Genesis Cloud Pricing 2026: EU H100 Cost vs Spheron

Genesis Cloud lists NVIDIA HGX H100 SXM5 at $2.19/GPU-hr on its own pricing page, a rate that reads like one of the cheapest H100 quotes in Europe. It's also a rate you can't actually buy at that unit size: the minimum order is a full 8-GPU node, so the real invoice is $17.52/hr whether your job uses one GPU or eight.

That single detail is the whole story of Genesis Cloud pricing in 2026. The company genuinely has an EU data-residency and sustainability pitch worth taking seriously. It's also, per its own Handelsregister filing, a company in formal liquidation that has been actively selling GPU capacity for nearly a year since the filing. Both facts matter to a buyer deciding whether to put a production workload on this platform. This post prices the H100, H200, and B200 lines, works through what the full-node minimum actually costs you, and compares every figure against Spheron's live rates, pulled from our current GPU pricing API on 30 July 2026.

Genesis Cloud's EU-First Pitch: Data Residency, Sustainability, and Full-Node Minimums

Genesis Cloud is a Munich-based provider that built its early business on Nordic capacity, first in Iceland, then a Norway data center developed with Bulk Infrastructure Group, per Genesis Cloud's own pricing page (genesiscloud.com/pricing). The pitch on that page is consistent with the rest of the EU sovereign-cloud field we've covered in our OVHcloud pricing breakdown and Scaleway H100 pricing post: Tier 3 EU data centers, 100% green-energy power, and ISO 27001 certification, with "EU sovereign cloud compliance" language on top.

That pitch lands at a useful moment. Gartner forecasts European sovereign cloud IaaS spending will hit $12.6 billion in 2026, up 83% year over year from $6.9 billion in 2025, and nearly double again to $23.1 billion in 2027, when it's projected to overtake North American sovereign spend for the first time, per webhosting.today's coverage of the Gartner numbers. Every EU-incorporated provider with real GPU capacity is competing for a slice of that budget, and Genesis Cloud is one of the smaller names in that field: it raised a relatively modest $6.6 million before its later troubles, per Bizety's reporting, a fraction of what Scaleway or OVHcloud have raised or generated in revenue.

What Genesis Cloud Actually Sells: H100 SXM5, H200, and B200 Nodes Only

Genesis Cloud's current catalog is narrower than either of those competitors. Per its pricing page and on-demand GPU product page, the lineup is NVIDIA HGX H100, HGX H200, and HGX B200 clusters connected with 3.2 Tbps InfiniBand, plus older RTX-class cards for cost-sensitive workloads. There's no PCIe or NVL H100 tier the way Spheron or OVHcloud list, and no A100 line still in active marketing. The InfiniBand fabric is a real spec for multi-node training jobs, but it's also the kind of spec that only pays off if you're actually running a multi-node job, not testing a single fine-tune.

The more consequential catalog detail is the minimum order size. All three current-generation lines, H100, H200, and B200, sell in full 8-GPU nodes only, with no single-GPU SKU, per Genesis Cloud's own product pages. That's a materially different buying model from a marketplace like Spheron, where you can rent 1x, 2x, 4x, or 8x H100 SXM5 configs depending on the job. We cover why that form-factor and config flexibility matters in our H100 NVL vs SXM5 vs PCIe guide; Genesis Cloud only sells the SXM5 variant, and only at node scale.

Genesis Cloud's Corporate Status: What "GmbH i.L." Means for Buyers

Genesis Cloud GmbH, registered in Munich under HRB 250051, entered formal liquidation under German law on 7 August 2025. The company's liquidators, Haiko Depping and Daniel Nießen, filed a "Continuation" notice as recently as 25 June 2026, meaning the entity remains registered and in active liquidation proceedings as of mid-2026, per Genesis Cloud's Handelsregister record on Northdata. This is not a rumor or a speculative reading of a filing. It's the company's own registered legal status, and it shows up on the company's own homepage: Genesis Cloud displays its name as "Genesis Cloud GmbH i.L.", the German legal marker for "in liquidation," per independent coverage from Bizety.

Any buyer evaluating Genesis Cloud for a production workload needs to treat that status as a first-order fact, not a footnote. A company in formal liquidation can, and often does, continue trading while the wind-down proceeds. That's apparently what's happening here. It's a different risk profile than an operating going concern, and it's worth weighing against whatever GPU-hour savings the platform offers, particularly for any workload with multi-month time horizons.

