Nvidia Confirms Lease of 1GW Texas Data Center, 30-Year Deal Valued at $50 Billion
0xBroomberg
Nvidia has confirmed it is the tenant of Hut 8's 1-gigawatt Beacon Point data-center campus in Texas, a deal worth up to $50.2 billion over 30 years — a signal that the chipmaker is betting its own balance sheet on the race to lock down AI power capacity.
How big is this lease?
The base term is 15 years at $19.6 billion, with a 3.0% annual rent escalator built in.
If all renewal options are exercised, the 30-year total reaches roughly $50.2 billion.
The campus is designed for 1 GW of capacity — enough to power a mid-sized city — and will house hundreds of thousands of Nvidia GPUs, built to Nvidia's DSX AI Factory reference architecture.
This means → this is not a standard lease. It is Nvidia using its own credit to underwrite the entire AI compute supply chain from the ground up.
Why is Nvidia leasing data centers itself?
The core logic is locking down power. The site already has grid-interconnection approval — an increasingly scarce asset as developers compete for electrical capacity across the U.S.
A senior executive familiar with the deal said: "They have the balance sheet to secure power, and through that, to ensure their own products get deployed."
In plain terms = Nvidia sells GPUs, but GPUs need electricity to run. Rather than wait for customers to find their own power, Nvidia is reserving the power first and handing customers a GPU-plus-facility package.
Who is Hut 8, and how did a miner become an AI landlord?
Hut 8 was originally a Bitcoin mining company that pivoted in recent years to AI data-center development.
In June it issued roughly $4.3 billion in bonds for Phase 1, at a coupon of 6.129%, and secured a Moody's investment-grade rating.
This means → the rating rests squarely on Nvidia's lease guarantee. Bankers say a data-center lease backed by a major tech company can cut financing costs by roughly half.
This reflects a broader trend: AI compute demand is pulling miners, property developers, and other non-traditional players into the data-center lane — and Nvidia's credit is their ticket in.
What is the "circular financing" risk?
After leasing the facility, Nvidia plans to sub-lease capacity to "Neocloud" partners — companies that buy Nvidia GPUs and sell AI cloud-computing services.
In plain terms = this creates a closed loop: Nvidia funds the facility → partners move in and operate → partners buy Nvidia chips. Nvidia is both landlord and chip supplier; the money circulates within its own ecosystem.
The worry: if AI demand slows and partners cannot pay rent, Nvidia faces a double hit — lost rental income and falling chip orders at the same time.
Is Nvidia's capital commitment still growing?
Its latest quarterly filing shows $32.4 billion in uncommenced data-center lease-payment obligations, up from just $7.4 billion a year earlier — more than a 4× increase in twelve months.
The Phase 1 Beacon Point lease signed in March is worth $9.8 billion, roughly one-third of that total.
Nvidia is also reportedly in talks to provide about $250 billion in funding support for OpenAI's large-scale data-center project in Ohio.
This means → Nvidia is binding its credit to AI infrastructure on an unprecedented scale. Whether it can keep expanding while staying financially healthy is the central test of this model's sustainability.
Is Google doing the same thing?
Google is similarly using its balance sheet to back data centers for its Tensor Processing Units — TPUs, Google's in-house AI chips.
This reflects a competition that has moved beyond "whose chip is faster" to "who can lock down power and facilities first."
Hut 8 says it will seek further financing for Phase 2, which would double the campus's scale — capital commitments in this race are still accelerating.
Content is for reference only, not financial advice.