NVIDIA, the company that has driven the current AI boom like no other, is now in the unusual position of having to rent its own GPUs. According to The Information, the company led by Jensen Huang has signed a multi-million dollar contract with the cloud startup Lambda, which specializes in AI computing services, to lease part of the graphics servers that Lambda previously purchased from NVIDIA itself.
The deal reflects not only the global shortage of high-performance chips but also the peculiar power dynamics shaping the AI economy.
An 18,000 GPU contract in two phases
Based on published information, the contract is divided into two stages:
- 10,000 GPUs on lease for four years, valued at approximately $1.3 billion.
- A second batch of 8,000 additional GPUs, likely older generations or lower-tier models, valued around $200 million.
In total, the agreement amounts to $1.5 billion, making NVIDIA Lambda’s biggest client to date.
Who is Lambda?
Founded in 2012, Lambda started as a startup focused on hardware and AI computation services. Its business involves leasing GPU server capacity to companies needing to train and deploy AI models. Its clients include industry giants like Microsoft, Amazon, and Google, along with key players such as OpenAI, Anthropic, and xAI (Elon Musk’s company).
In addition to hardware supply, NVIDIA is an investor in Lambda, strengthening strategic ties between the two amid high demand and supply bottlenecks for GPUs.
NVIDIA becomes a customer… of itself
The most striking detail is that NVIDIA, which manufactures the most sought-after chips globally (like the H100 and A100 GPUs), has chosen to lease computing capacity from an entity that already owns its hardware.
These leased servers won’t be available for the open market but will primarily be used for NVIDIA researchers’ internal purposes, similar to Amazon or Microsoft, which also use Lambda’s services for their own projects.
This situation echoes the case of CoreWeave, another NVIDIA-backed cloud startup, which signed similar agreements and even used its H100 GPU reserve as collateral for a $2.3 billion loan in 2023.
Context: chip shortages and the AI race
NVIDIA’s decision to lease its own hardware through Lambda highlights the intense pressure in the AI computing market.
- Major cloud providers (AWS, Azure, Google Cloud) dominate much of the latest-generation GPU output.
- Startups like CoreWeave and Lambda have carved out a niche thanks to their ability to secure early chip batches, partly backed directly by NVIDIA.
- Meanwhile, demand for AI model training (from chatbots to vision and image generation systems) continues to grow exponentially, forcing even the world’s largest GPU manufacturer to rely on leasing agreements to meet its needs.
Preparing for Lambda’s IPO
An important aspect of this deal is that it arrives at a pivotal moment for Lambda: the startup is preparing to go public in the coming months. The contract with NVIDIA enhances its appeal to investors by securing a top-tier client and a stable, multi-year revenue stream.
In a market where credibility hinges on installed capacity and flagship clients, Lambda gains strategic backing that could significantly boost its valuation on financial markets.
Conclusion
The NVIDIA-Lambda agreement reveals a paradox of the AI era: even the market leader in GPUs can’t fully supply itself.
This move also underscores the rising importance of cloud startups like Lambda and CoreWeave, which—supported by NVIDIA—are shaping an ecosystem where infrastructure becomes the real competitive edge.
In this new landscape, control over access to the latest GPUs, and under what conditions, will be decisive in determining winners and losers in the global AI race.
Frequently Asked Questions (FAQ)
1. Why is NVIDIA leasing its own GPUs instead of manufacturing them for itself?
Because demand far exceeds production capacity. Leasing GPUs to Lambda allows NVIDIA to meet its immediate research needs without waiting for new chip batches.
2. How does Lambda differ from other cloud providers like AWS or Google Cloud?
Lambda specializes in AI-optimized infrastructure, with more flexible pricing and configurations compared to hyperscalers. It also maintains exclusive agreements with NVIDIA that boost its appeal.
3. What impact does this deal have on the sector?
It consolidates Lambda’s position as a key player in AI cloud services and reinforces the trend of NVIDIA-backed startups like CoreWeave becoming strategic intermediaries for GPU access.
4. What does this mean for researchers and companies needing GPUs?
Access will likely remain limited and costly in the short term, with increasing reliance on intermediaries that have secured NVIDIA inventory.
via: tomshardware and the information