Compute Exchange launches a secondary GPU Marketplace
Compute Exchange
Sign Up
Reserved GPU Rental/RESERVED H200 PCIe

RESERVED H200 PCIe

NVIDIA H200 NVL 141GB (PCIe)

The PCIe form of the H200 (NVIDIA H200 NVL) delivers the same 141GB of HBM3e in air-cooled PCIe hosts — the memory headroom of H200 without SXM infrastructure.

AVAILABLE TERM LENGTH
1MO3MO6MO12MO24MO36MO

Broadly available across mid-tier providers. Shorter commits provision faster than SXM, and supply runs looser than H200 SXM.

Reserve CapacityCompare Providers
TECHNICAL SPECIFICATIONS
RESERVED
VRAM
141 GB HBM3e
MEMORY BANDWIDTH
4.8 TB/s
FP 16 TENSOR
1,671 TFLOPS (sparse)
FP 8 TENSOR
3,341 TFLOPS (sparse)
TDP
Up to 600W
FORM FACTOR
PCIe Gen5 (dual-slot)
INTERCONNECT
NVLink bridge (900 GB/s) / PCIe 5.0
ARCHITECTURE
Hopper
Partner Network

AGGREGATED ACROSS
LEADING NEOCLOUDS

Compute Exchange aggregates reserved capacity from a verified network of leading AI-native cloud providers and hyperscalers. All partners undergo identity, capacity, SLA, and operational verification before quotes surface on the network.

You receive a normalized comparison across providers in a single quote response — rather than evaluating each neocloud's contract structure, billing model, and SLA terms in isolation.

WORKLOAD FIT

RESERVED H200 PCIe
USE CASES

Large-context inference
RAG pipelines
Air-cooled data centers
Memory-bound fine-tuning
WHY RESERVE

RESERVED H200 PCIe
VS ON-DEMAND

H200 NVL rides the same HBM3e supply constraints as the SXM part. Reserving locks both capacity and unit economics for inference fleets that need the 141GB footprint without rebuilding around SXM.

Frequently Asked Questions

RESERVED H200 PCIe
KEY QUESTIONS

Is H200 PCIe the same silicon as H200 SXM?+
Same GPU and the same 141GB of HBM3e at 4.8TB/s — NVIDIA ships it as the H200 NVL, a dual-slot PCIe Gen5 card with NVLink bridging. Peak tensor throughput runs slightly below SXM and power tops out around 600W, in exchange for deployment in standard air-cooled PCIe hosts.
How does supply compare to H200 SXM?+
Looser. The PCIe form deploys across a wider provider base, so short-term reservations quote and provision faster than SXM. HBM3e remains the binding constraint for both variants, though — during allocation crunches, both tighten together, and reserved capacity holds priority.
What workloads fit H200 PCIe reservations best?+
Large-context inference and RAG pipelines in air-cooled facilities — anywhere the 141GB footprint matters but SXM infrastructure is not on the table. For multi-node training at scale, the SXM variant's NVSwitch fabric is the better reservation.
Ready to Reserve?

LIVE QUOTE FOR
RESERVED H200 PCIe

Compute Exchange returns indicative pricing within 24 hours, anchored to your specific quantity, region, and condition. We do not publish active counterparty listings.

Request a Quote