Hyperscalers List GPUs at Nearly 2× the GPU-Cloud Price, Fluence Research Finds

요약:The AI Compute Race report prices GPU compute outside the biggest buyers: a three-year commitment cuts the median price by 48%, while 17 of 22 tracked GPU models rose in price over the past year

The AI Compute Racereport prices GPU compute outside the biggest buyers: a three-year commitment cuts the median price by 48%, while 17 of 22 tracked GPU models rose in price over the past year

Fluence has published The AI Compute Race, a report that prices GPU cloud compute for every company that is not a frontier lab or a hyperscaler. It was produced with research partners NEAR AI, io.net, Boundless, Phala, Gensyn, Fortytwo, Theta, Compute Labs and GAIB. Its central finding: where a company buys a GPU-hour sets the price.

Across H100, H200 and B200 models, hyperscalers list GPUs at 1.98 times the GPU-cloud median, from 1.4 times on the B200 to 2.8 times on the H100. The same H200 lists from $4.08 to $10.60 an hour.

Commitment length matters almost as much as provider choice. Against each providers own on-demand rate, a one-year reservation cuts the median price by 21% and a three-year term by 48%. Spot capacity lists at 53% of on-demand, for workloads that can be interrupted.

Buyers who expect GPU prices to fall may need to recheck their budgets. Seventeen of the 22 GPU models tracked by GetDeploying cost more than a year earlier, and the median H100 rents are 13% to 30% more, depending on the window measured.

The report also traces why prices are rapidly rising:

  • Demand moved to open models.Chinese models share of OpenRouter tokens among companies in the US, Europe and the Global South rose from 6–13% to 57–67% in under eight months, according to CNBC. Open weights let any company with GPUs serve a near-frontier model on hardware it controls.
  • The biggest buyers locked up the newest supply.The four largest hyperscalers guide $720–740 billion of 2026 capital spending, and the four largest disclosed deals by AI labs and Meta with neoclouds total about $233 billion, three of them “up to” amounts and much of it for capacity not yet delivered.
  • The GPU-hour is becoming a financial asset.CME, ICE and Nodal Exchange have announced GPU futures, each settling on a different price index, while lenders are already lending against GPUs: Lambdas investment-grade-rated loan priced at 6.78% and Sharon AIs, with no rating disclosed, at 9.95%.

“Open-weight models are close behind the best closed ones, and companies are switching to them. Any company with GPUs can now serve a near-frontier model on hardware it controls. The GPUs are the hard part,” said Evgeny Ponomarev, Co-founder of Fluence.

The report closes with four ways to pay less for the same compute: compare the same chip across provider types, commit when the workload is steady, match the GPU generation to the model, and look beyond the hyperscalers, since much of the newest neocloud capacity is contracted to labs years ahead.

The full report is available at fluence.ai/compute-race-2026.

About Fluence

Fluence Cloud is a GPU cloud provider with capacity from independent data centers (fluence.ai). Founded in 2017 by Dmitry Kurinskiy, Evgeny Ponomarev and Tom Trowbridge, the company has raised $14 million from investors including 1KX, Multicoin, Distributed Global, Blockchange, Tiger Global and Protocol Labs.

면책 성명

본 기사의 견해는 저자의 개인적 견해일 뿐이며 본 플랫폼은 투자 권고를 하지 않습니다. 본 플랫폼은 기사 내 정보의 정확성, 완전성, 적시성을 보장하지 않으며, 개인의 기사 내 정보에 의한 손실에 대해 책임을 지지 않습니다.
전편

Aster CEO, 토큰이 몇 분 만에 32% 급락하자 부정행위 의심

다음

모든 전기 태양에서…일론 ‘당연하다’