How Nvidia became the world’s first $US4 trillion company

Nvidia became the first company in history to reach a market value of $US4 trillion, beating rivals Apple and Microsoft to the milestone.

Nvidia CEO Jensen Huang holds a Nvidia's Drive Thor processor as he delivers a keynote address at the Consumer Electronics Show (CES) in Las Vegas, Nevada in January.

Nvidia became the first company in history to reach a market value of $US4 trillion, beating rivals Apple and Microsoft to the milestone in Wednesday morning trading on the tech-heavy Nasdaq exchange.

The AI chip giant has seen its fortunes surge over the past three years thanks to the rise of generative artificial intelligence, an emerging technology that promises to revolutionize business and remake how humans interact with technology across the globe.

Nvidia, based in Santa Clara, Calif., designs the chips, known as graphics processing units, or GPUs, that power the AI industry. The rally in Nvidia’s shares caps a remarkable run and comes barely two years after the company notched a $1 trillion closing valuation for the first time. The AI chip maker which closed at $162.88 a share and just shy of the $4 trillion mark, is now worth as much as the 214 smallest companies in the S&P 500 combined, according to Dow Jones Market Data.

Apple and Microsoft have come close to the $4 trillion mark. Apple closed with a market cap of $3.915 trillion in late 2024, and Microsoft had a valuation of $3.708 trillion last week.

Founded by Taiwanese-American engineer Jensen Huang in 1993, Nvidia initially had a much more niche purpose: making personal-computer game graphics better. Its ascendance from a graphics-chip designer catering mainly to videogamers into the digital arms dealer at the red-hot center of the AI boom has been meteoric.

(The logo of Nvidia is seen during a production preview exhibition in Taipei.

Earlier this year Nvidia faced one of its biggest challenges ever when DeepSeek, a lower-priced Chinese chatbot, launched after being developed much more quickly and cheaply than most large language models.

DeepSeek, which was developed using about 2,000 Nvidia H800 processors -- significantly fewer than most competitors use -- proved that developers could create powerful AI software more cheaply and without using as much computing power as previously believed necessary. Nvidia’s stock tumbled nearly 20% on that news but has more than recovered on the promise of higher spending by the company’s biggest customers.

In its early years, Nvidia grew with the gaming industry, becoming, alongside Advanced Micro Devices, a market leader. But Huang had bigger ambitions. There was a wider range of computational tasks that he thought his chips could accelerate.

About a decade and a half ago, Nvidia planted the seeds for that to happen by developing software that allowed people to use its chips for purposes other than graphics processing.

The development of cloud computing and an increased focus on scientific computing led Huang’s company to a new frontier: data centers powered by Nvidia’s processors. Their uses expanded from there, including for cryptocurrency mining, where they flourished for a time. It also led to their wide use in machine learning, computer vision and other nascent AI applications.

Starting in the early 2010s, the company began producing increasingly powerful chips, naming each new “architecture,” or design, for famous physicists and other scientists such as James Clerk Maxwell, Johannes Kepler, Alan Turing and Ada Lovelace. For roughly two decades, Nvidia fell into a regular pace of releasing a new generation of chips every two to four years.

More recently, as AI technology has ramped up, that cycle has shortened dramatically -- now, the company says that it aims to maintain a cadence of releasing a new generation of chips every year.

In March 2022, the company released its Hopper generation of chips, which proved a massive hit with AI software developers for its improvements in memory and computing power. Nvidia sold millions of its Hopper H100 chips in 2023 alone, helping power the company past a market value of $2 trillion.

Last year, Nvidia released its newest microarchitecture, Blackwell, which represented its most powerful GPUs yet, including the B200, which crams 208 billion transistors onto a chip roughly the size of four Scrabble tiles arranged in a square.

Demand has been nearly insatiable for these chips, as large customers such as Meta Platforms, OpenAI and Google-owner Alphabet race to accumulate the firepower to develop the most sophisticated chatbots and other large language models ever made.

Any AI chip Nvidia could make was quickly snapped up, even though they cost tens of thousands of dollars each. Tech companies boasted about how many thousands of Nvidia chips they were buying; sprawling data centers were built to house them. Huang took on an almost Steve Jobsian aura -- part-celebrity, part-oracle -- cutting a distinctive figure with his signature black leather jackets.

Nvidia CEO Jensen Huang.

There have been some ups and downs. Both the Biden and Trump administrations have placed strict limits on Nvidia’s ability to sell many of its most powerful AI chips in China, treating them almost as strategic national-security assets.

In October 2023, the Biden White House curbed China sales of the company’s H800 chip, a processor that Nvidia had begun offering just seven months before. Then, in April of this year, the Trump administration limited sales of another chip, the H20, to China, leading Nvidia to take a $5.5 billion write-down.

“With the current export controls, we are effectively out of the China data-center market, which is now served only by competitors such as Huawei,” an Nvidia spokesman said Wednesday.

Nvidia’s revenue rose in tandem with its stock price. Two years ago, it had $7.2 billion of revenue in its May quarter. This year it had $44.1 billion -- an enormous number for a business with a gross profit margin above 70%.

The Wall Street Journal