NVIDIA

NVIDIA Corporation is an American accelerated-computing and AI infrastructure company whose technologies span GPUs, CPUs, networking, interconnects, complete computing systems and software. Founded as a PC graphics company in 1993, NVIDIA has evolved into a major provider of the infrastructure used for artificial intelligence, scientific computing, gaming, visualization, robotics and autonomous systems.

Last updated: August 17, 2026

About NVIDIA

NVIDIA is headquartered in Santa Clara, California, and trades on Nasdaq under the ticker NVDA. The company was founded by Jensen Huang, Chris Malachowsky and Curtis Priem, with Huang continuing to serve as president and CEO.

Although NVIDIA remains strongly associated with graphics processing units, the modern company operates across a much larger computing stack. Its technologies combine processors, high-speed interconnects, networking, integrated systems and software designed for workloads that benefit from large-scale parallel computing.

NVIDIA's current position is distinctive because it increasingly competes at the level of the complete accelerated-computing system, rather than selling individual processors in isolation.

Its fiscal 2026 Form 10-K describes NVIDIA as a data-centre-scale AI infrastructure company and identifies artificial intelligence, scientific computing, gaming, professional visualization, robotics and automotive computing among the major applications for its platforms.

What Does NVIDIA Do?

NVIDIA designs computing platforms for workloads that require large amounts of parallel processing.

Its core activities span several areas:

• designing GPUs and other processors;

• developing high-speed interconnect technologies;

• providing InfiniBand and Ethernet networking;

• building integrated AI and accelerated-computing systems;

• developing CUDA and other software platforms;

• supporting gaming and professional graphics;

• providing automotive and robotics computing platforms;

• developing simulation and physical-AI technologies.

The GPU remains central to NVIDIA's technology, but it is no longer sufficient to explain the business.

A modern AI workload can require processors, interconnects, networking, memory movement, system architecture and software to work together efficiently. NVIDIA increasingly develops technologies across each of those layers.

Who Uses NVIDIA Technology?

NVIDIA serves a broad range of customers and users.

NVIDIA reports customers and partners across major cloud platforms, enterprises, AI developers, startups and public-sector organizations.

NVIDIA Business Model

NVIDIA follows a fabless, platform-based computing model.

The company designs processors, systems, networking technologies and software but relies on external manufacturing partners for semiconductor fabrication, packaging, assembly and testing.

This allows NVIDIA to concentrate its internal resources on architecture, product development, systems engineering and software instead of operating leading-edge semiconductor fabrication plants.

Its business model also extends beyond selling hardware.

CUDA, development libraries, enterprise software and application frameworks make NVIDIA hardware easier to use across different workloads. Networking and interconnect technologies allow the company to address performance problems that appear when computing expands from one processor to large clusters.

The relationship can be simplified as:

processors → interconnects → networking → systems → software → applications

NVIDIA increasingly designs several of these layers together.

Hardware and Software Reinforce Each Other

NVIDIA's hardware provides computing capability, while software gives developers practical ways to use it.

CUDA is particularly important because applications, libraries and existing development workflows can influence which hardware platforms organizations choose.

A larger installed base can encourage additional software support, while a larger software ecosystem can make the underlying hardware more useful.

This relationship is a strategic characteristic of NVIDIA's business model rather than a separately reported revenue metric.

Does NVIDIA Manufacture Its Own Chips?

NVIDIA designs its processors but does not manufacture them in company-owned leading-edge semiconductor fabs.

It relies on external partners for processes including fabrication, advanced packaging, assembly and testing.

The model lets NVIDIA focus heavily on design and software, but it also creates dependence on external manufacturing capacity and supply-chain execution.

How Does NVIDIA Make Money?

NVIDIA generates most of its revenue from computing and networking products, with Data Center now representing the largest part of the business.

For fiscal 2026, NVIDIA reported total revenue of $215.9 billion.

Data Center therefore represented roughly nine-tenths of fiscal 2026 revenue.

Gaming remains a significant NVIDIA business, but it no longer explains the overall economic scale of the company.

Data Center and AI Infrastructure

Data Center includes both computing and networking products.

In fiscal 2026, NVIDIA reported approximately $162.4 billion in Data Center compute revenue and $31.4 billion in Data Center networking revenue.

Networking is therefore not a minor accessory to NVIDIA's accelerator business. It has become an important part of the infrastructure required to connect large AI systems.

For the first quarter of fiscal 2027, ended April 26, 2026, NVIDIA reported $81.6 billion in total revenue, including $75.2 billion from Data Center.

