By Yang Jianyong

Since the advent of generative AI, technology behemoths have been intensifying their efforts in AI infrastructure development to bolster the competitiveness of their large-scale models, resulting in an exponential surge in demand for computing power. This trend has translated into windfall gains for chip and computing power server manufacturers.

On a global scale, tech giants' heavy reliance on NVIDIA's high-performance chips for model training has positioned NVIDIA as the prime beneficiary. Jensen Huang, NVIDIA's CEO, confidently projects that the company's AI chip revenue will skyrocket to $1 trillion by 2027.

In the domestic arena, Foxconn Industrial Internet and Inspur Information stand out as the leading providers of computing power servers. Foxconn Industrial Internet is poised to surpass the trillion-yuan revenue threshold this year, achieving double-digit growth in both revenue and profitability. The driving force behind this remarkable performance is the global expansion of AI expenditures, which directly benefits computing power service providers and propels them into a new era of development.

In the era of large-scale models, the demand for computing power, particularly AI servers, continues to escalate. As the world's second-largest server provider, Inspur Information has made significant strides in the computing power market, with its net profit surging by 269.6% to 2.95 billion yuan in the first half of 2026.

However, the revenue growth rate lags far behind the profit growth rate, which, to some extent, hampers its long-term development potential. The market not only values growth but also emphasizes the formation of competitive barriers through technological innovation to achieve sustainable and high-quality growth.

To enhance its technological prowess and meet the burgeoning demand for computing power services from large-scale models, Inspur Information plans to issue additional shares to raise 9 billion yuan for further investment in the computing power market. This strategic move will enable the company to seize industry development opportunities and fuel the high-quality growth of its core business.

The raised funds will be earmarked for the R&D and industrialization of a new generation of AI infrastructure tailored for large-scale Agent applications, the R&D and industrialization of liquid-cooled native computing power infrastructure for green smart computing clusters, the R&D of key technologies for multi-Agent collaborative computing and storage integration infrastructure, as well as supplementary working capital. These initiatives received the green light at the extraordinary shareholders' meeting on October 8.

It is worth noting that with the continuous evolution of large-scale models, they have transitioned from initial knowledge Q&A (question-and-answer) sessions to Agent AI, which is now capable of executing tasks. This shift has further driven up Token usage and continuously amplified the overall demand for computing power in the market.

According to IDC data, the number of globally active Agents stood at 28.6 million in 2025, is projected to reach 79.4 million in 2026, and is expected to soar to 2.216 billion by 2030. Under this trend, Agent applications are accelerating their penetration into various industries.

Meanwhile, the large-scale deployment of Agents imposes higher demands on the efficiency of computing power infrastructure, resource scheduling capabilities, and platform application development, deployment, operation, maintenance, and governance. Inspur Information's R&D and industrialization project for a new generation of AI infrastructure for large-scale Agent applications primarily addresses the diverse product demands arising from such deployments.

Crucially, in the Agent era, the focus of AI computing power has shifted from model training to large-scale inference deployment. Token generation rate and usage cost have emerged as core competitive factors in the industry, with leveraging Scale up super-node architecture to enhance inference performance becoming a mainstream trend.

By focusing on super-node development and through the R&D and industrialization of highly scalable super-node rack servers, Inspur Information aims to enhance the customization and large-scale production capabilities of its super-node rack server products. This strategic move will enable the company to seize opportunities in the emerging super-node market and consolidate its technological and product advantages in the high-end computing power infrastructure sector.

Finally, the rapid advancement of AI large-scale model technology has ushered in tremendous development opportunities for various industries, further fueling sustained growth in demand for computing power and increasing the need for efficient AI servers.

The domestic AI server market size was approximately 254 billion yuan in 2025 and is expected to balloon to 1.65 trillion yuan by 2030, with a compound annual growth rate of 45.5%.

It is undeniable that the demand for AI computing power will experience even more rapid growth. For computing power providers, they are poised to reap the benefits of this increased demand, which will positively impact their performance. Overall, Foxconn Industrial Internet and Inspur Information, as two quintessential domestic computing power service providers, have successfully capitalized on the rapid growth window in computing power.

Yang Jianyong, a contributor to Forbes China, expresses views that are solely his own. He specializes in in-depth observations of technology fields such as AI large-scale models, artificial intelligence, industrial robots, IoT, cloud computing, and smart hardware.