iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
On the published H100 list prices observed on October 18, 2026, Lambda is less expensive than CoreWeave for comparable on-demand capacity: Lambda’s eight-GPU H100 SXM plan is $3.99 per GPU-hour, while CoreWeave’s North America HGX H100 listing is $49.24 per eight-GPU instance-hour, or $6.16 per GPU-hour. CoreWeave also lists an eight-GPU spot rate of $19.71 per instance-hour and a separate single-GPU inference rate of $6.16 per GPU-hour. AWS cannot be assigned one fair number without choosing an exact EC2 GPU instance, region, operating system and purchase option; its calculator is required for an apples-to-apples quote.
Published H100 prices at a glance
The figures below are provider list prices observed on October 18, 2026. They are not guaranteed quotes: cloud rates can vary by region, plan, availability and billing arrangement.
| Provider and product | Published price | Normalized comparison | Pricing type and scope |
|---|---|---|---|
| CoreWeave HGX H100 | $49.24 per hour | $6.16 per GPU-hour (49.24 ÷ 8) | Eight NVIDIA H100 GPUs, North America, on-demand |
| CoreWeave HGX H100 spot | $19.71 per hour | $2.46 per GPU-hour (19.71 ÷ 8) | Eight GPUs, North America, spot |
| CoreWeave H100 inference | $6.16 per GPU-hour | $6.16 per GPU-hour | Single-GPU inference rate, North America; separate from the HGX instance listing |
| Lambda H100 SXM, eight-GPU plan | $3.99 per GPU-hour | $31.92 for eight GPUs per hour | Eight-GPU plan, on-demand list price |
| Lambda H100 SXM, four-GPU plan | $4.09 per GPU-hour | $16.36 for four GPUs per hour | Four-GPU plan, on-demand list price |
| Lambda H100 SXM, two-GPU plan | $4.19 per GPU-hour | $8.38 for two GPUs per hour | Two-GPU plan, on-demand list price |
| Lambda H100 SXM, one-GPU plan | $4.29 per GPU-hour | $4.29 per GPU-hour | One-GPU plan, on-demand list price |
| Lambda A100 SXM, eight-GPU plan | $2.79 per GPU-hour | $22.32 for eight GPUs per hour | Eight-GPU plan, on-demand list price; different GPU generation |
On the comparable eight-GPU H100 entries, Lambda’s $31.92 equivalent hourly total is $17.32 below CoreWeave’s $49.24 on-demand instance price. That is about 35% less than CoreWeave’s instance rate, although the providers may differ in hardware configuration, software environment, location and included services. Comparing only the number can therefore hide practical differences.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallWhat does a CoreWeave H100 cost?
On-demand HGX H100
CoreWeave lists an eight-GPU NVIDIA HGX H100 instance in North America at $49.24 per instance-hour. Dividing by eight gives $6.155, displayed as approximately $6.16 per GPU-hour. The original instance size matters: this is an eight-GPU allocation, not a promise that one GPU can be rented at one-eighth of the instance price.
#1 Best Overall
- ATTENTION BUYERS: EXPORTING this item outside the U.S. requires strict compliance with the Bureau of Industry and Security (BIS) U.S. Export Administration Regulations (EAR). Non-compliance carries fines up to $1 MILLION PER VIOLATION AND UP TO 20 YEARS IMPRISONMENT for both the buyer and seller. To uphold our regulatory obligations, we strictly follow BIS "Know Your Customer" guidance by screening each order and immediately aborting transactions showing unresolved regulatory red flags. ***
- [Double-Flow-Through Design] The RTX PRO 6000 Blackwell features a double-flow-through cooling design, optimizing efficiency and airflow to sustain peak performance under 600W power loads. | [5th Gen Tensor Cores] Deliver up to 3X the performance of the previous generation and support for FP4 precision for faster AI model processing times with reduced memory usage, enabling local fine-tuning of LLMs and generative AI | [4th Gen Ray Tracing Cores] Double the ray-triangle intersection rate of the previous generation to create photoreal, physically accurate scenes and immersive 3D designs with RTX Mega Geometry, which enables up to 100X more ray-traced triangles.
