NVIDIA ARM chips gain fast space on AI boom servers and $ 95 billion market

The global server market recorded historical growth in the first quarter of 2025, reaching nearly $ 100 billion in revenue. Growth is driven by massive investments in Artificial Intelligence Infrastructure (AI), according to IDC.

One of the fastest growing segments is that of accelerated ARM servers, whose shipments increased impressive 70% this year.

And most of these ARM servers are NVIDIA GB200 NVL72 rack scale solutions based on the platform Grace Blackwellwhich combines the CPU nvidia grace with eight gpus b200 by servedr.

Related News:

Record growth

Global server purchases totaled US $ 95.2 billion, an increase of 134.1% over the same period as 2024. This is the highest quarterly growth ever recorded.

With the result, the annual forecast for 2025 was reviewed up: $ 366 billion, a 44.6% jump compared to 2024.

The wide implantation of GPU accelerated AI servers includes those used by hypereschalters. And the Nvidia ARM Grace CPUs contributed to a 70% increase in ARM server shipments this year, compared to the same period of the previous year.

ARM and X86: Architectural dispute

X86 architecture servers should move $ 283.9 billion by 2025, growing 39.9%. Already systems with CPUS ARM and other nonx86 should grow 63.7%reaching US $ 82 billion.

The servers of IA and HPC with GPU should grow 46.7%representing almost half of sector spending. The trend was amplified by the need for enormous computational power to support new AI workloads and training pipelines.

Until the end of 2025, ARM systems are expected to represent 21.1% of the total units of servers sent globally. This is considerably lower than the Long Term expectations of the 50% market penetration ARM.

However, 21.1% is still a huge slice and, considering that most of these CPUs are Nvidia’s Grace processors, it is also great news for the green giant.

Amazing projections

Until 2029, the server market should reach $ 588 billion. Accelerated X86 servers (with AMD / Intel GPUS or AI accelerators) will grow from $ 112 billion to $ 324 billion.

The accelerated ARM servers will triple, from $ 32 billion to $ 103 billion.

It is important to remember that Accelerated Arm machines will be released between 2027 and 2028 can use not only NVIDIA processors, but also CPU designs of the “NVLink Fusion field”, such as Fujitsu, Marly, MediaTek and Qualcomm.

Of course, it is not yet known if they will be able to conquer a significant slice of the market, but they have been considered in some way in predictions. And predictions may not be fulfilled because it is optimistic or by pessimistic.

LLMS Participation

The non -X86 accelerated server segment (including FPGA and Asic) should also grow. The forecast is more modest and the segment should reach $ 31 billion by 2029.

Demand for accelerated AI servers will be driven by more advanced LLMS and LRMs, in addition to speculation that General Artificial Intelligence (AGA) is possible.

AGI would require even more computational performance than current AI technologies, according to IDC. To better understand the difference, we have an article on the site explaining what an AGE would be.

The launch of Deepseek’s R1 reasoning model generated concerns about the need to invest in so much infrastructure. However, Deepseek R1 needed more infrastructure than reported, which calmed the market.

Furthermore, The evolution of simple chatbots for reasoning models and Ai Aggência will require much greater processing capacityespecially for inference.

Traditional servers

At the same time, spending on traditional servers (unfeed X86) will continue to constantly increase from US $ 91 billion to US $ 130 billion. However, they must represent a lower portion of the general market value.

Unfortunately, IDC does not make predictions about the adoption of general -based ARM -based servers, such as ARM’s own, personalized CSPS CPUs of hyperscala CSPs or companies such as AMPERE Computing.

Highlights

The United States lead with growth of 59.7%, representing 62% of global revenue. China continues with a high of 39.5%, accounting for more than 21% of sales. And considering that the United States and China are in an AI arms race, it is unlikely that both countries reduce their spending.

Other highlights go to the Japan (+33.9%) and the rest of Pacific Asia (+10.8%). Europe, the Middle East and Africa (emea) represent +7.0% and Latin America only +0.7%.

The worst result, however, goes to the Canada, with a drop of 9.6%. However, it is important to highlight that, in part, this is due to a very large increase last year, and now the curve corrects itself.

Source: IDC.