Nvidia CEO Jensen Huang told officials in Chapel Hill that nations must treat artificial intelligence like essential power infrastructure, pegging the buildout cost at up to $60 billion per gigawatt, as the company reported second-quarter revenue of $96.22 billion amid surging global demand.
The Infrastructure Price Tag and National AI Strategy
Jensen Huang introduced a new metric for the artificial intelligence economy during the G20 Innovation Ministerial in Chapel Hill, North Carolina. Speaking to global officials, the Nvidia CEO established a benchmark for sovereign AI investment: building one gigawatt of AI infrastructure costs roughly $50 billion to $60 billion. Industry circles have dubbed this calculation Jensen Math
since mid-2025.
That price tag accounts for an entire technological ecosystem rather than a simple chip purchase. The figure wraps together the cost of Nvidia graphics processing units, specialized data centers, the cooling systems required to keep equipment from melting, and the massive electrical power infrastructure needed to run it all.
Huang argued that governments must view these outlays with the same urgency applied to power grids and national highways. He described artificial intelligence as the great equalizer,
asserting that it democratizes problem-solving capacity across borders regardless of whether a nation possesses legacy research institutions.
The ministerial featured a coordinated push from tech leadership and government officials. OpenAI CEO Sam Altman, Anthropic co-founder Tom Brown, and U.S. Commerce Secretary Howard Lutnick joined Huang on stage. Together, the delegation pressed for light-touch regulations focused on real-world harms rather than pre-emptive restrictions based on worst-case scenarios, contrasting sharply with the European Union’s comprehensive AI Act approach.
Financial Results and the AI Inflection Point
The policy pitch arrived alongside financial results that reinforced the industry’s rapid expansion. Nvidia posted adjusted earnings per share of $2.22 on total revenue of $96.22 billion for its fiscal 2027 second quarter, topping Wall Street consensus estimates of $2.10 in earnings and $92.17 billion in revenue.
Company executives signaled that the commercial adoption of artificial intelligence has crossed a fundamental threshold. Nvidia CEO Jensen Huang stated in the earnings release that it is performing useful work, noting that its tokens are productive and profitable, and adding that compute has now become revenue, alongside the observation that the market has reached an inflection point where demand accelerates across multiple frontier labs, startups, and open-model ecosystems worldwide.
Jensen Huang, Nvidia CEO, stated that the artificial intelligence infrastructure buildout is moving at full speed, and that Vera Rubin, which is now in full production, was built precisely to power this moment.
Nvidia guided for $108 billion in revenue for the third quarter, plus or minus 2%, while maintaining a gross margin target of 74%. The corporation also repurchased $26 billion of stock during the quarter, leaving $99 billion authorized under its share buyback program. Huang’s personal net worth rose to $182 billion as Nvidia’s market capitalization climbed to $5.32 trillion.
The Emerging Central Processing Unit and Agentic AI Opportunity
While graphics processing units generate the bulk of corporate revenue, Nvidia’s newer central processing unit (CPU) business is scaling to power agentic artificial intelligence. Autonomous AI agents require CPUs to handle task planning and access external data sources while GPUs handle core reasoning.
With respect to CPU use, an agent is essentially what people call a harness. And so the harness runs on CPU… The world has a billion human users. I sense that the world is gonna have billions of agents. Not today. I mean, we are gonna grow into it. Jensen Huang, Nvidia CEO
Nvidia initially projected $20 billion in CPU sales for the current fiscal year. CFO Colette Kress now expects that revenue stream to more than double in fiscal 2028. Bank of America analysts project the broader CPU total addressable market to expand from $35 billion in 2025 to $170 billion by 2030.
Market Scale and Sovereign Commitments
The sheer magnitude of the spending cycle has prompted debate among financial analysts regarding long-term sustainability. D.A. Davidson’s Head of Technology Research Gil Luria observed that market skeptics struggle to comprehend the growth runway given the current scale, noting that the market is behaving as though these numbers are so immense that there is simply nowhere left for them to grow.
Yet corporate and sovereign capital expenditures continue to expand. Nvidia has also taken on a growing role in financing the data centers that host its chips.
As nations evaluate their long-term digital strategies ahead of the full G20 leaders’ summit in December 2026, Huang’s infrastructure framework establishes the economic baseline for sovereign computing power. Whether governments choose to subsidize gigawatt-scale data centers or rely on private sector buildouts, the capital requirements outlined in Chapel Hill ensure that artificial intelligence remains at the center of global fiscal and industrial planning.