- [PCIe Gen 5] Support for PCIe Gen 5 provides double the bandwidth of PCIe Gen 4, improving data-transfer speeds from CPU memory and unlocking faster performance for data-intensive tasks like AI, data science, and 3D modeling. | [GDDR7 Memory] With 96 GB of GPU memory and 1.8 TB ps bandwidth, it can tackle massive 3D and AI projects, fine-tune AI models locally, explore large-scale VR environments, and drive larger multi-app workflows.
- [DisplayPort 2.1] Achieve unparalleled visual clarity and performance, driving high resolution displays at up to 8K at 240 Hz and 16K at 60 Hz. Increased bandwidth enables seamless multi-monitor setups while HDR and higher color depth support ensures superior color accuracy for precision work, such as video editing, 3D design, and live broadcasting.
- [NVIDIA Blackwell Streaming Multiprocessor] The new SM features increased processing throughput, and new neural shaders that integrate neural networks inside of programmable shaders | DLSS 4: Multi Frame Generation ensures ultra-smooth frame pacing for lifelike simulations. | [Universal MIG] Divide a single RTX PRO 6000 Blackwell into multiple isolated instances, each with dedicated resources, allowing for concurrent execution of multiple workloads, optimized GPU utilization, and secure isolation of different applications or users. [WARRANTY] 3 YR Manufacturer's Warranty. Bulk OEM Packaging. Retail Packaging is NOT included.
Spot HGX H100
The same CoreWeave listing shows $19.71 per eight-GPU hour for spot, equivalent to approximately $2.46 per GPU-hour. Spot pricing is a separate purchase mode. Before using it for training, set an interruption and checkpointing policy, and confirm current availability and terms with CoreWeave. A low hourly rate is useful only if your workload can resume safely when capacity is reclaimed.
Inference rate
CoreWeave also lists H100 inference at $6.16 per GPU-hour in North America. Although that number matches the rounded on-demand normalization above, it describes an inference offering rather than the eight-GPU HGX instance. Ask which serving configuration, minimum allocation, storage and networking charges apply before treating the two figures as interchangeable.
How much is a Lambda H100 per hour?
Lambda publishes H100 SXM prices by plan size. Larger plans have a lower per-GPU rate:
| Plan | Rate per GPU-hour | Hourly GPU total | Difference from the one-GPU rate |
|---|---|---|---|
| 1 GPU | $4.29 | $4.29 | Baseline |
| 2 GPUs | $4.19 | $8.38 | $0.10 less per GPU-hour |
| 4 GPUs | $4.09 | $16.36 | $0.20 less per GPU-hour |
| 8 GPUs | $3.99 | $31.92 | $0.30 less per GPU-hour |
For a single H100, the published Lambda rate is $4.29 per GPU-hour. For a distributed eight-GPU job, the listed plan rate falls to $3.99 per GPU-hour. Lambda also advertises no egress fees, but you should still check storage, attached services and any current plan restrictions when estimating the full bill.
Rank #2
- NVIDIA Ampere Architecture-based CUDA Cores - Double-speed processing for single-precision floating point (FP32) operations and improved power efficiency provide significant performance improvements for graphics and simulation workflows, such as complex 3D computer-aided design (CAD) and computer-aided engineering (CAE), on the desktop.
- Second-Generation RT Cores - With up to 2X the throughput over the previous generation and the ability to concurrently run ray tracing with either shading or denoising capabilities, second-generation RT Cores deliver massive speedups for workloads like photorealistic rendering of movie content, architectural design evaluations, and virtual prototyping of product designs. This technology also speeds up the rendering of ray-traced motion blur for faster results with greater visual accuracy.
- Third-Generation Tensor Cores - New Tensor Float 32 (TF32) precision provides up to 5X the training throughput over the previous generation to accelerate AI and data science model training without requiring any code changes. Hardware support for structural sparsity doubles the throughput for inferencing. Tensor Cores also bring AI to graphics with capabilities like DLSS, AI denoising, and enhanced editing for select applications.
- Third-Generation NVIDIA NVLink - Increased GPU-to-GPU interconnect bandwidth provides a single scalable memory to accelerate graphics and compute workloads and tackle larger datasets.
