Lambda Stack

What it does

Lambda Stack is a curated set of Ubuntu packages for machine learning and deep learning, including PyTorch, TensorFlow, JAX, CUDA, NCCL, and the NVIDIA Container Toolkit. You start with an Ubuntu system and end with a configured environment that includes NVIDIA drivers, firmware, and AI libraries. Each package is tested for compatibility across Lambda systems, including NVIDIA HGX B200 and H200 SXM GPUs.

Researchers and engineers use it to skip manually assembling a compatible CUDA and framework stack. It comes preinstalled on Lambda instances and clusters, or you can install it on other Ubuntu servers and workstations with an install script. Once installed, the packages are available from Python virtual environments, GPU-accelerated Docker containers run without extra setup, and a single apt command upgrades the packaged AI software.

It suits ML engineers, researchers, and teams running GPU workloads on Ubuntu. Installation requires supported NVIDIA GPU hardware you own or rent; cloud compute and hardware are separate purchases. Non-interactive installation requires agreeing to the cuDNN license first. The documentation covers Ubuntu installs only.

What it can do

• Preinstalled PyTorch, TensorFlow, JAX, and Triton packages with CUDA support
• Automatic detection of hardware and installation of appropriate NVIDIA drivers and firmware
• Interactive and non-interactive install script for Ubuntu servers and workstations
• Python virtual environment access to Lambda Stack packages
• GPU-accelerated Docker containers via the included NVIDIA Container Toolkit
• Single apt command to upgrade AI software packages

Before you choose

• Running Lambda Stack requires NVIDIA GPU hardware; the software does not supply compute.
• Non-interactive installation requires you to read and agree to the cuDNN license first.
• Lambda's GPU instance and cluster rates are billed separately from the free software and exclude applicable sales tax, VAT, or GST.

Listed under

• Coding
• Sandboxes and infrastructure
• Cuda
• Deep learning
• Developers
• Developer tools
• GPU
• Machine learning

Get started

Non-interactive installs require agreeing to the cuDNN license; Lambda Stack comes preinstalled on Lambda cloud instances

1. Run the Lambda Stack install script on an Ubuntu system, using the non-interactive command with I_AGREE_TO_THE_CUDNN_LICENSE=1 for servers or the interactive command for workstations.
2. Reboot the system after an interactive installation so drivers and firmware take effect.
3. Create and activate a Python virtual environment with --system-site-packages to access Lambda Stack libraries.

Confirm the applicable plan, usage charges and commercial terms with the provider.

We couldn’t find published prices for Lambda Stack. See their pricing page ↗

Sources for this overview

Product pages, documentation, and pricing checked 29 Oct 2026. Each source supports part of the overview; availability and prices can change.

• Lambda Stack AI Software for Deep Learning & Machine Learning ↗lambda.ai
• GPU cloud pricing: rent NVIDIA H100, H200, and B200 | Lambda ↗lambda.ai
• Lambda Docs ↗docs.lambda.ai

Company website

lambda.ai: product, pricing and policy pages

YC directory mirror

Batch and company name

Social accounts

Accounts the company links, and its posts on X

Human review

Checked by a person

knowledge_base

Last checked 29 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.