Gaming, Professional Visualization and Automotive

Gaming generates revenue primarily through GeForce graphics products and related technologies.

Professional Visualization serves designers, engineers and creators using workstation-class graphics and accelerated computing.

Automotive includes computing and software used in automated-driving development and related vehicle platforms.

These markets are smaller than Data Center by revenue but remain strategically relevant because they extend NVIDIA's architecture across consumer, professional and physical-computing environments.

A Note on NVIDIA's Reporting Categories

NVIDIA uses several different reporting frameworks, and they should not be treated as interchangeable.

For fiscal 2026, its two operating segments were:

• Compute & Networking

• Graphics

The company also reported revenue across end markets including Data Center, Gaming, Professional Visualization, Automotive, and OEM and Other.

Beginning in fiscal 2027, NVIDIA announced a transition toward a new market-reporting framework built around:

• Data Center

• Edge Computing

Data Center will be further divided into Hyperscale and ACIE, covering AI Clouds, Industrial and Enterprise. Edge Computing includes areas such as gaming systems, workstations, robotics and automotive platforms.

Products and Services

NVIDIA's product portfolio is easier to understand as a technology stack than as a list of individual product names.

GeForce RTX

GeForce is NVIDIA's major consumer graphics platform.

It primarily serves PC gamers and creators but increasingly also supports AI workloads that can run locally on consumer PCs.

GeForce remains important to NVIDIA's brand and technology ecosystem even though gaming now represents a much smaller share of company revenue than Data Center.

Blackwell

Blackwell is NVIDIA's current-generation platform for large-scale AI and accelerated computing.

Rather than being positioned only as a new GPU, Blackwell combines processors with technologies including interconnects, networking and system-level infrastructure.

Vera Rubin

Vera Rubin represents the next generation of NVIDIA's AI infrastructure roadmap.

NVIDIA describes Rubin as a co-designed platform incorporating CPUs, GPUs, networking, interconnect technologies and other infrastructure components.

In May 2026, NVIDIA said Vera Rubin was ramping into full production. Its Q1 fiscal 2027 filing states that Rubin shipments are expected to begin in the second half of fiscal 2027.

CUDA

CUDA is NVIDIA's parallel-computing platform and programming model.

Introduced in 2006, CUDA allowed developers to use NVIDIA GPUs for workloads beyond traditional graphics, including scientific computing and machine learning.

Its importance to NVIDIA comes from the developer tools, libraries and software ecosystem built around the platform.

Networking and Systems

NVIDIA's networking portfolio includes technologies derived from its Mellanox acquisition, along with NVLink, InfiniBand, Ethernet networking and BlueField products.

These technologies matter because large AI workloads can span many processors and servers.

As systems scale, data movement and communication become part of the performance problem rather than a separate infrastructure concern.

DRIVE, Omniverse and Physical AI Platforms

NVIDIA DRIVE supports automated-driving development.

Omniverse provides simulation and real-time 3D technologies relevant to industrial digital twins and physical-AI development.

These products extend NVIDIA's platform beyond conventional data centres into vehicles, robots and other systems that interact with the physical world.

Customers and Markets

NVIDIA operates across both B2B and B2C markets.

Its largest business increasingly serves enterprise and infrastructure customers rather than consumers.

Cloud Providers and AI Companies

Major cloud platforms, AI developers and model providers use NVIDIA computing and networking infrastructure to train and run AI workloads.

These customers are particularly important to the Data Center business.

Enterprise and Industrial Customers

Companies use NVIDIA technology for enterprise AI, simulation, data processing, engineering and scientific computing.

Manufacturing and industrial customers are also part of NVIDIA's physical-AI and digital-twin strategy.

Consumer Gaming and Creators

GeForce serves consumers, gamers and digital creators.

Although gaming represents a smaller share of NVIDIA's current revenue than Data Center, it remains an important market and a major part of NVIDIA's consumer brand.

Automotive and Robotics

NVIDIA serves automotive manufacturers, autonomous-driving developers and robotics companies through its computing, simulation and edge platforms.

These markets are strategically important but remain less established financially than the current Data Center business.

Competitors

NVIDIA competes across several different layers of the computing market.

Its competitors therefore cannot be reduced to one list of GPU manufacturers.

AI Accelerators and GPUs

NVIDIA competes with semiconductor companies developing GPUs and other accelerators for AI, data-centre and high-performance-computing workloads.

AMD and Intel are among the established semiconductor companies competing in parts of these markets.

