Cloud GPU: on-demand, reserved and spot, and why neoclouds change the board (with 2026 prices)

Contents

Notation: amounts in N USD or N €, decimal point. The dollar sign is not used (on this site it is the formula delimiter). All cloud prices change without notice; the date of the source is given next to each figure.

TL;DR

An H100 SXM5 on-demand costs between 2.49 USD/GPU-hour (Lambda, a neocloud) and 12.29 USD/GPU-hour (Azure ND H100 v5) depending on the provider, a spread of 5× between the neocloud floor and the hyperscaler ceiling. The reserved mode (a 1-3 year commitment) lowers the price by 20-40 % and is the number comparable to amortising your own iron; the spot/preemptible mode takes another 40-65 % off, but with a risk of interruption on less than 2 minutes’ notice. For the on-prem vs cloud analysis of this series (article sovereign on-premise vs hyperscalers), neocloud reserved is the correct reference: at 1.49-2.10 USD/GPU-hour with a 3-year commitment, neoclouds compete head-on with the TCO of your own iron at medium utilisation.


The three cloud GPU purchasing models

On-demand

Pay per use with no commitment, billed by the hour or by the minute. Maximum flexibility; maximum price. Applies to: development, experimentation, occasional capacity peaks and any workload whose lifetime is unpredictable. The absence of commitment has a cost: the on-demand premium over reserved ranges from 25 % to 60 % depending on the provider.

Reserved / committed

A 1 or 3 year contract that guarantees capacity in exchange for a discount on the on-demand rate. At hyperscalers it goes by the name Reserved Instance (AWS), Committed Use Discount (GCP) or Azure Reservation. At neoclouds it is negotiated directly and can include the choice of node, network (Ethernet or InfiniBand) and region. Typical discounts:

TermDiscount on on-demand (market reference)
1 year20-37 %
3 years30-45 %

Reserved is the model comparable to on-prem: in both cases a fixed cost is incurred up front (amortised capex on-prem; committed payment in the cloud). The formula for the effective cost per GPU-hour is the same in both cases:

$$\text{cost/GPU-hour}_{reserved} = \frac{\text{committed price (USD/GPU-h)} \times 8760}{\text{contracted hours}}$$

Unlike on-prem, cloud reserved includes neither hardware capex nor operational opex, but neither does it include the benefits of running your own iron at high load (see GPU utilisation as a FinOps lever).

Spot / preemptible

Idle provider capacity offered at a 40-65 % discount on on-demand, on the condition that it can be reclaimed with 30 seconds to 2 minutes of notice (AWS Spot, GCP Spot VMs). The discount is real and reproducible; so is the interruption risk.

Workloads suitable for spot:

WorkloadRequires checkpointingSuitable for spot
LLM pre-training (checkpoint-resume every 15-30 min)yesyes
LoRA/QLoRA fine-tuning with checkpointsyesyes
Offline batch inference (embeddings, evaluations)no (idempotent)yes
Hyperparameter searchno (repeatable)yes
Production serving with a latency SLOnot applicableno
Production serving with an availability SLOnot applicableno

A fine-tune of a 7B model on 4× H100 over 40 hours on spot achieved real savings of 81 % against the equivalent on-demand (Spheron Blog, May 2026). The typical interruption rate on high-reliability H100 instances is 2-8 % per day, which gives a 92-98 % probability of completing a 24-hour shift without a cut (Thunder Compute, Jun. 2026).


The neoclouds: GPU-specialised clouds

Definition and differentiators

Neoclouds (or GPU-specialised clouds) are infrastructure providers focused exclusively on GPU compute for AI workloads, without the diversity of services of the hyperscalers. SemiAnalysis coined the term in 2024 and identifies CoreWeave, Nebius, Lambda and Crusoe at the top tier (SemiAnalysis, 2024).

Structural differences with respect to the hyperscalers:

DimensionNeoclouds (CoreWeave, Lambda, Nebius…)Hyperscalers (AWS, GCP, Azure)
H100 on-demand price2.49-6.16 USD/GPU-hour6.88-12.29 USD/GPU-hour
Interconnect networkInfiniBand NDR 400 Gbps (CoreWeave, Voltage Park), Ethernet 200+ Gbps (Lambda, Nebius)Proprietary Ethernet (EFA, GVNIC) or InfiniBand (only p5/H100 on AWS)
Access to new hardwareNVIDIA priority access (CoreWeave is an NVIDIA shareholder); H200/B200 catalogue available earlierSlower catalogue; B200 in reserved or limited capacity in 2026
Contracts1-36 months, negotiated directlyStandard RI (1-3 years), no node negotiation
Additional servicesManaged Kubernetes, storage, bare-metalComplete ecosystem (IAM, VPC, databases, ML services…)
EgressZero-egress or flat fee (Lambda, Voltage Park, Nebius)0.08-0.12 USD/GB (AWS, GCP, Azure)
SovereigntyVaries: Nebius (EU), Crusoe (US, clean), CoreWeave (US/EU)US CLOUD Act; no guarantee of EU jurisdiction

The neocloud market is estimated at 20,000 M USD of revenue in 2026 and is growing towards 180,000 M USD in 2030 (Synergy Research Group, 2026). CoreWeave passed 5,000 M USD of ARR faster than any other cloud platform. Neoclouds price their H100s 60-85 % below AWS and Azure for the same silicon (Signisys, 2026).

