Just now, AMD's market capitalization has exceeded the $1 trillion mark for the first time in history.

On Monday, AMD's market capitalization surpassed the $1 trillion mark for the first time, making it one of the few chipmakers to reach this milestone as investors bet on its expanding role in artificial intelligence computing.

As of press time, AMD's share price has risen 9.6% to $613.31, hitting an all-time high.

AMD's stock price has risen for five consecutive days, with a cumulative increase of about 25%. For investors, the recent rebound is undoubtedly a huge boost. Last month, AMD released its second-quarter earnings report that beat expectations, but its performance outlook fell short of some investors' expectations, leading to a drop in share price.

The company's second-quarter revenue reached 11.54 billion U.S. dollars, a 50% increase from 7.69 billion U.S. dollars in the same period last year, with artificial intelligence chips being the core driving force for growth. Sales from AMD's data center segment hit 6.7 billion U.S. dollars, a sharp 107% increase compared to the same period last year.

This chipmaker has delivered exceptional performance this year, with its share price rising by more than 180%. However, despite the staggering increase, AMD still lags far behind NVIDIA, the leader in AI chips, which has a market capitalization of about 5.4 trillion U.S. dollars and occupies most of the AI data center chip market share.

Despite strong public opposition to data centers and concerns about the underlying security of large language models, the AI construction that has driven the surge in chip stocks this year shows little sign of slowing down.

Lisa Su, CEO of AMD, said on the company's earnings call last month that she expects data center sales to double by 2027. Last week, Jensen Huang, CEO of NVIDIA, shared his views on chip demand, saying he expects the company's chip sales to double next year.

This milestone rally has drawn a successful full stop for the Santa Clara, California-based company.

AMD is widely seen as the biggest challenger to AI leader NVIDIA. It becomes the fourth U.S. chipmaker to top the $1 trillion market capitalization, following NVIDIA, Broadcom and Micron.

NVIDIA crossed this milestone in 2023 and is now the most valuable company in the world.

"Capital is flowing back into AI trades," said Thomas Hayes, chairman of New York-based Great Hill Capital LLC.

In recent months, investors' enthusiasm for chipmakers has cooled as the market pays close attention to the spending of hyperscale data center operators in the AI field. Rising oil prices linked to the US-Iran conflict and market expectations that interest rates will remain elevated for a long time have exacerbated this pressure.

Hayes said that now investors believe AI is the only segment that can maintain growth amid the economic slowdown triggered by the Federal Reserve. "AMD is a typical example."

Over the past year, AMD has accelerated the release of AI products, no longer just selling individual chips, but providing complete systems — combining processors, network devices and related hardware — to compete with NVIDIA.

On the other hand, the growing demand for central processing units used in servers to handle inference in tandem with graphics processing units has helped AMD capture market share from Intel.

AMD forecast last month that its quarterly revenue would top Wall Street expectations, but still fall short of investors' high hopes.

However, the stock is set to rise 185% in 2026, far outperforming the tech-heavy Nasdaq Index and placing it among the top performers in the S&P 500. AMD is currently trading at about 41 times its 12-month expected earnings — lower than its 10-year average of 44 times, but much higher than NVIDIA's recent trading level of 16.3 times expected earnings.

As AMD's rival, Intel saw its share price rise even more, up 12.03% to close at $121.67. Arm Holdings' share price rose 15.47%, but neither came close to AMD's closing gain. Intel has a market capitalization of $639.32 billion, and Arm's market capitalization stands at $339.93 billion.

An enterprise rising with the AI wave

AMD's success is largely driven by the booming development of artificial intelligence.

This did not happen overnight. AMD entered the AI field relatively late, after it had devoted most of its GPU development resources to more traditional high-performance computing applications and national supercomputing projects. That changed in 2023, when AMD launched the Instinct MI300A — a graphics card that, on paper at least, offers higher performance, larger memory and greater bandwidth than NVIDIA's then-new H100 and H200 series GPUs.

While AMD's products performed better on paper, their software optimization was insufficient, as much of the software was originally designed for NVIDIA accelerators from the very beginning.

Over the past three years, AMD's GPU team has been working hard to improve ROCm, supported by large-scale deployments of MI300 and MI350 series GPUs by companies including Microsoft, OpenAI, Oracle, Anthropic and Meta, with lower prices and higher memory capacity serving as key differentiators.

Key partnerships with Neoclouds, as well as smaller cloud service providers such as TensorWave and Vultr, have helped the open source community understand AMD's accelerators.

AMD claims that by mid-2026, it will not only close the performance gap with NVIDIA, but also launch a new rack-scale computing platform in the third quarter that is larger and faster than NVIDIA's existing Blackwell-based rack systems and Vera Rubin.

