Nvidia’s Revenue Per Gigawatt Seen Tripling to $40B as AI Infrastructure Value Soars

Nvidia’s Revenue Per Gigawatt Seen Tripling to $40B as AI Infrastructure Value Soars

TREE NEWS reports: Nvidia is evolving from a GPU chip vendor into a full-stack AI infrastructure platform spanning compute, networking, security and complete AI factory deployments — and the value it captures per project is rising fast. In a recent research note, Citigroup maintained a Buy rating on the stock with a $315 price target, arguing that the AI wave is simultaneously expanding both the scale of demand and the revenue opportunity per deployment.

Central to the thesis is a striking metric: Nvidia expects the revenue opportunity per gigawatt of AI infrastructure to climb from roughly $18 billion in the Hopper era to $25 billion with Blackwell, and to reach $40 billion in the Rubin generation. The increase reflects not just faster GPUs, but the bundling of CPUs, LPUs, networking gear and full AI factory infrastructure into single solutions — allowing Nvidia to capture more of each project’s total spend.

Buyback Signals Cash-Flow Confidence

Nvidia added a $150 billion share repurchase authorization, bringing total buyback capacity through fiscal 2028 to $235 billion. Against a market capitalization of roughly $5.52 trillion, that is a meaningful component of capital allocation. Citigroup reads the expansion as a signal of management’s confidence in sustained AI-driven free cash flow growth, suggesting the company can keep investing heavily in AI infrastructure, software and emerging applications while still returning substantial capital to shareholders.

Order data supports the demand narrative. One frontier AI lab has directly contracted 2.6 gigawatts of Nvidia AI infrastructure, with equipment scheduled for delivery before 2028. Indirect agreements with cloud service providers and neoclouds total more than $180 billion in value.

From GPUs to AI Security and Open Models

Nvidia is also pushing into AI agent security, compliance and governance with its Open Agent Safety Platform, which includes the OpenShell open-source runtime and a Sentry hardware monitoring system running on BlueField-4 DPUs. As AI agents enter mission-critical enterprise workflows, demand for policy controls and unauthorized-behavior detection rises — widening Nvidia’s addressable market beyond raw compute.

The open-model ecosystem is expanding rapidly: open models now account for about 75% of token generation, up from roughly 40% a year ago, while overall token consumption has grown 25-fold year over year. Nvidia argues that greater model openness accelerates AI adoption, and rising token demand pulls through underlying compute demand. The company continues to invest in language AI, physical AI, robotics and autonomous driving, and plans to acquire Hugging Face to deepen its developer-ecosystem ties.

Market Implications

For equities, the note reinforces the bull case that Nvidia’s growth is not merely cyclical but structural, tied to the buildout of AI factories worldwide. A rising revenue-per-gigawatt figure implies that even if deployment growth moderates, dollar content per project can keep climbing — a powerful earnings lever. The aggressive buyback also provides a floor under per-share metrics and signals that management sees the stock as undervalued relative to future cash generation.

For bonds, Nvidia’s massive cash generation and shareholder returns could keep its credit profile strong, though heavy capex across the AI supply chain may spur more debt issuance from customers and partners. In commodities, sustained AI infrastructure demand supports copper, power equipment and cooling-related inputs. In currencies, continued US tech leadership tends to support dollar-denominated asset flows, though the effect is indirect. Crypto markets may take a secondary cue: rising AI compute demand competes for GPU supply and capital, but also validates decentralized compute narratives that pitch idle GPU capacity as an alternative.

Key Takeaways for Investors

• Watch the revenue-per-gigawatt trajectory: the jump from $18B to $40B is a structural margin and content story, not just a volume story.
• Buybacks matter: $235 billion of capacity through FY28 supports per-share earnings and signals management confidence.
• Platform expansion widens the moat: security, networking and full AI factory solutions increase switching costs and share of wallet.
• Open models are a tailwind, not a threat: more openness drives token consumption and compute demand.
• Second-order plays: power, cooling, networking and decentralized compute networks may benefit as AI capex broadens.