NVIDIA overtakes Microsoft to become world's most valuable company

NVIDIA took the top spot in the global ranking of publicly traded companies by market capitalization on Tuesday. At the close of trading on Wall Street, the company was valued at about $3.34 trillion — trillion in the US sense — ahead of Microsoft at roughly $3.32 trillion and Apple at around $3.29 trillion.

The milestone was not driven by a new product announcement that day, but by a conviction that has spread across markets in recent months: much of the artificial intelligence race depends on NVIDIA chips. Its shares rose 3.5% during the session and are up nearly 174% since the start of 2024.

From video game graphics to the center of the AI economy

NVIDIA was founded in 1993 and initially built its business around GPUs, the graphics processors used in computers and game consoles. Unlike a conventional CPU, a GPU can perform many simple operations simultaneously. That capability, which is useful for rendering graphics, proved especially valuable for training neural networks.

The current boom is not just about silicon. For years, NVIDIA has developed CUDA, a programming platform that allows researchers and companies to use its GPUs for scientific computing and artificial intelligence. That ecosystem has created an advantage that is difficult to replace overnight: switching suppliers can mean adapting software, libraries and entire data centers.

That is why companies such as Microsoft, Amazon, Google and Meta are buying large quantities of its processors for their data centers. They are also needed by companies developing generative models, from text-based systems to tools that create images, audio or video.

Results that explain the valuation

The company's market value is underpinned by business growth that is unusual even in the technology sector. In the first quarter of fiscal 2025, whose results NVIDIA released on May 22, the company reported $26 billion in revenue, up 262% from a year earlier. Its data center division contributed $22.6 billion, nearly 87% of total revenue.

Quarterly net income was $14.9 billion, more than seven times the figure recorded in the same period of the previous fiscal year. The numbers have changed the company's scale: just two years ago, NVIDIA was known mainly for its graphics cards for gamers and creative professionals; now its data center business dominates its accounts.

The company also carried out a ten-for-one stock split in June, approved in May. The move did not change the company's underlying value, but it lowered the per-share price and made the stock easier for small investors to buy. The main driver, however, remains the expectation that big technology companies will maintain — or increase — their investment in AI infrastructure.

Blackwell sets up the next buying cycle

Current demand has centered on processors such as the H100 and H200, which are widely used to train large language models. NVIDIA unveiled Blackwell, its new data center architecture, in March. The company expects its first products to reach customers this year.

Blackwell combines multiple pieces of silicon to create more powerful computing systems and is designed to reduce the cost and energy consumption of large-scale AI workloads. For data center operators, that efficiency matters as much as performance: running an AI assistant for millions of users requires chips, electricity, cooling and high-speed networks.

The arrival of a new generation also gives NVIDIA a tool for sustaining its sales momentum, although it comes with a requirement: the company must meet its production schedules while customers wait for increasingly capable equipment.

Market leadership does not eliminate the risks

NVIDIA's overtaking of Apple and Microsoft reflects the weight Wall Street assigns to AI infrastructure, not a guarantee that the ranking will hold. Its valuation depends on whether technology companies' multibillion-dollar investments translate into profitable products and services for their own customers.

NVIDIA also faces competition on several fronts. AMD is seeking to expand its presence in AI accelerators; Intel is trying to regain ground in data centers; and major platforms are designing their own chips for specific workloads. Google uses its TPUs, Amazon is developing Trainium and Inferentia, and Microsoft is working on its own accelerators.

In the short term, none of those alternatives has altered NVIDIA's commercial dominance in training the most advanced models. But the industry needs more than computing capacity: it needs that capacity to be profitable, available and less energy-intensive. The position NVIDIA reached today makes it the leading market barometer for the bet on artificial intelligence.