Lambda Scores $320 Million for AI Infrastructure Cloud

Lambda, a cloud firm offering enterprises access to artificial intelligence (AI) infrastructure via a cloud based on NVIDIA graphics processing units (GPUs), has scored $320 million in Series C funding on a valuation of $1.5 billion. The company has raised over $388 million since 2021.

The round highlights burgeoning demand for supercomputing components from NVIDIA (Nasdaq: NVDA), which has seen its corporate value rise to the trillion-dollar level thanks demand for its market-leading chips.

Indeed, customers can’t seem to get enough NVIDIA processing power to train the large language models (LLMs) and inferencing needed for generative AI (GenAI) applications. Lambda, founded in 2012, has capitalized on the chance to offer a cloud-based share of NVIDIA’s most powerful chips, including the H100, H200, and GH200, all linked with InfiniBand.

NVIDIA H100 Instances from $2.49/hr.!

Lambda was founded in 2012 to, according to its website, “train machines how to see and learn.” By 2017 it was offering its own supercomputing hardware designs in the form of blades and workstations to enterprise customers, and by 2018 it launched the Lambda Cloud. The company now hawks reserved and on-demand time on its infrastructure, starting as low as $2.49 per hour for an NVIDIA H100 instance.

Lamdba claims to have had over 100,000 customer sign-ups to Lambda Cloud, and its hardware and cloud offerings have attracted over 5,000 customers across a range of vertical markets – though the press release mentions manufacturing, healthcare, pharmaceuticals, financial services, and the U.S. government specifically. Named customers include Intel, Microsoft, Amazon Research, Kaiser Permanente, Stanford, Harvard, Caltech, and the Department of Defense.

“AI is fundamentally changing the world,” said Lambda CEO and co-founder Stephen Balaban in a video announcing the funding. “If you flash forward a couple of years from now, we believe that it’s going to affect the world to such an extent that there’s going to need to be the computational power of one GPU in the hands of every person on the planet.”

Lambda CEO Stephen Balaban. Source: Lambda

Balaban notes that the round will bring even more GPUs into the Lambda stable. “As part of this round, we’re going to be deploying tens of thousands of NVIDIA GPUs into the Lambda Cloud,” he said in his vlog.

A Growing Market for AI Infrastructure

Lamdba is among the first startups to take a lead in the market for AI infrastructure – particularly in rentable GPUs. Other firms include CoreWeave, a former bitcoin mining firm now in the GPU rental business. As Iain Martin wrote in Forbes last September: “Other miners like Hive, Crusoe, and Hut 8 have also converted GPUs now essentially useless for ethereum mining after the cryptocurrency’s September 2022 energy-saving overhaul to renting them by the hour to chip-hungry startups.”

Other firms, such as Together AI, offer cloud-based and on-premises AI infrastructure, including access to H100s. In November 2023, together raised $102.5 million in Series A financing from an investor pool including Kleiner Perkins, NVIDIA, and Emergence Capital.

Lambda’s Series C round was led by Thomas Tull’s US Innovative Technology with participation from new investors B Capital, SK Telecom, funds and accounts advised by T. Rowe Price Associates, and existing investors Crescent Cove, Mercato Partners, 1517 Fund, Bloomberg Beta, and Gradient Ventures, among others.

Startup Profile: Lambda

HQ location: San Jose, Calif.

Employees: 50-200 on LinkedIn

CEO and Co-Founder: Stephen Balaban

Target market: GPU-accelerated compute architecture via on-premises hardware and cloud services for generative AI processing.

Prominent investors: Thomas Tull’s US Innovative Technology, B Capital, SK Telecom, Crescent Cove, Mercato Partners, 1517 Fund, Bloomberg Beta, Gradient Ventures, among others.

Funding raised to date: $320 million Series C; $24.5 million Series A and $44 million Series B --- plus at least three other rounds not publicly disclosed.