Custom AI Chips

Large cloud companies and technology firms increasingly develop custom accelerators for particular workloads.

Some of these organizations may use NVIDIA hardware while simultaneously building internal alternatives.

That makes certain major customers potential competitors in selected parts of the infrastructure stack.

Networking and Interconnects

NVIDIA also competes in high-performance networking and interconnect technologies.

As AI systems grow larger, networking can influence overall system performance, making this an increasingly important competitive area.

Software Ecosystems

Competition extends beyond hardware.

Alternative programming frameworks, software stacks and hardware-portable development environments can reduce the importance of a single processor ecosystem if they become easier or more widely adopted.

Company History

NVIDIA's development is best understood through a small number of strategic turning points rather than every product generation.

CUDA Was the Critical Platform Shift

CUDA was one of the most important changes in NVIDIA's history because it expanded the role of the GPU beyond graphics.

Developers could use GPU parallelism for scientific computing, numerical workloads and machine learning.

The long-term effect was not just a new software product. It helped transform NVIDIA from a graphics-hardware company into a programmable computing-platform company.

Mellanox Expanded NVIDIA Into Networking

NVIDIA completed its acquisition of Mellanox Technologies in 2020 for approximately $7 billion.

As AI workloads expanded from individual accelerators to large clusters, high-performance networking became increasingly important.

The acquisition gave NVIDIA more control over the communication layer connecting large computing systems.

Blackwell and Rubin Mark the Infrastructure Era

Blackwell and Vera Rubin reflect the latest stage of NVIDIA's evolution.

Instead of optimizing only the processor, NVIDIA increasingly designs processors, interconnects, networking, systems and software together.

This has moved the company closer to competing at the level of the entire AI data-centre architecture.

Leadership

Jensen Huang

Jensen Huang is NVIDIA's co-founder, president and CEO and has led the company since its founding.

That continuity is unusual for a technology company that has moved through several major computing eras.

CUDA, accelerated computing, data-centre expansion and NVIDIA's current AI infrastructure strategy all developed under the same CEO.

Founders

NVIDIA was founded by Jensen Huang, Chris Malachowsky and Curtis Priem.

Malachowsky and Priem are historically important to the company's formation, while Huang remains the founder most directly involved in current corporate leadership.

Ownership, Subsidiaries and Corporate Structure

NVIDIA is a publicly traded company listed on Nasdaq under the ticker NVDA.

Its business is reported through major operating segments and market categories rather than as a collection of independently operated consumer brands.

For fiscal 2026, the company's two operating segments were:

• Compute & Networking

• Graphics

NVIDIA also owns technologies and businesses acquired through previous transactions, including the networking capabilities added through its acquisition of Mellanox Technologies.

The company follows a fabless operating structure, meaning semiconductor manufacturing is largely performed through external foundries and manufacturing partners rather than company-owned leading-edge fabs.

This distinction is central to NVIDIA's corporate structure: it controls architecture, product design and software while depending on a wider manufacturing ecosystem to physically produce its processors and systems.

NVIDIA Strategy

NVIDIA's current strategy centres on expanding accelerated computing from individual processors into complete infrastructure.

Full-Stack AI Infrastructure

The company increasingly designs multiple layers of an AI system together.

That includes:

• GPUs and CPUs;

• interconnects;

• networking;

• rack-scale systems;

• software libraries;

• enterprise AI tools.

The strategic objective is broader than selling faster accelerators. NVIDIA aims to optimize more of the environment in which those accelerators operate.

AI Factories

NVIDIA uses the term AI factory to describe data-centre infrastructure designed to train models and run AI services at large scale.

The term is NVIDIA's own positioning, but it reflects the company's move toward selling and designing larger portions of the computing environment.

Vera Rubin is being positioned around this rack- and data-centre-scale approach.

Agentic AI

NVIDIA is positioning its newer infrastructure around AI systems capable of reasoning and performing multi-step tasks.

The strategic opportunity for NVIDIA is that increasingly compute-intensive inference could create demand for larger amounts of infrastructure after models have been trained.

Commercial adoption and long-term market size remain uncertain.

Physical AI and Robotics

NVIDIA is also extending its platform into robots, autonomous systems and industrial machines.

Its strategy combines:

1.                        data-centre training;

2.                        simulation;

3.                        AI software development;

4.                        deployment on edge or embedded computing systems.

Physical AI is an important strategic direction for NVIDIA, but it remains less mature commercially than its established Data Center business.