- 48 Gigabytes (GB) of GPU Memory - Ultra-fast GDDR6 memory, scalable up to 96 GB with NVLink, gives data scientists, engineers, and creative professionals the large memory necessary to work with massive datasets and workloads like data science and simulation.
What does an AWS GPU instance cost?
AWS does not provide one universal “GPU-hour” price. An EC2 quote depends on the exact GPU instance family and size, AWS region, operating system, tenancy, purchase option and billing duration. AWS states that On-Demand Instances let customers pay by the hour or second, with a 60-second minimum; qualifying Linux and other operating-system partial hours are billed per second. AWS expresses the charge per instance-hour, not as a universal per-GPU rate.
Use the AWS calculator for a defensible comparison
- Choose the exact EC2 GPU instance and note the number and model of GPUs, memory and any attached CPU or networking capacity.
- Select the same region used for the CoreWeave or Lambda quote. AWS pricing is region-specific.
- Choose the operating system and tenancy. Linux and other qualifying configurations can use per-second billing after the 60-second minimum.
- Select On-Demand, Spot, Reserved or another commitment option. Do not compare an AWS commitment price with an on-demand provider price without labeling the distinction.
- Enter expected runtime, data transfer, block storage, snapshots and other services in the AWS Pricing Calculator.
- Record the resulting instance-hour price and divide by the number of GPUs only after confirming that the GPUs are equivalent in model and memory.
Without those inputs, quoting an AWS number would imply a precision the published information does not support. An AWS instance with a different H100 configuration, region or operating system can produce a materially different total.
Normalize the prices before choosing a provider
Match GPU model and memory
Start with the same accelerator generation and memory capacity. An H100 SXM plan and an HGX H100 system may both be described as H100 offerings while differing in interconnect, host platform or allocation model. If your workload requires H100 80 GB, record that requirement explicitly and reject prices for a different memory tier.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Convert instance totals to GPU-hours, but retain instance size
Use this formula:
GPU-hour rate = instance-hour rate ÷ number of GPUs in the instance
Rank #3
- NVIDIA Volta GV100 Architecture — 4,608 CUDA Cores, 640 1st-Gen Tensor Cores delivering 14 TFLOPS FP32 and 112 TFLOPS deep learning performance for AI training, inference, HPC, and scientific computing workloads
- 32GB HBM2 ECC Memory — 900 GB/s Bandwidth — High-bandwidth memory on a 4096-bit bus with ECC error correction provides the memory capacity and throughput required for the largest AI models, simulations, and datasets
- PCIe 3.0 x16 Interface — 250W TDP — Standard PCIe Gen3 connectivity with passive cooling designed for enterprise rack server deployment in HPE ProLiant, Dell PowerEdge, and Supermicro platforms with adequate chassis airflow
- NVLink — Scale to 96GB Unified Memory — Connect two V100 GPUs via NVLink at 300 GB/s bi-directional bandwidth to scale GPU memory from 32GB to 96GB for larger AI training and HPC workloads
- Multi-Precision Computing — Supports FP64 (7 TFLOPS), FP32 (14 TFLOPS), FP16 (112 TFLOPS) and INT8 precision modes for flexible deployment across training, inference, and scientific simulation workloads
For CoreWeave, $49.24 ÷ 8 = $6.155, rounded to $6.16. For the spot listing, $19.71 ÷ 8 = about $2.46. Keep the original eight-GPU requirement beside the normalized figure; a distributed job may need all eight GPUs and cannot necessarily scale down to one.
Separate purchase modes
Maintain separate rows for on-demand, spot, reserved or committed capacity and inference. Spot is not a substitute for guaranteed capacity, and an inference rate may include a different service design from a training instance. A lower number is meaningful only when the allocation and reliability match your workload.
Record geography
CoreWeave publishes separate North America and Europe tables, and AWS prices vary by region. Lambda’s displayed plan should also be checked for the location available to your account. Put the region in every cost worksheet and quote date; otherwise a later comparison can mix unlike prices.
Illustrative monthly costs
The following arithmetic uses a 730-hour month at continuous utilization. It is an illustration, not a provider invoice, and excludes storage, networking, taxes, idle time and other services.
