and invest in its AI lab and developer experience platform.

Q: Should European enterprises consider Verda for AI workloads with data residency requirements?

Verda's Finland-based infrastructure and partnership with sovereign AI firm Aleph Alpha make it a credible option for European enterprises with data residency needs. However, as an early-stage company scaling rapidly, enterprises should validate reliability and SLA commitments before migrating production workloads.