AIForge Works
AIForge Works
← Back to Blog
AIForge Works Research
August 31, 2026

Weekly Pricing Pulse

By Bobby Clay, Founder, AIForge Works

Pricing evidence through August 30, 2026

9 providers harvested. 10 New Silicon catalog entries, 4 restored / expanded visibility entries, 7 Deprecated SKUs, and 0 on-demand per-GPU pricing observations over 1%.

The Pulse weekly cloud pricing graphic
Get the Pulse
Weekly GPU pricing by email

One email after the Monday harvest review: material price moves, catalog churn, OCPI context, and buyer notes.

Prefer LinkedIn? Subscribe to the LinkedIn edition.Subscribe on LinkedIn

About Pulse

The Weekly Pricing Pulse tracks what changed in published GPU cloud pricing across AWS, Azure, GCP, OCI, CoreWeave, Lambda, Vultr, Nebius, and Crusoe across the latest verified comparison period. AIForge Works compares each new harvest with the prior verified harvest and highlights buyer-relevant movement: comparable on-demand price changes, New Silicon, restored / expanded visibility, Deprecated SKUs, and likely price/SKU rolls.

This is not a live quote, workload-specific recommendation, or personalized account analysis. For workload-specific recommendations, use Cloud Advisor. To inspect neighboring SKUs, regions, or commitment terms, open the Multi-Cloud Pricing Analyzer. To compare your own GPU spend against a comparable floor, use GPU Audit.

Editor's note

Published prices were effectively flat this week. The market was not. Across the nine clouds we track, no comparable normalized public on-demand GPU rate cleared our 1% reporting threshold. Yet the transaction-based Ornn × ICE OCPI moved across all four tracked GPU families: B200 fell 8.9% to $6.05 while its lowest reviewed provider floor held at $6.69; H100 SXM rose 4.9% to $3.01 while its floor held at $3.85.

The catalog moved, too. Ten AWS B300 observations became newly visible: the same eight-GPU instance configuration starts at $25.10 per GPU-hour in Hyderabad and $35.14 in São Paulo, about 40% higher in São Paulo. Four observations returned across Lambda and Vultr, while seven others were no longer visible.

Treat that as catalog visibility, not capacity. Newly listed does not mean orderable, and a returning SKU is not a launch. Check the GPU, region, and terms your workload needs; the catalog starts the conversation, but it is not the quote.

Ornn OCPI context

The public catalog was quiet, but the broader GPU pricing benchmark moved. Over the seven days to August 30, H100 SXM and H200 rose, while B200 and A100 SXM4 declined. B200 had the largest move of these four benchmarks, down 8.9%.

OCPI, the Ornn Compute Price Index, is a transaction-based benchmark that is separate from the public on-demand catalog prices tracked elsewhere in this Pulse. These are the latest settled daily readings through August 30, 2026 at 20:00 UTC, retrieved from Ornn on August 31. Changes below compare August 23 with August 30; the catalog comparison in this issue uses the separate August 24 and August 30 harvests. Ornn methodology and API documentation.

GPU familyLatest OCPIChange vs. August 23
H100 SXM$3.01 / GPU-hour+4.9%
H200$4.50 / GPU-hour+1.8%
B200$6.05 / GPU-hour-8.9%
A100 SXM4$1.03 / GPU-hour-3.7%

For every provider row, the displayed value is the lowest eligible reviewed public offer found for that exact OCPI GPU cohort across the regions covered by the harvest. It is not a floor across GPU families, and it is not an availability guarantee. The corrected comparison includes AWS and Azure where exact-cohort evidence qualifies. Azure is absent from B200 because the reviewed Azure evidence is GB200, not an exact B200 match. The AWS H100 row carries a source-topology caveat because the provider source identifies H100 80 GB hardware but does not literally state the SXM form factor.

The review has eligible provider floors from eight providers for H100 SXM, seven for H200, five for B200, and seven for A100 SXM4. Missing evidence does not mean a provider lacks that GPU. Select a GPU to compare provider floors, then use the paired graphs to inspect the dated market trend.

