Acasia

★ 3/5 · Infrastructure/Database

GPU cloud compute provider offering serverless GPU endpoints with zero-markup pricing that passes through bare hardware costs directly to customers.

Pros

Zero markup pricing (bare hardware costs only)
serverless GPU endpoints reduce ops overhead
smart inference optimization automatically matches workloads to hardware

Cons

Emerging provider with smaller GPU selection than hyperscalers
potentially limited region availability
less ecosystem integration than AWS/GCP/Azure

Cost

The only rate published on the site is $1.28 per GPU-hour for a 48-GPU A100 SXM cluster in Santa Clara; bare metal clusters, marketplace rentals and newer Blackwell hardware are quoted on request, and no subscription tiers are listed.

Verdict

Best for cost-conscious AI startups and teams running inference workloads that benefit from transparent, zero-markup GPU pricing.

Visit Acasia