Science and Technology

05/26/26
Cloud providers are selling hourly access to NVIDIA's B200 Blackwell GPU, with prices ranging from roughly $2.14/hr on budget platforms to $14.24/hr on hyperscalers.
The market includes neo-cloud specialists like Verda, Lambda Labs, and Runpod competing against hyperscalers AWS and GCP, which charge significantly higher rates.
As of mid-July 2026, B200 adoption is still in early stages with limited inventory, causing prices to vary wildly by provider tier and availability.
Reaching above $7.37/hr would signal that enterprise demand or supply constraints are pushing B200 rates into the premium tier currently dominated by AWS, indicating strong scarcity value.
Note: The resolution likely depends on whether the metric measures the highest available premium rate (like AWS) or the median market rate; current data suggests the median is well below the threshold while the premium ceiling is well above it.
Lambda Labs lists the lowest verified on-demand rate at $4.99/hr, while AWS p6-b200 charges $14.24/hr, representing a 2.9× price spread.
smartcloudsuites.comspheron.networkOn-demand B200 capacity ranges from $5.89 to $8.60 per GPU-hour across three sourced providers, with availability remaining limited.
computetape.comThe current cheapest on-demand B200 rate is $2.14/hr on Verda, while the market median across 20 configurations is $6.11/hr.
computeprices.comgpus.ioAI-generated briefing. AI can make mistakes. This is not financial advice.
AWS p6-b200 on-demand pricing is already at $14.24/hr, nearly double the $7.37 threshold, suggesting that if the market aggregates premium hyperscaler rates, the average could exceed the target.
spheron.networksmartcloudsuites.comOn-demand pricing increased by approximately 6% from June 2025 to mid-2026, rising from $5.84/hr to $6.22/hr, indicating upward momentum in compute costs as demand grows.
getdeploying.comEarly OEM quotes place the B200 192GB SXM hardware cost at $45,000–$50,000, a significant increase over prior generations that supports higher hourly rental rates to cover capital expenditure.
northflank.comAI-generated briefing. AI can make mistakes. This is not financial advice.
The current market median for B200 is $6.11/hr, which is $1.26 below the $7.37 threshold, and the cheapest verified on-demand rate is $2.14/hr, creating a large buffer against the target.
gpus.iocomputeprices.comPricing is projected to stabilize around $2.50–$3.00/hr at major providers by Q4 2026 as TSMC ramps Blackwell production and supply enters the market, moving further away from the $7.37 target.
inworld.aiMost accessible neo-cloud providers (Lambda at $4.99/hr, Runpod at $4.99/hr, Spheron at $3.70/hr) are priced well below $7.37, keeping the weighted average for general users low.
AI-generated briefing. AI can make mistakes. This is not financial advice.