Ornn OCPI context

OCPI compared to Exact-GPU Provider Floors

Public catalog snapshot · August 30, 2026
Exact GPU cohort comparison
H100 SXM
Each provider row is its lowest eligible reviewed public offer for this exact OCPI GPU cohort, never a floor across GPU families. This is not an exhaustive search of each provider's catalog.
OCPI-H100 SXM
Transaction-based - Ornn x ICE - latest August 30, 2026 20:00 UTC
$3.01/ GPU / hr
Lowest reviewed provider floor for this exact GPU cohort
Nebius - eu-north1
$3.85/ GPU / hr
Spread to OCPI
$0.84 28% above
Providers with eligible exact-GPU offers 8 / 9
Hyperscalers: AWS $6.88 to Azure $12.29 (+128.6% to +308.3% vs OCPI)
Neoclouds: Nebius $3.85 to CoreWeave $6.16 (+27.9% to +104.5% vs OCPI)
Provider floors for this exact GPU cohort
Each row is the provider's lowest eligible reviewed offer - normalized per GPU/hr
Hyperscalers
4/4 matched
AWSHyperscalers
$6.88/GPU/hr
+$3.87 / +128.6%
p5.48xlarge-8 GPUs
US East (N. Virginia)
OCIHyperscalers
$10.00/GPU/hr
+$6.99 / +232.2%
B98415-8 GPUs
ap-hyderabad-1
GCPHyperscalers
$11.06/GPU/hr
+$8.05 / +267.5%
a3-highgpu-8g-8 GPUs
us-west1
AzureHyperscalers
$12.29/GPU/hr
+$9.28 / +308.3%
Standard_ND96isr_H100_v5-8 GPUs
westus3
Neoclouds
4/5 matched
NebiusNeoclouds
$3.85/GPU/hr
+$0.84 / +27.9%
1gpu-16vcpu-200gb-1 GPU
eu-north1
CrusoeNeoclouds
$3.90/GPU/hr
+$0.89 / +29.6%
h100-80gb-sxm-ib.8x-8 GPUs
Europe (Iceland)
LambdaNeoclouds
$3.99/GPU/hr
+$0.98 / +32.6%
gpu_8x_h100_sxm5-8 GPUs
Region not published
CoreWeaveNeoclouds
$6.16/GPU/hr
+$3.15 / +104.5%
hgx-h100-8 GPUs
global
VultrNeoclouds
N/A
No exact-GPU match
No eligible reviewed offer for this exact OCPI GPU cohort.-Coverage gap
N/A
Methodology: This comparison covers publicly listed whole-GPU H100 SXM offers that match the OCPI family. For each provider, the displayed floor is the lowest eligible offer found for this exact H100 SXM cohort; it is not a floor across GPU families. It excludes fractional vGPU slices, private quotes, negotiated discounts, taxes, provider-specific fees, zero-priced listings, spot/preemptible pricing, and reserved or committed terms. For multi-GPU instances, AIForge Works normalizes the listed hourly instance price to dollars per GPU-hour by dividing by GPU count.
What each provider floor means: Each row is that provider's lowest eligible reviewed public offer for the exact OCPI GPU cohort shown. It is never a floor across GPU families, an exhaustive market floor, or a region-matched comparison to OCPI. Other products or regions may have lower prices.
Sources and limits: This Pulse uses the August 30, 2026 harvest for public catalog floors and the Ornn x ICE OCPI reading for August 30, 2026 at 20:00 UTC for every published GPU family. OCPI is a transaction-based Ornn x ICE index; AIForge Works floors are independently indexed public catalog offers. This page is informational only: it is not a live quote, availability guarantee, purchase recommendation, or procurement, legal, financial, or technical advice. Verify current provider availability, terms, fees, taxes, discounts, and contract pricing directly with the provider before acting. Full OCPI methodology lives at index.ornn.com.

OCPI vs Exact-GPU Provider Floors by Week

Transaction-index prices vs exact-GPU provider floors from dated public catalog comparisons.

Each latest offer point is the provider's lowest eligible reviewed offer for that exact OCPI GPU cohort, never a floor across GPU families. Historical points retain their frozen comparison basis.

