Denvr Dataworks Pricing 2026: H100 and GB200 Cost vs Spheron

Denvr Dataworks publishes real, self-serve rates for A100, H100, and GH200. Then, right where it matters most, the rate card stops: H200 is "Reserved only," and GB200, B200, and B300 all route to "Request a Quote." That split matters more than any single number, because it means Denvr isn't uniformly cheap or uniformly opaque. It's cheap and transparent on older silicon, and closed on the newest.

We pulled Denvr's numbers straight from denvr.com/pricing and matched them against Spheron's live rates from the Spheron GPU pricing API, split strictly by instanceType so we're comparing on-demand to on-demand and spot to spot, not mixing the two. If you're also comparing other neoclouds with the same published-vs-sales-gated split, Crusoe Cloud's pricing follows a nearly identical pattern, and our Nebius H100 and H200 breakdown covers a provider that publishes both tiers.

Denvr Dataworks Pricing 2026: The Published Rate Card

Denvr's pricing page lists exactly three GPU configurations with a self-serve hourly rate: A100 SXM, H100 SXM, and GH200. Everything bills per minute on-demand with no long-term commitment required, and Denvr accepts prepaid credit cards or invoiced bank transfer as payment (Denvr pricing).

On-Demand: A100 SXM, H100 SXM, and GH200 Per-GPU Cost

Source: Denvr Dataworks pricing page. Notice the interconnect scales with the tier: A100 nodes get IB 1600G, H100 nodes double that to IB 3200G, and the single-GPU GH200 instance ships with IB 400G, which is plenty for a standalone Grace Hopper box but not built for multi-node training.

There's a separate number worth knowing about: pricing tracker ComputePrices lists a single-GPU A100 SXM tier from Denvr at $0.58/hr, tracked as of 27 Jul 2026, which it calculates as 67% below the average A100 SXM on-demand rate across the providers it monitors (ComputePrices A100 SXM tracker). That's a different SKU from the 8-GPU node above, single-GPU boxes instead of an 8-GPU chassis, but it's the cheapest A100 rate in this entire comparison by a wide margin.

Pricing fluctuates based on GPU availability. The prices above are based on 29 Jul 2026 and may have changed. Check current GPU pricing → for live rates.

H200, GB200, and B200: Reserved-Only, No Self-Serve Rate

Denvr's H200 SXM listing (8-GPU, 141GB HBM3e per GPU) is marked "Reserved only" on the pricing page, no hourly figure attached. GB200, B200, B300, and Denvr's Gaudi accelerators don't appear on the rate card at all; they route straight to "Request a Quote" (Denvr pricing).

That's a meaningful line to draw. A100, H100, and GH200 are older, more available silicon where Denvr can price aggressively and still fill capacity. H200 and Blackwell-generation parts are the supply-constrained tier where allocation gets matched to committed customers through a sales process instead of a walk-up rate card. It's the same pattern most of the neocloud market follows for its newest hardware, our B200 cloud pricing roundup shows several providers doing the same thing.

What's Included (and Not): Networking, Storage, Support Tiers

Every published Denvr instance bundles compute, the listed NVMe storage, and the listed InfiniBand interconnect into one hourly rate; there's no separate line item for networking or local storage on the on-demand tiers. What isn't itemized anywhere on the public pricing page is support tier, SLA terms, or uptime guarantees. Those aren't unusual omissions for a self-serve GPU rate card, but they're also not something you can compare against Spheron's numbers without asking Denvr directly.

Reserved Contracts: Save Up to 20%, Terms Not Published

Denvr advertises "Save up to 20% with multi-month or annual commitments" directly on the pricing page (Denvr pricing). There's no rate table showing what that 20% actually applies to, no minimum commitment length, and no published overage terms if you exceed your reserved quantity. You'd need to talk to Denvr's sales team to find the real number.

The math you'd want to run once you have that number: does the discount clear your actual utilization? A 20% cut on a resource you use 60% of the time is a worse deal than paying on-demand for the hours you actually need. Our serverless vs on-demand vs reserved GPU guide walks through that break-even calculation in more detail, and it applies here regardless of which provider's reserved tier you're evaluating.

The Sales-Quote Gap for GB200, B200, and B300

If your workload needs Blackwell-generation hardware today, Denvr's public pricing page doesn't help you comparison shop. GB200, B200, and B300 all sit behind "Request a Quote," and you won't see a number until you've already started that conversation.

This is where the comparison with Spheron gets genuinely interesting, because Spheron's GB200 access isn't a pure self-serve checkout either. Spheron's GB200 NVL72 page runs on a request-and-confirmation model: you submit GPU quantity and workload details, and the team confirms availability within one business day with pricing based on quantity, commitment length, region, and networking. That's still faster and more transparent than an open-ended sales quote, but it's not a live hourly rate you can pull off a page the way you can for H100 or A100.

Market context helps here too. Across the wider GB200 NVL72 market, published or estimated rates run roughly $10.50/GPU-hr on CoreWeave up to around $16/GPU-hr on Oracle and $27/GPU-hr on Azure, according to our own GB200 NVL72 pricing guide. Neither Denvr nor Spheron publishes a number in that range for GB200 specifically right now, which is the honest takeaway: this is the one GPU tier in this whole comparison where nobody wins on transparency.

Denvr vs Spheron: Cost Per GPU-Hour Side by Side

This is where published numbers meet published numbers. Spheron's rates below are pulled live from the Spheron GPU pricing API as of 29 Jul 2026, filtered strictly by instanceType so on-demand (DEDICATED) and spot offers are never mixed together.