Profiles of the main neoclouds

ProviderMain hardwareNetworkJurisdictionRelevant note
CoreWeaveH100, H200, B200, GB200InfiniBand NDR (Quantum-2 SHARP)US / EUNVIDIA shareholder; priority access to hardware; largest neocloud by fleet
Lambda LabsH100, A100, B200EthernetUSZero-egress; no spot; per-minute billing
NebiusH100, H200, B200, B300InfiniBandEU (Amsterdam)Spin-off of Yandex Cloud; zero-egress; object egress 0.015 USD/GiB; public on-demand price
CrusoeH100, A100, MI300XEthernetUS100 % renewable/flare gas energy; AMD MI300X at 3.45 USD/hr
Voltage ParkH100Ethernet (1.99 USD) / InfiniBand 3200 Gbps (2.49 USD)USMerged with Lightning AI in Jan. 2026; no minimums; non-profit Navigation Fund
Together AIH100, A100EthernetUSInference platform + raw compute
GMI CloudH100, H200EthernetAsia/USH200 on-demand from 2.60 USD/hr
RunPodH100, H200, B200, A100Ethernet / Secure CloudGlobalCommunity + Secure Cloud with SLA; spot available

Note: the “zero-egress”, “renewable” and priority hardware access claims are marketing from each provider and have not been verified by independent audits. They are reproduced with that caveat.


2026 price table: H100 SXM by provider and mode

Source and date given for each figure. Prices change without notice.

H100 SXM5 (80 GB HBM3) — USD per GPU-hour

ProviderOn-demand1-year reserved3-year reservedSpot/PreemptibleSource and date
CoreWeave~6.16~2.10-3.00 (negotiated)~1.49-2.10 (negotiated)not availableThunder Compute, Jun. 2026
Lambda Labs2.49-3.44~1.89 (37 % disc.)~1.84 (PCIe 3-yr)not availableSpheron Blog, May 2026
Nebius3.85discount up to 35 % (contact)discount up to 35 % (contact)2.15 (preemptible)nebius.com/prices, Jun. 2026
Voltage Park1.99 (Ethernet) / 2.49 (InfiniBand)contact (≥6 months)contactnot availablevoltagepark.com/pricing, Jun. 2026
RunPod (Secure)3.29not availablenot availableavailable (~59 % disc.)Spheron Blog, May 2026
Spheron2.50volume discount (contact)volume discount (contact)1.03Spheron pricing, May 2026
AWS (p5.48xlarge)6.88~2.97 (3-yr, ~23.78 USD/h node ÷ 8)~2.97~1.14 (p5 spot, 9.10 USD/h node)Vantage, Jun. 2026
GCP (A3 High)10.98~8.78 (1-yr CUD)n/a~3.69Spheron Blog, May 2026
Azure (ND H100 v5)12.29~6.15-8.60 (1-3 yr reservation)~6.15-8.60~2.27 (node spot ÷ 8)Vantage Azure, Jun. 2026

All amounts in USD/GPU-hour, rounded to 2 decimals. The “3-year reserved” column is the number comparable to the amortisation cost of your own hardware.

AWS p5.48xlarge: an 8× H100 node; on-demand 55.04 USD/h → 6.88 USD/GPU-h; 3-yr reserved 23.78 USD/h → 2.97 USD/GPU-h; spot 9.10-25.35 USD/h (varies by region and availability). GCP a3-highgpu-8g: 8× H100 SXM5; on-demand 87.84 USD/h → 10.98 USD/GPU-h; spot ~3.69 USD/GPU-h (us-central1, May 2026). Azure ND96isr H100 v5: 8× H100; on-demand 98.32 USD/h → 12.29 USD/GPU-h; node spot 18.17 USD/h → ~2.27 USD/GPU-h (source: Vantage, Jun. 2026, us-east-1).