Formal Liquidation Since August 2025, Still Selling Capacity in 2026

Genesis Cloud was founded in 2018, headquartered in Munich, and raised around $6.6 million in funding, a small round for infrastructure-heavy AI compute, per Bizety's reporting. Bizety frames the liquidation as reflecting pressures common across the neocloud segment: aggressive capital expenditure on GPU hardware, high operating costs, and sharply declining GPU rental prices squeezing margins on both older and newer hardware generations.

None of that has stopped the company from actively pricing and marketing HGX H100, H200, and B200 clusters through 2026, per its own pricing page. The liquidators' continuation filing in June 2026 confirms proceedings are ongoing, not resolved. If your procurement process has a vendor-viability gate, that gate should flag Genesis Cloud today regardless of how competitive the headline per-GPU rate looks. Vendor solvency belongs in the same regulatory diligence checklist EU teams already run for data residency and compliance before committing a workload to any provider.

We checked genesiscloud.com directly while researching this post, on 30 July 2026, and the entire site was unreachable: the bare domain times out with a Cloudflare 522 origin-timeout error, and every path under www.genesiscloud.com, including the pricing page and all three GPU product pages, returns a cached 404 served from Genesis Cloud's own CDN. That's not a one-off dead link. It's the marketing and pricing surface of a company in active liquidation going dark, which is the kind of vendor-viability signal a procurement checklist is meant to catch. Anyone trying to verify pricing directly before signing a contract should try genesiscloud.com themselves first, and treat a continued outage as a reason to escalate the vendor-viability question, not a reason to skip it.

H100 and H200 Pricing Compared to Spheron and the Hyperscalers

Here's the direct number comparison, with Genesis Cloud's figures pulled from its own product pages and Spheron's pulled live from our pricing API on 30 July 2026, split strictly by instanceType (DEDICATED for on-demand, SPOT for spot) before taking a minimum, so a spot rate never gets mislabeled as on-demand.

Pricing fluctuates based on GPU availability. The prices above are based on 30 Jul 2026 and may have changed. Check current GPU pricing → for live rates. Genesis Cloud's figures are compiled from Genesis Cloud's own pricing and product pages; as of this writing, genesiscloud.com itself is returning errors sitewide (see the outage note above), so verify current figures directly with Genesis Cloud once the site is back before budgeting.

H100 SXM5: Genesis Cloud $2.19/GPU vs Spheron's Live On-Demand Floor

Genesis Cloud's $2.19/GPU-hr for H100 SXM5, per its pricing page, undercuts Spheron's live 8-GPU on-demand floor of about $2.54/GPU-hr. That's a real, if narrow, win for Genesis Cloud, and we're not going to bury it. But it's a win that only applies if you're buying the full 8-GPU node on both platforms. Spheron's true single-GPU H100 SXM5 on-demand rate currently runs $4.06-$4.41/GPU-hr, and Spheron's H100 PCIe tier undercuts Genesis Cloud outright at $2.01/GPU-hr on-demand with no node minimum at all, a config Genesis Cloud doesn't currently sell. For a hyperscaler baseline, Azure's ND H100 v5 runs $12.29/GPU-hr on-demand, per our own Azure H100 pricing breakdown, which puts both Genesis Cloud and Spheron well below the hyperscaler tier regardless of config.

H200: Genesis Cloud $2.80/GPU vs Spheron On-Demand and Spot

H200 follows the same shape. Genesis Cloud lists HGX H200 at $2.80/GPU-hr on-demand, per its H200 product page, again as an 8-GPU-only SKU, working out to $22.40/hr for the node. Spheron's current H200 SXM5 on-demand floor sits higher at $4.54/GPU-hr, since Spheron doesn't currently list an 8x dedicated H200 config, only 1x. Spheron's H200 spot floor of $3.31/GPU-hr closes most of that gap for workloads that can tolerate reclamation, though spot capacity can be pulled without notice, so it's not a fair substitute for guaranteed on-demand access on a production inference endpoint.

