The USA Leaders

March 24, 2026

In early 2024, OpenAI CEO Sam Altman reportedly explored plans to raise to $7 trillion in global investment to expand semiconductor manufacturing and AI infrastructure. 

At roughly the same time, Nvidia CEO Jensen Huang suggested that early forms of general-purpose AI capability were already emerging in advanced systems.

The conversation about artificial intelligence has shifted. The debate is no longer only about algorithms or computing power. It now concerns the physical resources required to sustain the next technological era.

What Is AGI?

Understanding what is AGI requires looking beyond software. Artificial General Intelligence refers to AI systems capable of performing any intellectual task a human can perform, rather than solving one narrow problem.

Today’s AI models remain specialized. A system that generates text cannot independently run a company’s logistics network. A coding model cannot design a pharmaceutical drug without human guidance. Each system performs one category of work extremely well, but struggles outside its training domain.

AGI represents a shift from narrow intelligence to general capability.

An AGI system could reason across disciplines, analyze new situations without retraining, and perform multiple types of cognitive work such as research, planning, financial modeling, and technical design.

In practical terms, this means an AGI system could assist in running businesses, accelerating scientific discovery, optimizing supply chains, or managing large operational systems.

Artificial general intelligence also represents a new layer of global infrastructure. Data centers, semiconductor factories, cooling systems, and electrical grids form the backbone of this system.

Modern AI already consumes extraordinary resources. Training advanced models requires massive computing clusters, enormous electricity demand, and billions of gallons of water for cooling. 

Semiconductor production depends on lithium, copper, and rare earth materials sourced through complex global supply chains.

In effect, humanity is building a new organ for civilization. Every organ requires energy to function. For AGI, that energy comes from the same natural resources societies already struggle to manage.

Why This Future Is Non-Negotiable

The environmental and infrastructure costs of artificial intelligence invite a simple response. Some observers argue that society should slow the pace of AI development.

However, several structural forces make that unlikely.

The first is the growing complexity ceiling. Modern systems such as climate models, global supply chains, and genomic research generate massive volumes of data that exceed the analytical capacity of human experts. 

Researchers increasingly rely on machine learning systems to process these datasets.

Second, artificial intelligence is already accelerating scientific discovery. AI models now assist with protein-folding predictions, materials research, and early drug-discovery pipelines.

Pharmaceutical and energy research teams use machine learning to analyze millions of potential molecular combinations in days instead of years.

Third, geopolitical competition drives continued investment. Governments increasingly view advanced computing infrastructure as a strategic capability comparable to telecommunications or nuclear energy.

The result is a difficult trade.

Society is allocating massive quantities of electricity, water, and semiconductor materials to build AI infrastructure. 

The expectation is that these systems may eventually accelerate discoveries in fusion energy, carbon capture materials, and advanced medical treatments.

In practical terms, humanity is using its current resources to create a tool that may help discover future ones.

Another way to frame the trade is simple.

Humanity is burning its remaining candles trying to build a sun.

NVIDIA’s Vision for Autonomous Companies

NVIDIA sits at the center of the modern AI ecosystem. Its GPUs power the majority of large-scale AI training systems used by technology companies and research labs.

Industry leaders increasingly discuss the possibility that advanced AI systems could manage complex business processes. 

Some projections suggest that future AI agents may oversee operational systems capable of generating hundreds of millions or even billions of dollars in annual value.

Large technology firms already experiment with autonomous agents that analyze data, plan workflows, and assist with decision support.

However, these systems remain tools rather than independent leaders. They can process information rapidly but cannot provide accountability, ethics, or long-term stewardship.

The discussion around what is AGI increasingly focuses on how organizations combine machine capability with human responsibility.

The Hidden Tax of AI Infrastructure

Artificial intelligence requires a vast physical infrastructure. Three major resource demands shape the economics of this system.

• The Hardware Tax

Training advanced AI models requires massive GPU clusters. Many current training runs already use tens of thousands of chips. 

Future frontier systems may require significantly larger clusters as model complexity increases.

Each semiconductor depends on materials such as lithium, cobalt, and copper. Mining operations that supply these materials often operate in regions with limited environmental oversight.

Another emerging factor involves sovereign compute. Governments increasingly regulate advanced semiconductors because they view AI capability as strategic infrastructure. 

