NVIDIA
4 tools · 3 companies · 4 verified facts — receipts on every one.
NVIDIA GPUs (Blackwell) TensorRT-LLM NeMo Triton Inference Server
Companies with NVIDIA in the stack
All NVIDIA facts
-
Cursor (Anysphere) · NVIDIA GPUs (Blackwell)
Trains mixture-of-experts models at low precision on large GPU clusters.
High confidenceSource last verified Jul 26, 2026 View fact with receipts →
-
Perplexity · TensorRT-LLM
Served LLMs on GPU pods before the in-house engine took over.
Historical High confidenceSource last verified Jul 26, 2026 View fact with receipts →
-
Perplexity · NeMo
Framework behind the post-training run that produced R1-1776.
Medium confidenceSource last verified Jul 26, 2026 View fact with receipts →
-
Shopify · Triton Inference Server
Self-hosts fine-tuned vision models across the GPU fleet.
High confidenceSource last verified Jul 26, 2026 View fact with receipts →