The processor market continues to grow significantly, driven primarily by the growing demand for generative AI applications. The market is expected to nearly double between 2024 and 2030, from $288 billion to $554 billion, driven by the widespread adoption of generative AI by businesses, individuals, and governments. 

2024 marks a turning point in the processor industry, with the GPU market exceeding the APU market for the first time. This shift stems from the need for high computing power, especially in the server sector, to run large language models (LLMs) such as ChatGPT, Gemini, and Copilot. The GPU market will face increasing competition from AI ASIC chips developed by hyperscalers such as Google and AWS, which are expected to grow rapidly over the next five years, with the main goal of reducing the current huge capital expenditure costs. Behind the server processor market, edge AI is rapidly expanding in APUs and consumer CPUs. The goal is clear: develop native AI in all electronic devices to provide the best user experience. Smartphones and laptops are at the forefront of embedded AI development, but new types of devices may also emerge.

Giants and new entrants reshape the processor industry

The processor market is highly concentrated in a few companies. Three of the five processor segments are dominated by a single company with more than 50% market share, including Intel with 66% of the CPU market and Nvidia with more than 90% of the GPU market. In contrast, the APU and AI ASIC & DPU markets are more fragmented, with more intense competition from a wide range of companies including Apple, Qualcomm, MediaTek, Google, Samsung, Huawei, NXP, Texas Instruments, and others.

In a highly competitive market, new players have also emerged from China, some of which have been successful, such as Xiaomi's success in smartphone APUs and NIO's success in automotive ADAS APUs. The diversity of APU market players can be explained by the diversity of target end markets. While the GPU market focuses primarily on servers, desktops, and workstations, the APU market covers smartphones, laptops, smartwatches, smart speakers, smart TVs, automotive, VR, and many other segments detailed in this report, and generally has established leaders in each market.

To dominate the market, processor vendors have a common goal of offering the most powerful solutions ahead of their competitors. To achieve this goal, each market faces its own challenges, both in terms of cost and in terms of physical and technological limitations. The common trend across all segments is to move year by year toward more advanced technology nodes, which were once limited to smartphone APUs but are now gradually being adopted by server CPUs and other CPUs. In this technological race, foundries play a central role. Without them, there would be no iPhones or Nvidia GPUs. Over the past 20 years, the number of foundries capable of producing the most advanced nodes has decreased by a factor of 10, and the transition to 2nm processes is likely to reduce this number further. TSMC's technological monopoly is at the heart of major geopolitical tensions, highlighting the need for Intel and Samsung to remain competitive.

The data center processor market is soaring

The data center processor market is expanding rapidly, driven by the growing demand for generative AI applications that require high-performance computing. The global data center processor market reached $147 billion in 2024 and is expected to reach $372 billion by 2030. GPUs and AI application-specific integrated circuits (ASICs), which are at the core of generative AI, largely drive the market and maintain double-digit growth rates.

CPUs and network processors (such as DPUs) are also crucial in this market and are growing steadily. In the AI field dominated by GPUs and AI application-specific integrated circuits (ASICs), FPGAs have experienced a sharp decline and are expected to remain stable in the medium term. The rapid expansion of the market for cryptocurrencies such as Bitcoin has driven the strong growth of crypto ASICs in crypto mining farms, which are essential for verifying cryptocurrency transactions.

Since 2022, Generative AI driven by OpenAI has transformed the data center processor market and greatly benefited Nvidia's GPUs. Faced with Nvidia's dominance and the strategic interests represented by artificial intelligence, hyperscale computing platforms such as Google and AWS are establishing partnerships with Broadcom, Marvell and Alchip to jointly design their own AI ASIC chips for greater autonomy.

According to data, the CPU server market, led by Intel and AMD, which together account for 80% of the market share, remains critical for general computing, but is under pressure. Arm-based CPUs from hyperscalers and new players are gaining momentum, especially Amazon's Graviton, Google's Axion and Nvidia's Grace, all of which claim to have significant energy efficiency advantages. The CPU market is expected to reach $35.6 billion by 2030.

Smaller market segments are also growing, as detailed in Yole Group's Generative AI 2025 report:

• Server FPGAs: $1.5 billion by 2030,

• DPUs and Networking ASICs: $17.7 billion by 2030,

• Crypto ASICs: $4.2 billion by 2030, influenced by changing mining dynamics.

In the wave of transition to AI ASIC chips, many startups such as Groq, Cerebras and Graphcore are seeking market position in innovative ways, triggering a wave of mergers and acquisitions and financing. This pursuit of performance efficiency is driving the transition to ARM-based CPUs, breaking Intel and AMD's long-standing lead in x86 architecture. With expertise in thermal solutions and high power capacity, cryptocurrency mining farms are now entering the AI market by deploying the most powerful GPUs.

Yole Group analysts point to consolidation and M&A as key drivers of computing innovation. In 2024 alone, Yole Group analysts have identified the following key announcements:

• SoftBank acquires Graphcore,

• AWS invests $700 million in Tenstorrent,

• Insurgency and Sapeon merge in South Korea,

• Meta attempted to acquire Furiosa for $800 million but failed.

These developments highlight the scarcity of competitive AI chip teams and the rising value of silicon expertise in AI infrastructure.

Finding new ways to break through

Chiplets play a vital role in GPUs, CPUs, and ASICs, optimizing yields and enabling increasingly larger chips with more advanced nodes. In 2024, the latest CPUs will use 3nm processes, while GPUs and AI ASICs will still use 4nm processes, although 3nm processes are expected to arrive as early as 2025 with the launch of AWS Trainium 3. To meet AI needs, computing performance has increased 8 times since 2020 and continues to accelerate, with Nvidia announcing that its Rubin Ultra will be launched in 2027 with FP4 inference speeds of 100 PetaFLOPs. 

However, as AI models become larger and the demand for low latency and high bandwidth continues to increase, memory plays a vital role in AI applications. HBM memory currently plays this key role in Nvidia, AMD, Google, and AWS solutions, but many AI ASIC startups, such as Groq and Graphcore, are working to build processors based on SRAM memory to improve performance.

As AI becomes a key asset in global digital strategies, governments are investing in dedicated AI data centers to ensure national computing capabilities. At the same time, the U.S. government continues to implement strict export controls, dividing the world into different regulatory tiers and restricting China's access to cutting-edge AI chips.

In response, the Chinese government is accelerating the development of its domestic semiconductor industry, while Nvidia is committed to developing chips that meet export standards. At the same time, Huawei is stepping up the development of its CPUs and AI ASICs, highlighting the strategic urgency of AI computing self-sufficiency.

"Strategic computing becomes the core of AI infrastructure," affirmed Adrien Sanchez of Yole Group.

Source: Content compiled from Yole

Reference link

https://www.yolegroup.com/press-release/generative-ai-at-the-core-of-a-372-billion-data-center-processor-revolution/

View more at EASELINK