A100 SXM: Live Rates Compared

Denvr wins the A100 comparison outright at both tiers it publishes. Its $0.58/hr single-GPU rate beats even Spheron's spot price, and its $1.35/hr 8-GPU node rate is well under Spheron's cheapest on-demand A100 offer. Spot pricing narrows the gap ($0.82/hr on Spheron vs $1.35/hr for Denvr's 8-GPU on-demand tier) but doesn't close it against Denvr's single-GPU rate. If A100 is genuinely all you need, Denvr's published card is hard to beat on price alone. For the SXM vs PCIe and MIG-partitioning tradeoffs that affect which A100 tier actually fits your workload, see our A100 deployment guide.

Pricing fluctuates based on GPU availability. The prices above are based on 29 Jul 2026 and may have changed. Check current GPU pricing → for live rates.

H100 SXM: Live Rates Compared

At the equivalent 8-GPU SXM tier, Denvr's $2.30/hr beats Spheron's $2.54/hr by about 9%. That's the honest read, not a rigged one. Where Spheron pulls ahead is form factor and billing flexibility Denvr doesn't offer at all: H100 PCIe on Spheron runs $2.01/hr on-demand and $1.67/hr spot, both cheaper than Denvr's SXM rate, and Denvr has no spot tier on any GPU to compete with Spheron's interruptible pricing. If your workload genuinely needs SXM-class NVLink bandwidth, Denvr is the cheaper published rate today. If it doesn't, PCIe on Spheron undercuts both.

Pricing fluctuates based on GPU availability. The prices above are based on 29 Jul 2026 and may have changed. Check current GPU pricing → for live rates.

GB200 and H200: Two Reserved-Only Tiers, Compared Honestly

H200 is the clearer gap: Spheron publishes an actual on-demand and spot number, Denvr doesn't. GB200 is the fair fight the assignment's angle is built on, neither company runs GB200 as a self-serve checkout item. Denvr's process is open-ended sales contact; Spheron's is a structured request form with a one-business-day confirmation window and pricing tied to quantity, term, region, and networking. Same lack of a public rate, different process for getting one.

Real-World Monthly Cost at 200 Hours/Month

200 hours a month is a reasonable proxy for a team running GPUs on weekdays during business hours, common for iterative fine-tuning or inference load testing. Per-GPU cost at the 8-GPU node tier:

• A100 SXM, Denvr: $1.35 × 200 = $270/month
• A100 SXM, Spheron on-demand: $2.23 × 200 = $446/month
• A100 SXM, Spheron spot: $0.82 × 200 = $164/month (cheaper than Denvr's on-demand rate)
• H100 SXM, Denvr: $2.30 × 200 = $460/month
• H100 SXM, Spheron on-demand: $2.54 × 200 = $508/month
• H100 PCIe, Spheron on-demand: $2.01 × 200 = $402/month (cheaper than Denvr's SXM rate)
• H100 PCIe, Spheron spot: $1.67 × 200 = $334/month
• GH200, Denvr: $3.87 × 200 = $774/month
• GH200, Spheron on-demand: $3.02 × 200 = $604/month

Spot pricing is what actually flips the A100 comparison in Spheron's favor, and PCIe is what flips H100. Neither number stays fixed for long: Spheron's rates move with real-time availability across 5+ providers, and Denvr's published card can change too. Run this with your own hours before you budget against it.

Pricing fluctuates based on GPU availability. The prices above are based on 29 Jul 2026 and may have changed. Check current GPU pricing → for live rates.

When Denvr's Model Makes Sense vs a Neocloud Marketplace

The tier-by-tier math above doesn't produce one universal winner, and pretending it does would be dishonest:

• You need A100 SXM and want the lowest published rate, full stop. Denvr's $0.58/hr single-GPU and $1.35/hr 8-GPU node rates beat Spheron at every tier we could match against them.
• You need H100 with SXM-class NVLink and want the cheapest self-serve node rate. Denvr's $2.30/hr edges out Spheron's $2.54/hr for the equivalent 8-GPU configuration.
• You don't need SXM interconnect, or your workload can tolerate spot interruption. Spheron's H100 PCIe on-demand ($2.01/hr) and spot ($1.67/hr) both undercut Denvr, and Denvr has no spot tier at any GPU tier to compare against.
• You need H200, GB200, or B200 today without an open-ended sales process. Spheron publishes live H200 rates outright, and its GB200 request flow at least comes with a one-business-day confirmation window and defined pricing inputs, versus Denvr's undated quote request.
• You want to compare multiple providers' rates in one place instead of one company's own data centers. Spheron aggregates offers across 5+ providers, so you're comparing regions and vendors side by side. Denvr is a single company's own infrastructure, which can mean tighter hardware control but also means you're only ever seeing one number, not a market.

For the hidden-cost and contract-term questions worth asking any provider before you sign, our AI buyer's guide covers what to check beyond the sticker price. If you want Denvr positioned against the broader market rather than just Spheron, the GPU cloud pricing comparison across 15+ providers puts every published rate side by side. Deployment docs for spinning up an instance on Spheron are at docs.spheron.ai.

Denvr's rate card is honest where it publishes, A100 and H100 SXM rates are real, comparison-shoppable numbers, and the A100 pricing in particular is tough to beat anywhere. It just stops publishing at H200. For a live H200 rate, or for GB200 access with a defined confirmation window instead of an open-ended sales call, check Spheron's numbers today.

Check current H100 and A100 pricing on Spheron →