B200: Genesis Cloud On-Demand vs 36-Month Reserved vs Spheron

Genesis Cloud's Blackwell tier prices at $3.75/GPU-hr on-demand, dropping to $2.80/GPU-hr on a 36-month reserved commitment, per its B200 product page. A 36-month reserved term is a long lock-in to sign with a counterparty already in formal liquidation, whatever the discount looks like on paper. Spheron's B200 on-demand rate currently runs $9.36/GPU-hr, only available as a 1x config right now, with spot pricing at $5.34-$5.37/GPU-hr. On raw on-demand price, Genesis Cloud's B200 rate is cheaper today, again with the node-minimum and vendor-status caveats attached.

What a Full-Node Minimum Costs You If You Only Need One GPU

Direct answer: if your workload genuinely needs 8 GPUs, Genesis Cloud's per-GPU rate is competitive with Spheron's own 8-GPU on-demand floor. If it needs one GPU, you're paying for eight of them regardless, and that turns a competitive per-GPU headline into one of the more expensive ways to get a single H100 in Europe.

Most GPU rental decisions aren't 8-GPU decisions. A single fine-tune, an inference endpoint serving one model, a small research job, all of these are one- or two-GPU workloads in practice. Genesis Cloud has no SKU for that. Its H100, H200, and B200 lines are 8x-node-only, so a team evaluating "H100 from $2.19/hr" on the pricing page needs to immediately re-read that as "H100 from $17.52/hr, minimum."

The Math: $17.52/hr Forced Spend vs a True Single-GPU Rate

Run the numbers side by side. A buyer who needs exactly one H100 SXM5 pays $17.52/hr on Genesis Cloud, the full 8-GPU node price, because there's no smaller unit to buy. The same buyer on Spheron pays roughly $4.06-$4.41/hr for a real single-GPU H100 SXM5 on-demand instance. That's about 4x more spend on Genesis Cloud for identical single-GPU capacity, even though Genesis Cloud's nominal per-GPU list price looked lower going in.

The same math compounds on H200 and B200. An 8-GPU H200 node runs $22.40/hr on Genesis Cloud regardless of how many of those GPUs you actually use; a B200 node runs $30.00/hr on-demand or $22.40/hr on the 36-month reserved tier. None of those numbers change if your workload only needs a fraction of the node. If you're not running distributed training across all 8 GPUs simultaneously, that unused capacity is dead spend, not idle infrastructure you can spin down independently, since Genesis Cloud sells and presumably bills the node as a single unit.

This is the practical reason node-minimum policies matter more than headline per-GPU rates for most buyers. A provider with a higher per-GPU price and a 1x-GPU floor, like Spheron's H100 PCIe tier at $2.01/GPU-hr, can be cheaper in total dollars than a provider with a lower per-GPU price and an 8x-GPU floor, the moment your actual GPU requirement is smaller than the minimum node.

EU Data Residency Without the Full-Node Markup

Data residency and full-node pricing are separate problems, and it's worth not conflating them. Genesis Cloud's ISO 27001 certification, green-energy claims, and EU-only data center footprint are real attributes that matter for teams with a genuine EU-jurisdiction requirement, the same category of requirement we cover in more depth in our EU AI Act compliance guide for GPU cloud and our broader European GPU cloud provider roundup. Neither of those attributes requires an 8-GPU minimum order to deliver; that's a Genesis Cloud catalog decision, not an inherent property of EU-based infrastructure.

Where Spheron Does and Doesn't Match Genesis Cloud's EU Placement Claims

Spheron aggregates capacity from 5+ providers, which includes EU-region placement options alongside Spheron's H100 SXM5 and PCIe availability, without forcing an 8-GPU commitment to access them. What Spheron does not claim is Genesis Cloud's specific ISO 27001 certification or its green-energy sourcing story; those are provider-specific attestations you'd need to verify directly with whichever underlying EU data center partner is fulfilling a given Spheron instance. If ISO 27001 on the GPU rental entity itself is a hard procurement requirement, that's a genuine point in Genesis Cloud's favor, separate from the liquidation status covered above. If the requirement is EU placement without a specific certification attached, on-demand H100 access on Spheron covers that without the node-minimum markup. For the full multi-provider view beyond Genesis Cloud and Spheron, our Nebius H100 and H200 pricing breakdown is the next EU-region comparison worth reading. Deployment and API details for Spheron itself are in the docs.

If your H100 or H200 workload doesn't need a full 8-GPU node and doesn't need to sign a multi-year commitment with a provider in formal liquidation, on-demand H100 access on Spheron is worth pricing out first.

Spheron's H100 instances →