Key Challenges and Risks

NVIDIA's size and current AI position do not eliminate significant operational and competitive risks.

Manufacturing and Supply-Chain Dependence

NVIDIA depends on external suppliers for semiconductor manufacturing, advanced packaging, assembly and testing.

Demand for NVIDIA products can therefore exceed the manufacturing capacity available to produce finished systems.

Its SEC filings identify supplier concentration, manufacturing capacity and complex product transitions as material risks.

Competition From Alternative Accelerators

NVIDIA competes with other GPU suppliers, specialized AI accelerators and custom chips developed by cloud companies.

Customers may use NVIDIA products for some workloads while deploying different hardware for others.

Software Competition

CUDA is a major strategic advantage, but software ecosystems can change.

Alternative frameworks and hardware-portable development environments could reduce the switching cost between accelerator platforms.

Data-Centre Power and Infrastructure Constraints

Large AI systems require substantial physical infrastructure.

NVIDIA has identified data-centre availability, energy supply and access to capital as factors that can affect customer deployments.

Demand for processors does not automatically create the electricity, buildings and network infrastructure required to operate them.

Export Controls and Regulation

Export restrictions affect where NVIDIA can sell certain advanced Data Center products.

The company excluded China Data Center compute revenue from its Q2 fiscal 2027 outlook, demonstrating how regulation can directly affect parts of its addressable market.

Rapid Product Transitions

NVIDIA operates on an aggressive product roadmap.

Moving customers from one architecture to another requires coordination across chips, networking, systems, software and manufacturing partners.

The broader the platform becomes, the more layers NVIDIA has to execute successfully at the same time.

Key Facts About NVIDIA

• NVIDIA's largest business is now Data Center rather than gaming.

• Data Center generated about $193.7 billion of NVIDIA's $215.9 billion fiscal 2026 revenue.

• NVIDIA follows a fabless semiconductor model and depends on external manufacturing partners.

• CUDA, introduced in 2006, helped transform NVIDIA GPUs from graphics processors into general-purpose accelerated-computing platforms.

• NVIDIA's 2020 Mellanox acquisition added networking capabilities that became increasingly important as AI systems grew larger.

• NVIDIA increasingly develops processors, interconnects, networking, systems and software as one computing platform.

• Some of NVIDIA's largest customers are also developing alternative AI accelerators.

• NVIDIA's competitive position depends on more than GPU performance; software, networking and developer adoption are also important.

• Blackwell and Vera Rubin reflect NVIDIA's shift toward rack-scale and data-centre-scale AI infrastructure.

• Jensen Huang has led NVIDIA since the company was founded in 1993.

NVIDIA Case Studies

Explore upcoming WeCaseStudy case studies examining NVIDIA's business growth, AI infrastructure strategy, CUDA ecosystem, Mellanox acquisition, Blackwell platform and other major strategic decisions.

Frequently Asked Questions

Is NVIDIA a hardware or software company?

NVIDIA operates across both hardware and software.

Its hardware includes processors, networking and integrated computing systems, while its software includes CUDA, development libraries and enterprise AI tools.

What is NVIDIA best known for today?

Consumers still strongly associate NVIDIA with GeForce graphics cards, but Data Center is now its largest business by a wide margin.

At the corporate level, NVIDIA is increasingly known for accelerated computing and AI infrastructure.

Why is CUDA important to NVIDIA?

CUDA allows general-purpose computational workloads to use NVIDIA GPU parallelism.

Its software tools and developer ecosystem helped NVIDIA expand from graphics into scientific computing, machine learning and AI.

Does NVIDIA manufacture its own chips?

NVIDIA designs its processors but uses external manufacturing partners for semiconductor fabrication, packaging, assembly and testing.

It therefore operates primarily as a fabless semiconductor and computing-platform company.

Is NVIDIA still mainly a gaming company?

No, not by revenue.

Gaming remains a major NVIDIA business, but Data Center accounted for roughly nine-tenths of fiscal 2026 revenue.

Who founded NVIDIA?

NVIDIA was founded in 1993 by Jensen Huang, Chris Malachowsky and Curtis Priem.

Huang continues to serve as president and CEO.

Resources:

• NVIDIA fiscal 2026 Form 10-K

• NVIDIA Q1 fiscal 2027 Form 10-Q

• NVIDIA Q1 fiscal 2027 results

• NVIDIA corporate timeline

• NVIDIA Jensen Huang biography

• NVIDIA CUDA Programming Guide

• NVIDIA Blackwell platform announcement

• NVIDIA Vera Rubin production announcement