Rent NVIDIA A2 in the Cloud

According to our July 14, 2026 snapshot, NVIDIA A2 cloud lease runs from $0.22 per hour on-demand. Cherry Servers is currently the lowest-cost, and leads on spot pricing at per hour (~up to 60% discount). Multi-month reserved contracts tack on another ~up to 40% saving.

Billing granularity and egress fees matter as much as the headline rate. Per-second billing saves money on short training runs; a provider with zero egress fees is often cheaper in practice than a slightly lower hourly rate at a provider that charges for data transfer.

The cheapest NVIDIA A2 cloud access right now is on Cherry Servers at $0.22/hr.

NVIDIA A2 specifications: 16 GB GDDR6 memory, 200 GB/s bandwidth, 18 FP16 TFLOPS, 4.5 FP32 TFLOPS, 60W TDP, built on Ampere (released 2021).

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.

Get NVIDIA A2 on Cherry Servers starting at $0.22/hr — live pricing and fast provisioning.

How fast is NVIDIA A2? The raw numbers: 18 TFLOPS FP16, 4.5 TFLOPS FP32, 200 GB/s memory bandwidth. In mixed-precision AI jobs, that translates to sustained throughput comfortably above older generations.

For model training, expect wall-clock times that scale predictably from those TFLOPS figures at large batch sizes. For low-latency inference, real-world latency is dominated by memory bandwidth and by how much of your KV-cache fits on-chip — so the 200 GB/s and 16 GB capacity matter more than headline TFLOPS.

Get NVIDIA A2 on Cherry Servers starting at $0.22/hr — live pricing and fast provisioning.

The 1 cloud vendors offering NVIDIA A2 right now: Cherry Servers. Lowest on-demand price: Cherry Servers, $0.22 per hour. Lowest spot price: , per hour.

Each vendor brings different trade-offs on network, support, and billing model. Community-driven marketplaces (Vast, RunPod, Genesis Cloud) tend to be cheapest; bare-metal specialists (FluidStack, Latitude.sh, Cherry Servers) tend to excel on interconnect; hyperscalers (Google Cloud, DigitalOcean) win on regions and enterprise tooling.

Get NVIDIA A2 on Cherry Servers starting at $0.22/hr — live pricing and fast provisioning.

NVIDIA A2 is ideal for Edge inference, entry-level AI. That covers the workloads where the card's balance of VRAM (16 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 $0.22 per hour from Cherry Servers, making it reachable for any team willing to move to the cloud for compute.

Deploy NVIDIA A2 on Cherry Servers for $0.22/hr — currently the only tracked provider offering this GPU.