Jensen Huang rarely posts: AI is an important force reshaping the world, like electricity and the internet—fundamental infrastructure.

On March 10th, NVIDIA CEO Jensen Huang rarely elaborated on the development logic of the AI industry in a personal signed article.

He pointed out that AI should not be understood as a single model or application, but as an emerging infrastructure system.

Artificial Intelligence (AI) is one of the most powerful forces shaping the world today. It is not just a clever application or a single model; it is an indispensable infrastructure, like electricity and the internet.

In his view, the AI industry is undergoing a technological infrastructure buildout comparable to an industrial revolution. Currently, hundreds of billions of dollars have been invested worldwide, but the overall development is still in its early stages.

Huang stated that AI is a “five-layer cake” infrastructure—energy, chips, infrastructure, models, and applications—and requires trillions of dollars more to build.

AI is shifting from “software” to real-time intelligent generation

Huang first explained the fundamental difference between AI and traditional software.

Over the past decades, software has been essentially “pre-recorded programs.” Developers write algorithms, and computers execute them according to rules. Data must be structured and retrieved via database queries. But AI has changed this pattern.

Huang wrote: “This is the first time in computing history that machines can understand unstructured information—images, text, sounds—and grasp their meaning.”

More importantly, AI does not read answers from databases but generates intelligence in real time.

He explained: “Every response is newly generated, and each output depends on the context. Computers are no longer just executing instructions; they are reasoning.”

Because intelligence is generated in real time, the entire computing architecture must be redesigned.

The “Five-Layer” Structure of the AI Industry

In the article, Huang proposed a framework for the AI industry: a five-layer technology stack—energy, chips, infrastructure, models, and applications. He emphasized that these five layers are strongly interconnected.

Energy is the most fundamental layer. Real-time generated intelligence requires real-time power. Every token generated involves electron movement, heat management, and energy conversion into computing power. Beneath this layer, there are no abstractions. Energy is the first principle of AI infrastructure and a hard constraint on how much intelligence the system can produce.

Chips sit above energy. These processors are designed to convert large-scale, efficient energy into computing power. AI workloads demand massive parallel computing, high-bandwidth memory, and fast interconnects. Advances in chips determine the speed of AI scaling and how affordable intelligence becomes.

Infrastructure is built on chips. This includes land, power delivery, cooling systems, construction, networks, and systems that coordinate thousands of processors into a single machine. These systems are “AI factories.” Their purpose is not data storage but the manufacturing of intelligence.

Models are above infrastructure. AI models can understand various types of information: language, biology, chemistry, physics, finance, medicine, and the physical world itself. Language models are just one category. Currently, some of the most disruptive work is happening in protein AI, chemical AI, physics simulation, robotics, and autonomous systems.

Applications are at the top, where economic value is created. Drug discovery platforms, industrial robots, legal assistants, and autonomous vehicles all fall into this category. Autonomous vehicles are AI applications embodied in machines; humanoid robots are AI applications embodied in bodies. They use the same tech stack but produce different results.

AI infrastructure buildout is still in early stages

Regarding industry scale, Huang provided a clear assessment.

He said: “We have currently invested only a few hundred billion dollars, but we will need to build infrastructure worth trillions of dollars in the future.”

Globally, chip factories, server assembly plants, and AI data centers are accelerating construction. Huang called this trend potentially “one of the largest infrastructure projects in human history.”

Meanwhile, this also creates new labor demands. Building AI data centers requires many skilled workers, including electricians, plumbers, network engineers, and equipment installers.

He emphasized: “You don’t necessarily need a PhD in computer science to participate in this transformation.”

Open-source models drive AI industry expansion

Huang also highlighted the role of open-source models in the AI ecosystem.

He pointed out that many AI models worldwide are open, and companies, research institutions, and countries rely on these models to participate in AI development. When open-source models reach advanced levels, they stimulate demand across the entire industry chain.

He cited an example: “DeepSeek-R1 is a typical case.”

After the model was made public, it promoted application development and increased demand for training compute, infrastructure, chips, and energy. In other words, breakthroughs in a single model can pull the entire industry chain downward.

AI’s impact extends beyond the software industry

At the end of the article, Huang emphasized that AI will not only transform the software industry but also impact energy, manufacturing, labor structures, and economic growth.

He said: “AI is an industrial-scale transformation that will change how energy is produced, how factories are built, how work is organized, and how economies grow.”

He believes that AI is still in its early stages. Much infrastructure remains unbuilt, and many talents are yet to be trained.

But the trend is very clear: “AI is becoming the infrastructure of the modern world.”

“AI as the ‘Five-Layer Cake’ Infrastructure”

March 10, 2026, Speaker: Jensen Huang

Artificial Intelligence (AI) is one of the most powerful forces shaping the world today. It is not just a clever application or a single model; it is an indispensable infrastructure, like electricity and the internet.

AI operates on real hardware, real energy, and real economic foundations. It sources raw materials and transforms them at scale into intelligence. Every company will use it, every country will build it.

To understand why AI is developing this way, we need to reason from first principles and examine the fundamental changes occurring in the computing field.