2026 price table: H200 and B200

H200 SXM (141 GB HBM3e) — USD per GPU-hour

ProviderOn-demandSpot/PreemptibleSource and date
Nebius4.502.45 (preemptible)nebius.com/prices, Jun. 2026
GMI Cloud2.60not availableGMI Cloud, May 2026
RunPod4.39not availableSpheron Blog, May 2026
Spheron4.54not availableSpheron pricing, May 2026
CoreWeave6.31not availableGMI Cloud, May 2026
AWS (p5e)~4.98limitedSpheron Blog, May 2026
Azure~13.78not availableSpheron Blog, May 2026

B200 SXM6 (192 GB HBM3e) — USD per GPU-hour

ProviderOn-demandSpotSource and date
Lambda Labs4.99-5.29not availableSpheron Blog, May 2026
Nebius (HGX B200)7.153.95 (preemptible)nebius.com/prices, Jun. 2026
RunPod5.89not availableSpheron Blog, May 2026
Spheron6.022.12Spheron pricing, May 2026
AWS (p6-B200)~14.24~3.24Spheron Blog, May 2026

B200 availability is still limited in Q2 2026; many providers offer only access by advance reservation. The prices are market estimates subject to rapid change.


How it fits into the on-prem vs cloud comparison

The cloud cost per GPU-hour is the number that goes head to head with the on-prem TCO computed in sovereign on-premise vs hyperscalers. The relationship between the three purchasing modes and the cost analysis is as follows:

Cloud modeWhat it represents in the comparisonOn-prem equivalent
On-demandmarginal cost of peak capacity; floor of build-vs-buy at 100 % utilisation— (not comparable: on-prem has no elasticity)
3-year reservedthe committed cost comparable to capex amortised over 3 yearsnode amortisation + fixed annual opex
Spotthe cloud price floor, but with interruption risk— (not comparable: on-prem is not interrupted)

The cost per token formula applies the same to cloud as to on-prem:

$$\text{cost/1M tokens}_{cloud} = \frac{\text{GPU-hour price (USD)} \times 10^6}{\text{throughput (tok/s)} \times 3600}$$

With an H100 SXM5 at 2,800 tok/s (typical throughput for a 70B model in FP8):

ModeReference priceCost/1M tokens (70B, 2800 tok/s)
Neocloud on-demand (Lambda)2.49 USD/GPU-h~0.247 USD (~0.23 €)
Neocloud 3-yr reserved (CoreWeave)~1.49 USD/GPU-h~0.148 USD (~0.14 €)
Neocloud spot (Spheron/Nebius)~1.03-2.15 USD/GPU-h~0.102-0.213 USD
AWS on-demand (p5)6.88 USD/GPU-h~0.682 USD (~0.63 €)
AWS 3-yr reserved (p5)~2.97 USD/GPU-h~0.295 USD (~0.27 €)
GCP spot (A3)3.69 USD/GPU-h~0.366 USD (~0.34 €)
On-prem amortised, high util. (80 %)~2.83 USD/GPU-h all-in~0.281 USD (~0.26 €)

Reference exchange rate: 1 USD = 0.924 €. The rate fluctuates; comparative calculations must be done in the same currency.

The key reading: at high utilisation, on-prem TCO (~2.83 USD/GPU-hour all-in according to Spheron, 2026) competes with the 3-year reserved of the neocloud (~1.49-2.97 USD/GPU-hour). On-prem wins at high utilisation when the all-in cost drops below 2.5 USD; neocloud reserved wins when utilisation is medium or low, because it does not pay for idle capex. Spot lowers the cloud floor but is not comparable: interruption makes it incompatible with continuous serving, which is the case where on-prem competes.

0246810USD/GPU-hour →Spot Spheron H1001.03Spot Nebius H1002.15Neo reserved 3-yr(CoreWeave)1.49Lambda on-demand2.49Voltage Park OD1.99On-prem TCO (80 %)~2.83AWS p5 on-demand6.88Azure ND H100 v512.2912.29 USD/GPU-h

Network factors: InfiniBand vs Ethernet

For distributed training and large-scale inference (tensor parallelism across nodes), the bandwidth of the interconnect network between GPUs is critical. Neoclouds differ in what they offer:

ProviderGPU-GPU networkBandwidthApplies to
CoreWeaveInfiniBand NDR (Quantum-2, SHARP)400 Gbps/GPUmulti-node HGX clusters
Voltage ParkInfiniBand Quantum-23200 Gbps (8-GPU node)the 2.49 USD/hr tier
Voltage ParkEthernet200 Gbpsthe 1.99 USD/hr tier
Lambda LabsEthernet200 Gbpsall nodes
NebiusInfiniBand400 Gbps/GPUHGX H100/H200/B200
AWS p5EFA (Elastic Fabric Adapter)3200 Gbps (node)EC2 clusters
GCP A3GVNIC + RoCE200 Gbps/GPUA3 High
Azure ND H100 v5InfiniBand NDR400 Gbps/GPUND96isr H100 v5

InfiniBand NDR with SHARP (in-network collective reduction) cuts the latency of allreduce operations in distributed training. For inference serving with tensor parallelism inside a node (NVLink), the inter-node network is less critical; for multi-node training or disaggregated prefill-decode (the three axes of the series), the network between nodes matters.


