Rent AMD Instinct MI350X in the Cloud

Cloud AMD Instinct MI350X pricing in July 2026: on-demand starts at per hour (cheapest at ), spot at per hour (cheapest at ), reserved ~up to 40% below on-demand.

For reference, 1 providers currently list AMD Instinct MI350X capacity. Differences in interconnect, regions, and SLA explain why rates span a wide band — the lowest-cost isn't always the right answer for multi-node training or regulated real-time serving.

See the AMD Instinct MI350X page for the full spec sheet and current provider list.

AMD Instinct MI350X specifications: 288 GB HBM3e memory, 8,000 GB/s bandwidth, 1,800 FP16 TFLOPS, 72 FP32 TFLOPS, 1,000W TDP, built on CDNA 4 (released 2025).

For engineers evaluating the card: VRAM capacity and bandwidth are usually the two figures that determine feasibility, while TFLOPS determine time-to-result. Dense FP16 training throughput scales near-linearly with TFLOPS on well-optimised kernels, and mixed-precision pre-training can effectively double effective compute when tensor cores are in use.

See the AMD Instinct MI350X page for the full spec sheet and current provider list.

AMD Instinct MI350X hits 1,800 TFLOPS of FP16 compute with 8,000 GB/s of memory bandwidth and 288 GB of VRAM. FP32 peaks at 72 TFLOPS.

Those figures place AMD Instinct MI350X in a useful performance band for generative AI work: strong enough to pre-training mid-to-large models in reasonable time, with enough bandwidth to keep real-time serving latency low. Actual tokens-per-second or images-per-second varies 2x depending on framework, quantisation, and model size — always benchmark with the exact stack you plan to ship.

See the AMD Instinct MI350X page for the full spec sheet and current provider list.

The short list of providers carrying AMD Instinct MI350X: DigitalOcean. We track 1 in total. The best-priced on-demand rate is at ( per hour), and the best-priced spot rate is at ( per hour).

When evaluating any of them, don't just compare price — check regional coverage, compliance (SOC 2, HIPAA, GDPR), multi-GPU interconnect, and the actual realised price once egress fees are factored in. A cheap hourly rate can evaporate fast once you move the outputs of a training run off the provider's network.

See the AMD Instinct MI350X page for the full spec sheet and current provider list.

AMD Instinct MI350X is ideal for Next-gen AI training, inference at scale. That covers the workloads where the card's balance of VRAM (288 GB), compute, and bandwidth pays off — chiefly AI pre-training and real-time serving at scales that don't require the absolute top-tier accelerators.

It's a data-center-segment card, so expect data-centre-grade reliability and form factor rather than gaming-class features. Cloud pricing starts at per hour from , making it reachable for any team willing to move to the cloud for compute.

See the AMD Instinct MI350X page for the full spec sheet and current provider list.