From Pre-Recorded Software to Real-Time Intelligence In much of computing history, software has been “pre-recorded.” Humans write algorithms, and computers execute them. Data must be carefully structured, stored in tables, and retrieved via precise queries. SQL became indispensable because it kept that world running.

However, AI breaks this pattern.

This is the first time in history that computers can understand unstructured information. They can interpret images, read text, understand sounds, and grasp their meanings. They can reason about context and intent. Most importantly, they can generate intelligence in real time.

Each response is newly created. Every answer depends on the context you provide. It is no longer software that retrieves pre-stored instructions but software that can reason and generate intelligence on demand.

Because intelligence is generated in real time, the entire underlying computing stack must be reinvented.

AI as Infrastructure When viewed from an industrial perspective, AI can be broken down into a five-layer technology stack.

Energy The most fundamental layer is energy. Real-time generated intelligence requires real-time electricity. Every token generated involves electron movement, heat management, and energy conversion into computing power. Beneath this layer, there are no abstractions. Energy is the first principle of AI infrastructure and a hard limit on how much intelligence the system can produce.

Chips Above energy are chips. These processors are designed to convert energy into large-scale, efficient computing power. AI workloads demand massive parallelism, high-bandwidth memory, and fast interconnects. Advances in chips determine the pace of AI scaling and how affordable intelligence becomes.

Infrastructure Above chips is infrastructure. This includes land, power delivery, cooling systems, construction, networks, and systems that coordinate thousands of processors into a single machine. These are “AI factories.” Their purpose is not data storage but the manufacturing of intelligence.

Models Above infrastructure are models. AI models can understand various types of information: language, biology, chemistry, physics, finance, medicine, and the physical world itself. Language models are just one category. Currently, some of the most disruptive work is happening in protein AI, chemical AI, physics simulation, robotics, and autonomous systems.

Applications At the top are applications, where economic value is created. Drug discovery platforms, industrial robots, legal assistants, and autonomous vehicles all fall into this category. Autonomous vehicles are AI applications embodied in machines; humanoid robots are AI applications embodied in bodies. They use the same tech stack but produce different results.

This is the “Five-Layer Cake” of AI: Energy → Chips → Infrastructure → Models → Applications.

Every successful application strongly drives each layer below it, extending all the way to the power plants that sustain it.

We are just beginning this buildout. So far, hundreds of billions of dollars have been invested, but trillions more are needed.

Globally, chip factories, computer assembly plants, and AI factories are rising at unprecedented scale. This is becoming the largest infrastructure project in human history.

The workforce needed for this buildout is enormous. AI factories require electricians, plumbers, pipefitters, steelworkers, network technicians, installers, and operators. These are high-skill, high-wage jobs, currently in short supply. You don’t need a PhD in computer science to participate in this transformation.

Meanwhile, AI is boosting productivity across the entire knowledge economy. For example, in radiology, AI can assist in reading scans, yet the demand for radiologists continues to grow. This is not a paradox.

The core role of radiologists is patient care, and reading images is just one task. As AI takes on more routine work, radiologists can focus on diagnosis, communication, and patient care. This increases hospital efficiency, allowing more patients to be served and more staff to be employed.

Productivity creates capacity, and capacity drives growth.

What has changed in the past year? Over the last year, AI has crossed an important threshold. Models have become sufficiently capable to deliver practical value at macro scale. Reasoning ability has improved, hallucinations have decreased, and factual grounding has significantly increased. For the first time, AI-based applications are generating real economic value.

Applications in drug discovery, logistics, customer service, software development, and manufacturing have demonstrated strong product-market fit. These applications are powering each underlying technological layer.

Open-source models play a crucial role here. Most models worldwide are free. Researchers, startups, large corporations, and entire nations rely on open-source models to participate in the AI wave. When open models reach cutting-edge levels, they change not only software but also activate demand across the entire tech stack.

DeepSeek-R1 is a powerful example. By enabling a robust reasoning model to be widely used, it accelerates application deployment and increases demand for training, infrastructure, chips, and energy at the bottom.

What does this mean? When you see AI as an indispensable infrastructure, its profound impact becomes clear.

AI started with large language models based on Transformer architecture. But it is much more than that. It is an industrial revolution that reshapes how energy is produced and consumed, how factories are built, how work is organized, and how economies grow.

The reason for building AI factories is because intelligence is now generated in real time; chips are being redesigned because efficiency determines the speed of AI expansion; energy is central because it sets the absolute upper limit of intelligence capacity; applications are accelerating because their underlying models have crossed a threshold and can deliver practical value at scale.

Each layer mutually reinforces the others.

This is why the scale of this infrastructure buildout is so enormous. It touches so many industries. It is not limited to any single country or field. Every company will use AI, every country will build it.

We are still in the early stages. Most infrastructure remains unbuilt, most talent untrained, most opportunities unexploited.

But the direction is very clear:

AI is becoming the foundational infrastructure of the modern world.

The choices we make now—how fast we build, how broadly we participate, how responsibly we deploy AI—will ultimately shape the future of this era.