Neoclouds and sovereignty: what changes

Unlike the US hyperscalers subject to the US CLOUD Act, some neoclouds offer options with jurisdiction outside the US:

ProviderEU regionUS CLOUD Act applies
NebiusAmsterdam (NL), Frankfurt (DE)No (Dutch legal entity)
CoreWeaveFrankfurt (DE), London (GB)Yes (US company)
Lambda LabsUS onlyYes
Voltage ParkUS onlyYes
ScalewayParis (FR), Amsterdam (NL)No (French company)
US hyperscalersEU regions availableYes (CLOUD Act applies)

For data subject to GDPR, only providers with an EU/EFTA legal entity guarantee the absence of the CLOUD Act. The full discussion of the sovereignty axis is in sovereign on-premise vs hyperscalers. For capacity planning with this data, see capacity planning for on-premise inference.


Decision table: purchasing mode by cost/risk/sovereignty axis

ModeCostOperational riskSovereignty (EU)When to use
Neocloud on-demandmedium (2-4 USD/GPU-h)low (no interruption)depends on the providerdevelopment, unpredictable peaks, serving with an SLO
Neocloud 1-yr reservedlow-medium (1.5-3 USD/GPU-h)low (no interruption, guaranteed capacity)depends on the providerstable serving, scheduled training; validated workload
Neocloud 3-yr reservedlow (1.5-2.1 USD/GPU-h)low, financial commitmentdepends on the providercomparable to amortised on-prem; very stable base load
Neocloud spotminimum (1-2.5 USD/GPU-h)high (interruption <2 min)depends on the providerbatch, fine-tuning with checkpoints, hyperparameter search
Hyperscaler on-demandhigh (6.9-12.3 USD/GPU-h)lowNO (CLOUD Act)integration with AWS/GCP/Azure services; no alternative
Hyperscaler reservedmedium-high (3-9 USD/GPU-h)lowNO (CLOUD Act)lock-in to the hyperscaler ecosystem is justified
Hyperscaler spotmedium (1.1-3.7 USD/GPU-h)high (interruption)NO (CLOUD Act)batch with no sovereignty requirement; more availability than neocloud spot
Amortised on-prem (>70 % util.)low (~2.3-2.8 USD/GPU-h all-in)low (no interruption)total (your own hardware)stable, high base load; maximum sovereignty

On-prem competes with neocloud 3-yr reserved when utilisation exceeds 65-70 % (see GPU utilisation as FinOps).


Computing the spot saving: an example with checkpointing

Reference parameters to validate whether spot is worth it:

$$\text{spot saving} = (\text{OD price} - \text{spot price}) \times \text{total GPU-hours} - \text{retry cost}$$ $$\text{retry cost} = \text{interruption rate} \times \text{lost work (GPU-hours)} \times \text{spot price}$$

With checkpointing every 30 minutes, work lost per interruption is ≤ 0.5 GPU-hours per event. At an interruption rate of 4 % per day on one H100 GPU:

ItemReference value
On-demand price (neocloud)2.50 USD/GPU-h
Spot price1.03 USD/GPU-h
Gross saving per GPU-hour1.47 USD
Expected interruptions in 40 h (4 %/day)~0.67 events
Re-run work per event (0.5 h × 1.03 USD)~0.34 USD
Net saving on a 40 h × 4 GPU job~234 USD (against 400 USD on-demand)

The 58 % saving only materialises if the job implements checkpoint-resume correctly. Without checkpoints, a single interruption forces a restart from scratch: the saving turns into an overcost from duplicated work.


Integration into the series TCO model

The cloud €/GPU-hour is the number that feeds the FinOps analysis of the series. The chain:

  1. Cost of cloud iron (this price table) → equivalent to amortised capex + opex of on-prem.
  2. Throughput (measured with genai-perf, track B of the series) → the denominator.
  3. Cost per token (cost per token and per request) → the final comparative metric.
  4. Utilisation (GPU utilisation as FinOps) → in on-demand cloud, utilisation does not penalise cost (you pay for what you use); in reserved and in on-prem it does penalise, because the cost is fixed.

The correct comparison is neocloud 3-yr reserved vs on-prem amortised over 3 years, with the same assumptions on throughput, egress and utilisation. Spot is the cloud price floor but not the comparable floor: its interruptible nature excludes it from any comparison with an on-prem cluster that serves continuous traffic.


Sources