Compared with NVIDIA's own state-of-the-art AI systems, AMD's Helios rack server has 50% higher HBM4 memory and scale-out bandwidth, and its AI training performance is improved by 15% to 25%. Moreover, if AMD's claims are true, Helios will be 30% more cost-effective than its rivals. While this figure is difficult to verify, it clearly signals to Wall Street why the largest and most influential AI labs and hyperscale data centers are flocking to this platform.

However, GPUs are not the only reason for the surge in AMD's valuation in recent years — they are just the most noticeable. In February this year, Aaron Rakers of Wells Fargo Securities predicted that by the fourth quarter of 2025, sales of Instinct GPUs would account for 2.5 billion to 2.6 billion U.S. dollars of AMD's total revenue of 10.3 billion U.S. dollars, roughly equivalent to the revenue of Epyc CPUs.

The rise of agent AI workloads such as OpenClaw and Claude Code has boosted demand for CPUs. While the models themselves still run on GPUs, the code they generate and the tools they execute still require CPUs to function. Competition in this segment is more intense. In addition to long-time rival Intel, AMD also has to compete with Arm, Qualcomm, NVIDIA and cloud service providers.

But as we have seen from the Instinct processors, AMD's Epyc processors have always been known for high cost-effectiveness, and their cost-performance ratio has been higher than that of competitors in recent generations, which has helped the chip design company significantly expand its market share.

In August, AMD's market share in the desktop CPU segment exceeded 35%. As we wrote in our reviews earlier this year, AMD now produces the most powerful desktop CPUs in the world, and has few rivals in the workstation space, thanks in part to repackaging unlocked Epyc processors into high-core-count Threadripper processors.

In the data center segment, AMD's market share jumped to 34.5%, while Arm-based CPUs from companies including Ampere, Amazon, Microsoft and Google accounted for another 13.6% of the market share. As Mercury Research, the analytics firm that provides these figures, noted, if only Epyc and Intel Xeon SP processors are compared, AMD's market share rises to 46.4%.

These gains have been fueled by Intel's struggles to launch competitive Xeon processors. Since Intel's Sapphire Rapids processors underperformed and gave AMD nearly two years to catch up, Intel has been working hard to narrow the gap.

Intel came close to success in 2024 with its Xeon 6 processors, which remain very popular as host CPUs for NVIDIA's small GPU systems, but executive shakeups and the cancellation of its mainstream Xeon 7 SP product line mean Intel is unlikely to have a competitive data center CPU until 2028.

AMD faces some pressure from NVIDIA's Vera and Arm AGI CPUs, but this is mostly concentrated in AI-centric applications such as GPU head nodes, agent sandboxes and key-value cache offload servers. With a modular design, AMD can support a wider range of target applications, thus covering a broader market.

But AMD's market share growth is not limited to CPUs and GPUs. Over the past decade, AMD has accumulated a rich technology portfolio through acquisitions, covering smart network interface cards, data processing units (DPUs), field-programmable gate arrays (FPGAs) and even system design. AMD's acquisition of Xilinx is particularly notable: although this acquisition is not its main source of revenue, it has made AMD the largest FPGA supplier in the market and provided the chip design company with a huge amount of intellectual property that keeps it highly competitive. For example, AMD's NPU (neural processing unit) technology is directly derived from Xilinx's intellectual property, and it is these technologies that ensure AMD obtains Microsoft Copilot+ certification.

Just as AMD has become an effective tool to put pressure on Intel's CPU prices, NVIDIA is now playing a similar role in the data center GPU space. When AMD released the Helios graphics card in July, it announced that Anthropic had agreed to deploy up to 2 gigawatts of Instinct accelerators to support AI training and inference. OpenAI, Oracle, Microsoft and Meta also plan to deploy these rack systems on a large scale.

"The growing number and scale of Helios and MI450 series deployments in the data center AI segment will drive significant growth in our (data center) business in the second half of this year, and accelerate growth in 2027," CEO Lisa Su told investors last month.

She added: "We now expect that by 2027, data center business revenue will more than double from the previous year."

However, although AMD's executives have remained optimistic, Wall Street's views on the company are still mixed, with the main concern being that AMD may put too many of its AI eggs in too few baskets.

On the other hand, AMD's product portfolio covers general-purpose computing, high-performance computing, artificial intelligence, embedded systems, desktop and gaming segments. Although its scale is far smaller than NVIDIA's, AMD is also much less exposed to the AI bubble.

This article is from the WeChat official account "Semiconductor Industry Watch", and is authorized for release by 36Kr.