OCPI index Catalog floor
OCPI index Catalog floor

H100 SXM

OCPI
$3.01
Provider floor
$3.85(Nebius)
$0.84 above
$4.13$2.035/18Nebius5/18 catalog floor: $2.9500 from Nebius5/18 OCPI: $3.1500$3.15$2.955/255/25 catalog floor: $2.9500 from Nebius5/25 OCPI: $2.85006/16/1 catalog floor: $2.9500 from Nebius6/1 OCPI: $2.7700$2.776/8Nebius6/8 catalog floor: $3.8500 from Nebius6/8 OCPI: $3.1000$3.856/156/15 catalog floor: $3.8500 from Nebius6/15 OCPI: $2.4800$2.486/226/22 catalog floor: $3.8500 from Nebius6/22 OCPI: $2.31006/296/29 catalog floor: $3.8500 from Nebius6/29 OCPI: $2.4500$2.457/67/6 catalog floor: $3.8500 from Nebius7/6 OCPI: $2.57007/137/13 catalog floor: $3.8500 from Nebius7/13 OCPI: $2.5500$2.557/207/20 catalog floor: $3.8500 from Nebius7/20 OCPI: $2.71007/277/27 catalog floor: $3.8500 from Nebius7/27 OCPI: $2.6000$2.608/98/9 catalog floor: $3.8500 from Nebius8/9 OCPI: $2.53008/178/17 catalog floor: $3.8500 from Nebius8/17 OCPI: $2.6400$2.648/248/24 catalog floor: $3.8500 from Nebius8/24 OCPI: $2.8700$2.878/30Nebius8/30 exact-GPU provider floor: $3.8500 from Nebius8/30 OCPI: $3.0100$3.01$3.85
OCPI index Catalog floor

H200

OCPI
$4.50
Provider floor
$4.29(Crusoe)
$0.21 below
$6.13$3.035/18Nebius5/18 catalog floor: $3.5000 from Nebius5/18 OCPI: $5.7200$5.72$3.505/255/25 catalog floor: $3.5000 from Nebius5/25 OCPI: $4.11006/16/1 catalog floor: $3.5000 from Nebius6/1 OCPI: $4.6600$4.666/8Crusoe6/8 catalog floor: $4.2900 from Crusoe6/8 OCPI: $4.6000$4.296/156/15 catalog floor: $4.2900 from Crusoe6/15 OCPI: $3.5900$3.596/226/22 catalog floor: $4.2900 from Crusoe6/22 OCPI: $3.44006/296/29 catalog floor: $4.2900 from Crusoe6/29 OCPI: $3.6100$3.617/67/6 catalog floor: $4.2900 from Crusoe7/6 OCPI: $4.03007/137/13 catalog floor: $4.2900 from Crusoe7/13 OCPI: $4.3900$4.397/207/20 catalog floor: $4.2900 from Crusoe7/20 OCPI: $4.90007/277/27 catalog floor: $4.2900 from Crusoe7/27 OCPI: $4.8600$4.868/98/9 catalog floor: $4.2900 from Crusoe8/9 OCPI: $4.67008/178/17 catalog floor: $4.2900 from Crusoe8/17 OCPI: $4.7500$4.758/248/24 catalog floor: $4.2900 from Crusoe8/24 OCPI: $4.4200$4.428/30Crusoe8/30 exact-GPU provider floor: $4.2900 from Crusoe8/30 OCPI: $4.5000$4.50$4.29
OCPI index Catalog floor

B200

OCPI
$6.05
Provider floor
$6.69(Lambda)
$0.64 above
$7.65$3.785/18Nebius5/18 catalog floor: $5.5000 from Nebius5/18 OCPI: $5.8200$5.82$5.505/255/25 catalog floor: $5.5000 from Nebius5/25 OCPI: $5.34006/16/1 catalog floor: $5.5000 from Nebius6/1 OCPI: $5.9900$5.996/8Lambda6/8 catalog floor: $6.6900 from Lambda6/8 OCPI: $4.6600$6.696/156/15 catalog floor: $6.6900 from Lambda6/15 OCPI: $4.5000$4.506/226/22 catalog floor: $6.6900 from Lambda6/22 OCPI: $4.29006/296/29 catalog floor: $6.6900 from Lambda6/29 OCPI: $4.4100$4.417/67/6 catalog floor: $6.6900 from Lambda7/6 OCPI: $5.14007/137/13 catalog floor: $6.6900 from Lambda7/13 OCPI: $6.7500$6.757/207/20 catalog floor: $6.6900 from Lambda7/20 OCPI: $7.14007/277/27 catalog floor: $6.6900 from Lambda7/27 OCPI: $6.5800$6.588/98/9 catalog floor: $6.6900 from Lambda8/9 OCPI: $5.70008/178/17 catalog floor: $6.6900 from Lambda8/17 OCPI: $5.9300$5.938/248/24 catalog floor: $6.6900 from Lambda8/24 OCPI: $6.6400$6.648/30Lambda8/30 exact-GPU provider floor: $6.6900 from Lambda8/30 OCPI: $6.0500$6.05$6.69
OCPI index Catalog floor

