According to recent TrendForce research, the rising demand for artificial intelligence (AI) servers is significantly shaping the strategies of leading North American cloud service providers (CSPs). These major CSPs—likely including Amazon Web Services (AWS), Microsoft Azure, Google Cloud, and Meta—are rapidly advancing their in-house development of application-specific integrated circuit (ASIC) chips to meet AI workloads' increasing performance and efficiency needs. These companies release upgraded versions of their proprietary ASIC chips roughly every 1 to 2 years, driven by the urgent need to enhance AI server processing power while controlling costs and power consumption. By designing their own ASICs, these CSPs aim to reduce dependence on external suppliers, improve system integration, and tailor performance to AI-specific tasks. Meanwhile, the AI server market in China is undergoing a notable transition. TrendForce projects a sharp decline in the market share of externally sourced chips from global suppliers like NVIDIA and AMD, from about 63% in 2024 to roughly 42% by 2025. This change reflects the growing capabilities and market influence of local Chinese chip manufacturers, bolstered by strong government policies promoting technological self-sufficiency and innovation within the semiconductor sector. Through these initiatives—including funding and supportive regulations—China’s domestic AI chip suppliers are expected to expand their market share to nearly 40% by 2025, nearly matching that of foreign vendors. These divergent trends in North America and China illustrate broader geopolitical and economic dynamics influencing the global semiconductor industry.

North American CSPs emphasize proprietary chip solutions to maintain technological leadership and security in AI infrastructure, while China accelerates its drive for domestic chip production to lessen reliance on foreign technology amid escalating trade tensions and export controls. The simultaneous intensification of in-house ASIC development in North America and the rapid growth of indigenous Chinese chip suppliers are transforming the global AI hardware ecosystem. This fosters stronger competition, promotes innovation, and contributes to a more diversified supply chain for AI computing components. AI servers are vital for supporting applications such as machine learning model training, inference, and large-scale data processing, requiring specialized hardware accelerators. ASIC chips, optimized for these specific workloads, outperform general-purpose CPUs and GPUs by boosting performance and reducing energy consumption. The frequent introduction of updated ASICs by North American CSPs reflects the swift advancement of AI algorithms and the increasing complexity of models that demand continual hardware improvements. Meanwhile, the rise of domestic Chinese chip makers signals the maturing competitiveness of China’s semiconductor design industry, offering credible alternatives to established global suppliers. Looking forward, these developments may lead to a more decentralized and diversified AI chip market, with regional innovation and production hubs emerging to meet localized demands and comply with diverse regulatory environments. Such shifts could profoundly affect global technology supply chains, economic relations, and the overall pace of AI progress worldwide. In summary, TrendForce’s findings highlight a pivotal moment in the AI server market: accelerated in-house ASIC innovation by North American CSPs combined with the strong expansion of China’s domestic chip suppliers are driving major shifts in market shares and technological capabilities. These dynamics are ushering in intensified competition, enhanced innovation, and a realignment of the global semiconductor landscape that will underpin the future of artificial intelligence.