NVIDIA CEO Jensen Huang's Latest Article: The "Five Layers of AI"

Original Title: AI Is a Five‑Layer Cake
Original Author: Nvidia
Translation: Peggy, BlockBeats

Editor's Note: Artificial intelligence is transforming from a cutting-edge technology to the infrastructure that underpins the modern economy. In its first in-depth article published on its official account, Nvidia attempts to systematically examine the industry structure of AI from first principles, covering a complete five-layer technology stack from energy and chips to data center infrastructure, models, and applications.

The article points out that AI is not just about the competition of software or models, but a global industrial construction involving energy, computing power, manufacturing, and applications. Its scale may become one of the largest infrastructure expansions in human history. Through the lens of this "five-layer cake," Nvidia seeks to illustrate that the true meaning of AI is not just smarter software but a foundational infrastructure revolution comparable in scale to the advent of electricity and the internet.

Below is the original text:

Artificial intelligence is one of the most powerful forces shaping the world today. It is not a clever application, nor is it a single model; it is an infrastructure as essential as electricity and the internet.

AI runs on real hardware, real energy, and a real economy. It turns raw materials into "intelligence" at scale. Every company will use it, and every nation will build it.

To understand why AI unfolds in this way, it is helpful to start from first principles and examine the fundamental changes that have occurred in the computing field.

From "Pre-fabricated Software" to "Real-time Generated Intelligence"

For most of the history of computing development, software has been "pre-fabricated." Humans first describe an algorithm, and then the computer follows the instructions. Data must be carefully structured, placed into tables, and retrieved by precise queries. SQL is indispensable because it makes this whole system work.

AI breaks this pattern.

For the first time, we have a computer that can understand unstructured information. It can see images, read text, listen to sounds, and understand their meaning; it can infer context and intent. More importantly, it can generate intelligence in real-time.

Each response is newly generated. Every answer depends on the context you provide. It is no longer software retrieving pre-existing instructions from a database; it is software reasoning in real-time and generating intelligence on demand.

Because intelligence is generated in real time, the entire computing technology stack that supports it must also be reinvented.

AI as Infrastructure

Looking at AI from an industrial perspective, it can actually be broken down into a five-layer structure.

Energy

At the bottom is energy.

Real-time generated intelligence requires real-time generated power. The generation of each token means electrons in motion, heat being managed, and energy being transformed into computational capability.

Below this layer, there is no abstraction. Energy is the first principle of AI infrastructure and the fundamental constraint on how much intelligence a system can produce.

Chips

Above energy is chips. The design goal of these processors is to convert energy into computational power with extreme efficiency at scale.

AI workloads require massive parallel computing, high-bandwidth memory, and fast interconnects. Progress at the chip level determines the speed of AI expansion and ultimately how cheap "intelligence" will become.

Infrastructure

Above chips is infrastructure. This includes land, power delivery, cooling systems, construction, network systems, and scheduling systems that organize tens of thousands of processors into one machine.

These systems are essentially AI factories. They are not designed to store information but to manufacture intelligence.

Models

Above infrastructure are models. AI models can understand various types of information: language, biology, chemistry, physics, finance, medicine, and the real world itself.

Language models are just one type. One of the most transformative works is happening in the following areas: protein AI, chemistry AI, physical simulation, robotics, autonomous systems.

Applications

At the top is the application layer, where economic value is truly generated. For example, a drug discovery platform, industrial robots, a legal copilot, self-driving cars.

A self-driving car is essentially a "machine-hosted AI application"; a humanoid robot is a "body-hosted AI application." The underlying technology stack is the same; only the final form of presentation is different.

So, this is the five-layer structure of AI: Energy → Chip → Infrastructure → Model → Application. Every successful application will pull all layers down to the bottom, all the way to the power plant that powers it.

Infrastructure Development Still in Its Early Days

We are just starting this development. The current investment scale is only a few hundred billion dollars, and in the future, we still need to build infrastructure at the level of tens of trillions of dollars.

On a global scale, we are seeing: chip factories, computer assembly plants, AI factories.

Infrastructure is being built at an unprecedented scale. This is becoming one of the largest infrastructure projects in human history.

Labor Demand in the AI Era

The workforce needed to support this development is very large.

AI factories need: electricians, plumbers, pipeline installers, structural steel workers, network technicians, equipment installers, operations and maintenance personnel.

All of these are highly technical, well-paid positions, and currently in extremely short supply. Engaging in this transformation does not necessarily require a computer science Ph.D.

At the same time, AI is boosting productivity in the knowledge economy. Take radiology, for example. AI has already started to assist in medical image interpretation, but the demand for radiologists is still growing.

This is not contradictory.

The true responsibility of a radiologist is to care for patients, and reading scans is just one part of the job. As AI takes over more and more repetitive tasks, doctors can spend more time on diagnosis, communication, and treatment.

The hospital's efficiency improves, allowing more patients to be served, thereby requiring more manpower. Productivity creates capacity, and capacity creates growth.

What Has Changed in the Past Year?

In the past year, AI has crossed a key threshold.

The models are now good enough to truly make an impact in large-scale scenarios.

· Significant improvement in reasoning ability

· Significant reduction in hallucinations

· Substantial enhancement in "grounding" with the real world

For the first time, AI-based applications are starting to create real economic value.

A clear product-market fit has emerged in the following areas: drug discovery, logistics, customer service, software development, manufacturing

These applications are driving the entire technology stack.

The Role of the Open-Source Model

The open-source model plays a key role in this. The vast majority of AI models worldwide are free. Researchers, startups, businesses, and even entire countries rely on open-source models to participate in the cutting-edge AI competition.

When open-source models reach the technological frontier, they not only transform software but also activate demand across the entire technology stack.

DeepSeek-R1 is a prime example. By making a powerful inference model widely available, it has driven rapid growth at the application layer while also increasing demand for training power, infrastructure, chips, and energy.

What Does This Mean?

When you view AI as infrastructure, everything becomes clear. AI may have started with Transformers and large language models, but it is much more than that.

It is an industrial-scale transformation that will reshape:

· how energy is produced and consumed

· how factories are built

· how work is organized

· the pattern of economic growth

AI factories are being built because intelligence can now be generated in real-time. Chips are being redesigned because efficiency determines the speed of intelligent expansion. Energy has become central because it determines how much intelligence a system can produce. Applications are exploding because models have finally crossed the threshold of "scale available."

Each layer reinforces the others.

That's why this construction is on such a massive scale, why it will simultaneously impact so many industries, and why it will not be limited to a single country or a single field.

Every company will use AI.

Every country will build AI.

We are still in the early stages.

Many infrastructures have yet to be built, much labor has yet to be trained, and many opportunities have yet to be realized.

But the direction is already very clear.

Artificial intelligence is becoming the foundational infrastructure of the modern world.

And the choices we make today, the speed of our build, the breadth of our engagement, the responsibilities we assume, will determine what kind of age we live in.

[Original Article]