NVIDIA's Nuclear-Level Strategy Just Revealed! Jensen Huang Announces Partnership with Altman to Secure Land and Power, Aggressively Targeting 16GW Computing Hegemony

Jensen Huang has just dropped a bombshell. NVIDIA has officially announced that it will directly step in to secure the absolute core physical resources of land, power, and facilities to build super AI factories.

In an AI-dominated economy, computing power directly translates into hard cash revenue. AI factories are the most representative infrastructure of this era, continuously converting computing power, energy, and data into intelligence that drives various industries.

A complete AI factory requires top-tier chips, packaging technology, memory, and networking, but even more critically, it relies on the three major physical cornerstones: land, power, and facilities, abbreviated in the industry as LPS. NVIDIA has decided to replicate its powerful approach to managing the chip supply chain directly into the competition for LPS resources, exclusively reserving capacity for NVIDIA AI factories.

The latest concrete move is NVIDIA’s formal partnership with SB Energy to secure massive LPS capacity at the highly favorable PORTS-Pike technology park in Ohio, USA. A super data center designed to host NVIDIA’s computing power will rise from this land, with OpenAI as its tenant.

The Computing Hunger of Frontier AI Labs

For the world's largest cloud service providers and corporate giants, securing land and power has always been the foundation of their infrastructure strategy. With deep pockets, professional infrastructure teams, and Evergreen Contracts, they can independently acquire LPS resources and then purchase NVIDIA’s computing network systems to operate AI factories. This remains NVIDIA’s primary business model for the future.

The situation for frontier AI labs is entirely different.

They have terrifyingly high demands for training and inference computing power, but their expansion speed is so rapid that their balance sheets and credit capabilities simply cannot keep up. These companies hold massive customer bases and skyrocketing revenues, but they lack decades-long Evergreen Contracts and investment-grade financing capabilities, making it impossible for them to independently handle the infrastructure required for AI factories.

The bottleneck limiting their explosion has completely shifted to a shortage of computing power; algorithms and customer demand are already more than sufficient. With more computing power, they can produce more intelligence, develop more products, attract more users, and generate more revenue. NVIDIA’s direct involvement aims to provide the underlying engine for this flywheel effect.

Locking in a Super Cash Cow for the Next 20 Years

OpenAI will build and operate a world-class AI factory at PORTS-Pike. It will fully adopt NVIDIA’s full-stack DSX AI factory platform, covering GPUs, CPUs, networking, and infrastructure software.

The initial deployment is expected to provide up to 4.25 gigawatts of AI factory capacity. The figures are staggering: each generation of NVIDIA systems deployed at PORTS-Pike represents a massive volume of approximately 1.5 million NVIDIA GPUs, generating huge revenues of $150 billion to $200 billion for NVIDIA. The entire campus has a lifecycle of up to 20 years, sufficient to support multiple generations of equipment upgrades.

The core economic logic here is very clear: by securing land and power, one locks in an AI factory site with an extremely long lifespan, where NVIDIA equipment can be continuously upgraded. Each new generation of equipment brings higher capacity, stronger intelligence, and better economic benefits.

In the future, NVIDIA may even take on the remaining 3.75 gigawatts of capacity at the campus.

Binding OpenAI’s $600 Billion Computing Blueprint

Looking at the big picture, OpenAI has committed to massively deploying NVIDIA’s AI infrastructure before 2030. Currently, OpenAI’s existing and planned commitments require approximately 12 gigawatts of NVIDIA computing power. If NVIDIA truly maximizes the scale of cooperation at PORTS-Pike, this figure could soar directly to 16 gigawatts.

Calculated at this terrifying scale, by 2030, this deal represents an approximately $600 billion order for NVIDIA computing power.

Addressing Four Core Questions

Following the announcement, there was intense external interest in the details, and Jensen Huang provided extremely clear answers.

First, what exactly did NVIDIA guarantee? NVIDIA provides 20-year support for approximately 4 gigawatts of physical infrastructure at PORTS-Pike, ensuring that this land is used exclusively for deploying NVIDIA computing power. This support is limited to rent and electricity payments for specific parts, plus specific residual value commitments. NVIDIA will not take on all costs of the site or all obligations of the tenant. These guarantees will come into effect in stages between 2028 and 2030 as the data centers become operational. As long as OpenAI pays rent on time, NVIDIA’s risk exposure will continue to decrease.

Second, what is NVIDIA’s motivation for providing guarantees? Land and power have become the biggest bottlenecks constraining AI factory construction. NVIDIA’s approach is to carefully select the top-tier sites and deploy multiple generations of computing equipment on them to meet customers' extremely robust demand.

The third point of contention is whether this constitutes circular financing. The answer is absolutely not. Rent is paid in full by OpenAI. NVIDIA merely leveraged its scale and forward-looking predictions to secure this land. This is consistent with their logic for managing the supply chain: whenever they see clear customer demand that can be converted into long-term productivity, they decisively act to lock in core resources.

The final question is: what if OpenAI stops using this site in the future? NVIDIA’s computing equipment is highly universal and easy to transfer. This capacity can be directly handed over to cloud service providers, enterprises, or other AI startups within the global ecosystem. The CUDA ecosystem gives these devices a common soul, turning them into high-value hard currency that can be rented out and financed at any time. The value of this prime location is definitely not tied to a single customer or a specific generation of equipment.

The Strongest Infrastructure Builder of the Intelligence Era

NVIDIA’s land acquisition strategy in the future will be very restrained and precise. The vast majority of customers will still handle land and power themselves. NVIDIA will only focus on top-tier sites with extremely stable demand capable of supporting multiple generations of equipment iterations.

From initial accelerated computing chips to systems, networking, the CUDA ecosystem, and now full-stack AI factories, NVIDIA has been evolving. Now they have taken a crucial new step, beginning to directly lock in the underlying physical resources required to build these factories.

NVIDIA has completely transformed into a full-stack AI infrastructure platform. They are heavily investing in the long-term future of AI factories, ensuring that customers can deploy the world’s most profitable computing machines generation after generation. By mastering land and power, NVIDIA is helping the world’s most innovative companies build super factories that drive the entire intelligence era.

AI Cambrian

Risk Warning and Disclaimer

The market involves risks, and investment requires caution. This article does not constitute personal investment advice, nor does it take into account individual users' special investment goals, financial status, or needs. Users should consider whether any opinions, views, or conclusions in this article align with their specific circumstances. Investment based on this content is at your own risk.