Rank #4
- Discrete graphics card memory 40 GB
- Memory bandwidth (max) 1555 GB/s
- Graphics processor family NVIDIA
- Graphics processor A100
| Configuration | Hourly calculation | 730-hour illustration |
|---|---|---|
| CoreWeave eight-GPU on-demand | $49.24 × 730 | $35,945.20 |
| CoreWeave eight-GPU spot | $19.71 × 730 | $14,388.30 |
| Lambda eight-GPU H100 plan | $3.99 × 8 × 730 | $23,301.60 |
| Lambda one-GPU H100 plan | $4.29 × 730 | $3,131.70 |
| Lambda eight-GPU A100 plan | $2.79 × 8 × 730 | $16,287.60 |
If your GPUs run for only half the month, halve the compute component; if jobs leave instances running while idle, billed hours remain unchanged. For spot, model interruption and restart overhead rather than assuming 730 successful hours.
Costs beyond the GPU-hour
- Storage: Include boot volumes, datasets, checkpoints, snapshots and retained artifacts. Training jobs that repeatedly write checkpoints can accumulate storage even when GPU use is unchanged.
- Networking and egress: Lambda explicitly advertises no egress fees. AWS and CoreWeave totals can include transfer charges depending on architecture, destination and service combination. Verify the current rate card for the selected region.
- Host and orchestration capacity: CPU, memory, local NVMe, Kubernetes or other control-plane resources may be bundled, metered separately or required to keep GPUs busy.
- Taxes and currency: Published list prices may exclude taxes and may be displayed in a particular currency. Apply the billing entity’s tax treatment rather than assuming the table total is final.
- Utilization: A $4 GPU-hour is not economical if data preparation, queueing or synchronization leaves the accelerator idle. Measure active GPU time and include warm-up and teardown hours.
Which price is best for common workloads?
Single-GPU development and testing
Lambda’s published one-GPU H100 rate is $4.29 per GPU-hour, making it the directly stated single-GPU option in this comparison. CoreWeave’s $6.16 figure is labeled as an inference GPU-hour, while its $49.24 listing is an eight-GPU instance. Confirm that the provider can supply the exact single-GPU development environment you need before comparing totals.
Eight-GPU distributed training
Lambda’s eight-GPU H100 plan totals $31.92 per hour at the published rate, versus CoreWeave’s $49.24 on-demand eight-GPU HGX listing. Choose based on more than price: verify interconnect performance, region, image availability, checkpoint storage and scheduling guarantees for your training framework.
Recommended Free Tools
Interruptible batch work
CoreWeave’s $19.71 eight-GPU spot listing is substantially below its on-demand rate. It can suit checkpointed batch jobs whose deadline tolerates restarts. Price the engineering time and extra runs required by interruptions, and keep an on-demand fallback if missing a deadline costs more than the hourly saving.
Production inference
CoreWeave publishes a distinct $6.16 per H100 GPU-hour inference figure. Compare the complete serving design: replicas, request volume, autoscaling behavior, model loading time, storage and egress. A raw GPU-hour comparison does not reveal cost per request or latency at your target traffic level.
A practical quote checklist
- GPU model, memory and interconnect recorded.
- Number of GPUs required simultaneously recorded.
- Region and quote date recorded.
- On-demand, spot or commitment status labeled.
- Operating system and instance family specified for AWS.
- Expected monthly GPU hours and utilization percentage estimated.
- Storage, transfer, taxes and ancillary services included.
- Interruption, checkpoint and recovery assumptions documented for spot.
- Inference measurements expressed as cost per request or token as well as GPU-hour.
Bottom line on CoreWeave vs. AWS vs. Lambda
For the published H100 list prices available on October 18, 2026, Lambda is the lower-priced eight-GPU on-demand option at an equivalent $3.99 per GPU-hour, compared with CoreWeave’s approximately $6.16 per GPU-hour from its $49.24 eight-GPU HGX instance. CoreWeave’s spot listing lowers the arithmetic to about $2.46 per GPU-hour, but it is a different, interruption-sensitive purchase mode. AWS may be cheaper or more expensive depending on the exact EC2 GPU instance, region, operating system and purchase option; use those inputs in the AWS Pricing Calculator before drawing a conclusion.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