A100 SXM4

OCPI
$1.03
Provider floor
$2.30(Crusoe)
$1.27 above
$2.53$0.785/18Crusoe5/18 catalog floor: $1.9500 from Crusoe5/18 OCPI: $1.2400$1.24$1.955/255/25 catalog floor: $1.9500 from Crusoe5/25 OCPI: $1.26006/16/1 catalog floor: $1.9500 from Crusoe6/1 OCPI: $1.3000$1.306/86/8 catalog floor: $1.9500 from Crusoe6/8 OCPI: $1.07006/15CoreWeave6/15 catalog floor: $2.2100 from CoreWeave6/15 OCPI: $1.0500$1.05$2.216/226/22 catalog floor: $2.2100 from CoreWeave6/22 OCPI: $1.02006/29Crusoe6/29 catalog floor: $2.3000 from Crusoe6/29 OCPI: $1.0200$1.02$2.307/67/6 catalog floor: $2.3000 from Crusoe7/6 OCPI: $1.01007/137/13 catalog floor: $2.3000 from Crusoe7/13 OCPI: $1.0400$1.047/207/20 catalog floor: $2.3000 from Crusoe7/20 OCPI: $1.01007/277/27 catalog floor: $2.3000 from Crusoe7/27 OCPI: $1.0500$1.058/98/9 catalog floor: $2.3000 from Crusoe8/9 OCPI: $1.06008/178/17 catalog floor: $2.3000 from Crusoe8/17 OCPI: $1.0200$1.028/248/24 catalog floor: $2.3000 from Crusoe8/24 OCPI: $1.0700$1.078/30Crusoe8/30 exact-GPU provider floor: $2.3000 from Crusoe8/30 OCPI: $1.0300$1.03$2.30

August 3 is omitted from the paired trend because that Harvest candidate did not reach acceptance.

Week at a glance

No provider had comparable on-demand offers over the 1% movement threshold. AWS, Lambda, and Vultr changed catalog visibility through New Silicon, restored / expanded visibility, and Deprecated SKU entries, shown below.

These movement counts apply only to comparable public-catalog on-demand prices. OCPI is a separate transaction-based market index and is shown independently.

  • On-demand pricing was quiet this week. We did not find any public on-demand GPU price changes above the 1% reporting threshold; the visible movement is catalog availability.

Provider roll-up

ProviderClassNewRestored / expandedGone>1%Rolls
AWShyperscaler100000
Lambdaneocloud02600
Vultrneocloud02100
Totals-104700

No buyer-visible GPU catalog additions, removals, or public on-demand price moves above 1% were found this week for Azure, GCP, OCI, CoreWeave, Crusoe, and Nebius.

Offers seen in earlier verified harvests returned to view this week: Lambda (2) and Vultr (2). Because these provider and SKU combinations appeared in earlier verified harvests, we treat them as restored catalog visibility rather than brand-new silicon launches.

Catalog changes this week

Catalog change counts summarize week-over-week visibility changes in the compared harvests. They show provider + region + instance/SKU observations that became newly visible or no longer visible between harvests. We do not describe these as confirmed capacity launches or retirements unless provider-published sources say so.

New Silicon

10 New Silicon GPU catalog entries were detected across AWS (10). Entries are grouped by provider so high-volume catalog churn does not hide smaller-provider changes.

New Silicon means a provider + region + instance/SKU observation newly visible in this harvest that is not paired with a same-provider, same-geo, same-instance deprecated SKU. It still does not by itself prove net-new capacity or a brand-new accelerator family.

AWS

10 New Silicon entries / 2 displayed groups
  • New Silicon: AWS p6-b300.48xlarge in Asia Pacific (Hyderabad). B300 288 GB, 8-GPU pack. 5 price variants: instance $201-$205/hr ($25.1-$25.58/hr per GPU).
  • New Silicon: AWS p6-b300.48xlarge in South America (Sao Paulo). B300 288 GB, 8-GPU pack. 5 price variants: instance $281-$285/hr ($35.14-$35.62/hr per GPU).

Restored / expanded visibility

4 previously seen public GPU catalog observations were visible again this week across Lambda (2), Vultr (2).

Each provider and SKU had appeared in an earlier verified comparison; the region shown is where the offer was visible this week. We classify these as restored or expanded catalog visibility, not a new hardware launch, proof of added capacity, or confirmation that the offer is currently orderable.

Lambda

2 Restored / expanded visibility entries
  • Restored / expanded visibility: Lambda gpu_2x_h100_sxm5 in Region not published. H100 80 GB, 2-GPU pack. On-demand instance price: $8.3800/hr ($4.1900/hr per GPU).
  • Restored / expanded visibility: Lambda gpu_8x_a100_80gb_sxm4 in Region not published. A100 80 GB, 8-GPU pack. On-demand instance price: $22.32/hr ($2.7900/hr per GPU).

