Behind Anthropic’s $35 billion cloud deal, the true landlord is Nvidia

The $35 billion cloud services mega-deal signed by Anthropic is, on the surface, an agreement with cloud provider Lambda, but the party actually holding the data center lease and acting as the "landlord" is NVIDIA.

According to the Wall Street Journal on Tuesday, Anthropic has signed a $35 billion cloud services agreement with Lambda, a cloud provider backed by NVIDIA. However, the data center powering the computing capacity is built by Hut 8 in Nueces County, Texas. Weeks earlier, NVIDIA signed a separate agreement with Hut 8 to secure capacity, actually holding the lease for the data center.

In this arrangement, Lambda acts more like an intermediate layer in the compute supply chain—it installs chips purchased from NVIDIA into space leased by NVIDIA, then delivers cloud services to Anthropic. NVIDIA is not only Lambda’s chip supplier and investor, but also now holds the data center lease, making multiple profits from this deal.

The $35 billion scale once again highlights the enormous demand for fundamental computing power from leading foundation model companies, and provides additional evidence for the sustained demand for NVIDIA GPUs.

$35 Billion Mega-Deal: NVIDIA Is the "Landlord"

The $35 billion cloud services agreement between Anthropic and Lambda is one of the largest cloud service transactions in the history of the AI sector.

The unique aspect of this transaction is its multi-layer structure: Anthropic is the end customer for cloud services, Lambda is the cloud provider signing externally, while the data center hosting the computing power is actually leased by NVIDIA.

The data center is built by Hut 8 in Nueces County, Texas. NVIDIA had already signed its own agreement with Hut 8 weeks before, securing capacity in advance. This means that when Lambda deploys chips purchased from NVIDIA into this space, the actual "landlord" is NVIDIA.

Lambda: The Middle Layer Between Compute Supply and Demand

Lambda is a cloud provider focused on GPU cloud computing and is backed by investment from NVIDIA. In this transaction, it plays the role of connecting chip supply with the computing power needs of model companies: sourcing GPUs from NVIDIA, installing them into data center space leased by NVIDIA, and delivering them as cloud services to Anthropic.

Reportedly, Lambda is in talks for a Pre-IPO financing round of up to $3 billion, possibly valuing the company at over $12 billion. In a November funding round, Lambda raised more than $1.5 billion. Last year, the company also signed an agreement with Microsoft to deploy AI infrastructure driven by tens of thousands of NVIDIA processors.

For NVIDIA, this deal means multilevel participation—it sells chips to Lambda, shares in its growth as an investor, and controls the critical data center link through the lease. Around this order, NVIDIA is able to further expand its influence in the AI compute supply chain, beyond just supplying chips.

Another Order in the Compute Arms Race

The $35 billion scale represents the continuation of Anthropic’s computing power expansion strategy. To support the soaring demands for model training and inference, Anthropic has recently signed several large-scale computing agreements, including a nearly $45 billion partnership with SpaceX, an $1.8 billion agreement with Akamai, and the procurement of chips and cloud services from Google. This Lambda order marks another key move in its compute arms race against competitors like OpenAI and Google.

The significance of this transaction is clear: although Anthropic is advancing chip self-development and diversifying suppliers, its computing power expansion will still require large-scale NVIDIA GPU procurement in the medium to short term. And in this order, NVIDIA acts simultaneously as the chip supplier, cloud provider investor, and data center lessee, making its involvement much deeper than pure hardware sales.