Lambda Labs Inc., a startup with a cloud platform optimized to run artificial intelligence models, today announced that it has raised $480 million from investors.

The Series D round reportedly values the company at $2.5 billion. It was led by Andra Capital and SGW with participation from more than a dozen other investors. The group included Nvidia Corp., Super Micro Computer Inc. and prominent computer scientist Andrej Karpathy.

“Lambda is investing billions of dollars to build the software platform and infrastructure powering AI,” said Lambda co-founder and Chief Executive Officer Stephen Balaban.

Developers use Lambda’s namesake cloud platform to train AI models, fine-tune them for specific tasks and run inference workloads. The company disclosed today that its data centers host more than 25,000 Nvidia Corp. graphics processing units. Lambda is currently in the process of adding the chipmaker’s newest and fastest GPU, the Blackwell B200, to its cloud. 

One of the platform’s main selling points is an offering called 1-Click Clusters. It allows developers to quickly provision AI clusters with up to 512 of Nvidia’s H100 chips. Introduced in 2022, the H100 is two generations behind the Blackwell B200 but remains popular among data center operators. 

The GPU servers in a 1-Click Cluster each include a 24-terabyte flash storage pool for storing AI applications’ data. Additionally, the cluster features three standard servers powered by central processing units. The latter machines process inbound network traffic from applications and run the management software that coordinates the GPUs.

Lambda’s cloud servers run an internally developed software bundle called the Lambda Stack. It’s based on Ubuntu, a popular Linux distribution, and includes preconfigured versions of several AI development tools. The software automates much of the manual work usually involved in updating those AI tools. 

Lambda offers its flagship cloud platform along several AI-optimized hardware products.

The company is preparing to launch an AI appliance called Scalar Server that enterprises can deploy in their on-premises data centers. According to Lambda, the system is equipped with eight H100 graphics cards. The company also offers AI appliances based on Nvidia’s DGX appliance architecture that likewise offer eight GPUs.

In addition to its server lineup, Lambda sells a collection of AI-optimized desktop computers known as the Vector series. Developers can use the machines for tasks such as testing newly trained AI models. The most capable desktop in the series, the Vector Pro, ships with four Nvidia chips and a collection of preinstalled AI development tools.

Lambda will use the proceeds from its funding round to enhance its cloud infrastructure and software. One focus of the engineering effort will be Lambda Chat, a service through which the company provides free access to open-source large language models.

Image: Lambda