Vultr

2 Restored / expanded visibility entries
  • Restored / expanded visibility: Vultr vcg-a16-12c-128g-32vram in Atlanta, US. NVIDIA A16 16 GB, 2-GPU pack. On-demand instance price: $0.9420/hr ($0.4710/hr per GPU).
  • Restored / expanded visibility: Vultr vcg-a16-12c-128g-32vram in Silicon Valley, US. NVIDIA A16 16 GB, 2-GPU pack. On-demand instance price: $0.9420/hr ($0.4710/hr per GPU).

Deprecated SKUs

7 Deprecated SKU GPU catalog entries were detected across Lambda (6), Vultr (1). Entries are grouped by provider so high-volume catalog churn does not hide smaller-provider changes.

Deprecated SKU means the exact provider catalog SKU was present in the prior harvest but absent from this one. It is no longer visible as that exact SKU in this week's public-catalog comparison, even when a same-instance replacement looks likely; this does not prove the SKU is unavailable to buy.

Lambda

6 previously listed offers not present this week
  • Previously listed GPU offer: Lambda gpu_8x_a100 from California, USA. Last seen on-demand: $15.92/hr (per-GPU: $1.9900/hr).
  • Previously listed GPU offer: Lambda gpu_2x_h100_sxm5 from Central Texas, USA. Last seen on-demand: $8.3800/hr (per-GPU: $4.1900/hr).
  • Previously listed GPU offer: Lambda gpu_8x_a100 from Germany. Last seen on-demand: $15.92/hr (per-GPU: $1.9900/hr).
  • Previously listed GPU offer: Lambda gpu_8x_a100 from Israel. Last seen on-demand: $15.92/hr (per-GPU: $1.9900/hr).
  • Previously listed GPU offer: Lambda gpu_8x_a100 from Virginia, USA. Last seen on-demand: $15.92/hr (per-GPU: $1.9900/hr).
  • Previously listed GPU offer: Lambda gpu_8x_a100_80gb_sxm4 from Virginia, USA. Last seen on-demand: $22.32/hr (per-GPU: $2.7900/hr).

Vultr

1 previously listed offers not present this week
  • Previously listed GPU offer: Vultr vcg-a16-12c-128g-32vram from Singapore, SG. Last seen on-demand: $0.9420/hr (per-GPU: $0.4710/hr).

Top on-demand movers by provider

No comparable offers met the 1% movement threshold for this section.

Up to 5 price observations per provider, ranked by confidence (high-confidence first), then by percentage change.

Methodology

Evidence through. Pricing evidence in this issue runs through August 30, 2026. This evidence-through date is recorded separately from the publication date shown in the article header.

What we compared. This Pulse compares the prior verified harvest (August 24, 2026) against the current harvest (August 30, 2026). The comparison spans 6 days; it is not described as a seven-day week.

How we compare prices fairly. Different providers package GPUs differently: some sell single-GPU instances, others sell 8-GPU instances. To compare apples-to-apples, we divide every instance price by the number of GPUs on that instance. All movement percentages in this Pulse are based on that per-GPU hourly rate, not the raw instance price.

What counts as a real move. We only flag price changes greater than 1%. Anything smaller is treated as rounding noise and excluded. Providers with zero comparable GPU offers in either week are excluded from GPU price-movement conclusions. New Silicon means a provider + region + instance/SKU observation newly visible in this harvest that is not paired with a same-provider, same-geo, same-instance deprecated SKU and has not been seen in the lookback window. Restored / expanded visibility means the provider and instance/SKU appeared in an earlier verified harvest and is visible again, sometimes in additional regions. Deprecated SKU means the exact provider catalog SKU was present in the prior harvest but absent from this one.

Scope and limits. The generated Pulse diff certifies on-demand movement only. All prices are in US dollars. Catalog entries missing a clean price or verified GPU specs are dropped before comparison. This Pulse compiles publicly published prices and is not a live quote, availability guarantee, or workload-specific advice.

This week's coverage. The August 30, 2026 harvest completed across 9 providers. After normalizing 6,305,742 raw catalog price observations against verified hardware specs, this Pulse identified 13,865 comparable public on-demand GPU observations for week-over-week movement analysis. Of those, 13,850 exact provider + region + instance/SKU observations were present in both weeks and eligible for direct movement comparison.

Next Pulse: Week of 2026-09-07.