NVIDIA's Jensen Huang's new article: The "Five-Layer Cake" of AI

Original Title: AI Is a Five‑Layer Cake

Original Author: Nvidia

Original Compiler: Peggy, BlockBeats

Editor's Note: Artificial intelligence is gradually evolving from a cutting-edge technology into the infrastructure that supports the operation of the modern economy. In its first long article published on its official account, Nvidia attempts to systematically outline the industrial structure of AI from first principles: from energy and chips to data center infrastructure, and then to models and applications, forming a complete five-layer technology stack.

The article points out that AI is not just a competition of software or models, but a global industrial construction involving energy, computing power, manufacturing, and applications, the scale of which may become one of the largest infrastructure expansions in human history. Through this "five-layer cake" perspective, Nvidia aims to illustrate that the true significance of AI is not just smarter software, but an infrastructure revolution on par with electricity and the internet.

Here is the original text:

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

AI operates on real hardware, real energy, and real economic systems. It transforms raw materials into scalable "intelligence." Every company will use it, and every country will build it.

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

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

For most of the history of computing, software has been "pre-packaged." Humans first describe an algorithm, and then the computer executes it according to instructions. Data must be carefully structured, stored in tables, and retrieved through precise queries. SQL is indispensable because it allows this entire system to function.

AI breaks this model.

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 meanings; it can reason about context and intent. More importantly, it can generate intelligence in real-time.

Every response is a new generation. Every answer depends on the context you provide. It is no longer software retrieving existing instructions from a database, but 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

From an industrial perspective, AI can actually be broken down into a five-layer structure.

Energy

At the bottom layer is energy.

Real-time generated intelligence requires real-time generated power. The generation of each token means electrons are moving, heat is being managed, and energy is being converted into computing power.

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

Chips

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

AI workloads require massive parallel computing capabilities, high-bandwidth memory, and high-speed interconnects. Advances at the chip layer determine the speed of AI expansion and how cheap "intelligence" will ultimately become.

Infrastructure

Above chips is infrastructure. This includes land, power delivery, cooling systems, construction engineering, network systems, and scheduling systems that organize thousands of processors into a single 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 layer is the applications layer, where real economic value is generated. For example, drug discovery platforms, industrial robots, legal copilots, and self-driving cars.

A self-driving car is essentially an "AI application carried by a machine"; a humanoid robot is an "AI application carried by a body." The underlying technology stack is the same; only the final presentation form differs.

Thus, this is the five-layer structure of AI: Energy → Chips → Infrastructure → Models → Applications. Every successful application pulls all levels downwards, all the way to the power plants at the bottom layer that supply it.

An Infrastructure Build Still in Its Early Stages

We have only just begun this construction. The current investment scale is only a few hundred billion dollars, while in the future, infrastructure worth trillions of dollars still needs to be built.

Globally, we are seeing: chip factories, computer assembly plants, AI factories.

An unprecedented scale is being constructed. This is becoming one of the largest infrastructure builds in human history.

Labor Demand in the AI Era

The scale of labor required to support this construction is enormous.

AI factories need: electricians, plumbers, pipe installers, steel structure workers, network technicians, equipment installers, and operations personnel.

These are all highly technical, well-paying positions, and there is currently a severe shortage. Participating in this transformation does not necessarily require a PhD in computer science.

At the same time, AI is driving productivity increases in the knowledge economy. Take radiology as an example. AI has begun to assist in medical imaging interpretation, yet the demand for radiologists continues to grow.

This is not contradictory.

The true responsibility of a radiologist is to care for patients, and reading images is just one of their tasks. As AI takes over more repetitive tasks, doctors can devote more time to judgment, communication, and treatment.

Increased efficiency in hospitals can serve more patients, thus requiring more manpower. Productivity creates capability, and capability creates growth.

What Changes Have Occurred in the Past Year?

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

Models are now good enough to truly function in large-scale scenarios.

• Significant improvements in reasoning capabilities
• Significant reductions in hallucinations
• Substantial enhancements in "grounding" with the real world

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

There is already a clear product-market fit in the following areas: drug development, logistics, customer service, software development, and manufacturing.

These applications are powerfully driving the entire underlying technology stack.

The Role of Open Source Models

Open source models play a key role in this. The vast majority of AI models in the world are free. Researchers, startups, enterprises, and even entire countries rely on open source models to compete in advanced AI.

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

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

What Does This Mean?

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

It is an industrial-level transformation that will reshape:

• The way energy is produced and consumed
• The way factories are built
• The way work is organized
• The patterns 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 intelligence expansion. Energy is becoming central because it determines how much intelligence the system can produce at most. Applications are exploding because models have finally crossed the threshold of "scale availability."

Each layer reinforces the others.

This is why this construction is so vast, why it will simultaneously impact so many industries, and why it will not be limited to any one country or field.

Every company will use AI.

Every country will build AI.

We are still in the early stages.

A lot of infrastructure has yet to be built, a lot of labor has yet to be trained, and a lot of 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 construction, the breadth of participation, and the responsibility of deployment will determine what this era will ultimately become.