Please Wait...
Please Wait...
Fourth-generation Tensor Cores and third-generation RT Cores for AI and graphics
The L40S features NVIDIA's Ada Lovelace architecture with 18,176 CUDA cores, 864GB/s memory bandwidth, and hardware support for FP8 precision. This enables exceptional performance for LLM training, multimodal GenAI workloads, and real-time ray-tracing applications.
Get started with
world’s largest privately-held cloud
infrastructure company
Vultr Cloud GPU Accelerated by NVIDIA L40S
Built to power the most demanding AI and graphics-intensive workloads for the data center.
Enterprises face mounting pressure to deploy AI infrastructure that keeps pace with rapid industry transformation. Generative AI models and LLM workloads demand universal computing solutions that provide accelerated compute, graphics, and video processing at scale.
Vultr Cloud GPU powered by NVIDIA L40S delivers up to 1.7x training performance versus previous-gen GPUs, with 48GB memory capacity and flexible deployment through GPU passthrough, 8-GPU bare-metal servers, and pre-configured images.