Export controls on high-performance GPUs already influence global AI development.

• The Energy Tax

AI training requires enormous electricity consumption. Estimates suggest that training frontier models may require power comparable to a small industrial facility.

Hyperscale data centers running AI inference workloads can consume a significant share of local grid capacity in technology hubs.

These projections vary by model architecture and efficiency improvements, but the overall trend points toward rapidly increasing energy demand.

• The Water Tax

Cooling systems represent another critical cost.

Large data centers rely on water-based cooling systems to maintain safe operating temperatures. 

Public disclosures indicate that Google’s global data centers consumed more than 5 billion gallons of water in 2022.

As AI infrastructure expands, water availability becomes a planning constraint in several regions.

• Sustainability Debt

The table below illustrates the difference between current AI infrastructure and projected requirements for larger-scale AI systems based on current engineering trends.

*These projections reflect industry direction rather than fixed outcomes. Efficiency improvements in chip design and cooling systems may reduce some of these pressures.

• The Cost Will Reach Corporate Balance Sheets

Large technology companies currently absorb most infrastructure costs associated with AI development.

Over time, these costs will move through the technology supply chain. Companies may experience this shift through higher SaaS pricing, increased AI API costs, and premium pricing for high-performance cloud computing.

Even organizations that do not operate their own data centers will still pay these infrastructure costs indirectly through software subscriptions and cloud services.

The Human Ledger

The infrastructure demands of AI affect real communities and ecosystems. Lithium extraction in Chile’s Atacama Desert consumes groundwater used by local agriculture. 

Cobalt mining in parts of Central Africa supplies critical materials used in electronics and battery systems.

These supply chains illustrate a recurring pattern in global technology development. 

High-value digital services often rely on resource extraction occurring far from the end user.

AI infrastructure may also reshape urban economies. 

Cities capable of hosting hyperscale data centers may experience economic growth and new tax revenue. At the same time, local utilities must expand electricity capacity to support those facilities.

This dynamic may eventually create competition between residential energy consumption and large-scale digital infrastructure.

The Labor Paradox

Artificial intelligence reshapes how organizations structure work.

Entry-level cognitive tasks such as data analysis, financial modeling, and routine coding increasingly rely on AI assistance.

A financial model that once required two days of manual work can now be generated in two hours using AI tools.

However, productivity gains introduce a structural challenge.

Historically, professionals learned their fields by performing entry-level work repeatedly. That process created the experience required for senior expertise.

When AI automates those early tasks, the traditional pathway toward mastery becomes less clear.

Some economists describe the potential outcome as a hollow economy, where organizations rely on a smaller group of senior professionals supported by extensive AI systems.

At the same time, technological change has historically created new types of work. The rise of computers eliminated some roles while generating entirely new industries.

The key difference with AI lies in the speed of transformation.

Another subtle loss may occur as well.

When the struggle of manual work disappears, professionals may lose the deep intuition that emerges from wrestling with raw data and imperfect information.

That intuition often produces the insights that shape strategy and innovation.

Solutions and Mitigation

While the resource costs of AI are high, several technological developments aim to reduce environmental impact.

Chip manufacturers continue to improve energy efficiency per computation with new semiconductor architectures.

Many hyperscale data centers now deploy liquid cooling technologies that reduce water consumption and improve energy efficiency.

Major cloud providers invest heavily in renewable energy infrastructure, including solar, wind, and small modular nuclear reactors.

Researchers also explore alternative cooling methods, including closed-loop water systems and advanced air cooling techniques.

These innovations may help reduce the environmental footprint of large-scale AI infrastructure over time.

Conclusion

The debate around artificial intelligence often focuses on technical capability. A deeper conversation concerns the resources required to sustain this technology.

Understanding what is AGI means recognizing both its promise and its cost.

Building global AI infrastructure requires electricity grids, water systems, semiconductor supply chains, and skilled human oversight. 

These systems represent one of the largest technological investments in modern history.

If society uses this capability only for marginal productivity gains, the environmental tradeoffs may prove difficult to justify.

However, if advanced AI systems accelerate breakthroughs in medicine, energy, and climate science, the investment may reshape the trajectory of human progress.

Technology companies will build the infrastructure. Governments will shape regulation. Leaders ultimately determine the purpose for which these systems are used.

Neha Shekhawat