China’s Computing Power Goals Represent Challenge to U.S. Tech Advantage

By Rick Herrmann   •
Credit: Adobe Stock

China aims to supercharge its core computing capabilities, targeting an increase of over one-third to either match or surpass the United States. Reportedly, a vital component of this endeavor is China’s emphasis on domestically produced and developed technology, underpinning their traditional commitment to self-reliance and resilience. This emphasis may be in part anticipation of further restrictions on U.S. technology exports to China, most recently updated October 17. China argued that restrictions interfere with normal trade and economic activities and “seriously threaten the stability of industrial supply chains.” China’s vision was articulated in an action plan released on October 10, 2023, by the Ministry of Industry and Information Technology in collaboration with six other departments. This announcement should spur U.S. policy makers to aggressively invest in compute and high-performance computing infrastructure to keep the U.S. in a leading position in scientific research and innovation.

The “High-quality Development of Computing Power Infrastructure (HQDCPI)” program, set for completion by 2025, has ambitious goals. The Chinese aim to achieve over 300 EFLOPS (exa-Flop) in compute power. That’s extremely fast; in tech- speak, “Exascale implies 1018 IEEE 754 Double Precision (64-bit) operations (multiplications and/or additions) per second (exaflops).” Significantly, 35% of this figure will be dedicated to intelligent computing or AI capabilities. Additionally, data storage capacities are expected to exceed a staggering 1800 exabytes. For context, this storage capability can encompass the entirety of human knowledge, or alternatively, manage and analyze mammoth data sets.

Today’s updated rules will increase effectiveness of our controls and further shut off pathways to evade our restrictions. These controls maintain our clear focus on military applications and confront the threats to our national security posed by the PRC Government’s military-civil fusion strategy.

U.S. Commerce Secretary Raimondo,
October 17

Setting the benchmark at 300 exa-FLOP serves a multi-fold purpose for China. It offers a tangible goal, establishes a new competitive standard against the U.S., and mirrors China’s overarching strategy to increasingly rely on homegrown technology. This enormous computing power and scale can drive revolutionary advances in various sectors. These include weather forecasting, pharmaceutical research, and aerospace simulations, among others. China’s Ministry has identified over 30 sectors that stand to benefit from this enhanced computing infrastructure, such as finance, energy, education, medicine, and transportation. Emerging applications might encompass digital twin modeling, AI training processes, video surveillance, and extensive scientific research.

China’s current computing capacity stands at about 197 exa-FLOPS. Achieving the goals set in the initiative implies the need for an additional 100 exa-FLOPS by 2025. Drawing a comparison, the world’s fastest computer today, Frontier, located at the U.S. Department of Energy (DOE) Oak Ridge National Labs, boasts a speed of 1.102 exaflops. Built at a cost of $600M and powered by 50,000 processors, replicating this level of power suggests China could be considering a potential $60B investment in its computing infrastructure over the coming 18 months – and, ostensibly, all using domestically produced technology.

This symbolizes China’s vast commitment to investing in, producing, and deploying crucial technologies locally. This initiative not only focuses on raw compute power but also encompasses memory, data transmission, and broader data center infrastructure, acting as the linchpin for China’s AI ambitions and benefiting giants like Alibaba and Tencent.

In contrast, the U.S. lacks a singular, cohesive aspirational target like China’s. Its benchmark revolves around the HPC T500 list. As it stands, Frontier is globally recognized as the fastest computing system. The U.S. claims a 30% stake in the T500 systems, with China closely trailing at 26.8%. However, in terms of sheer performance, the U.S. leads significantly, boasting 45.9% to China’s less than 15%.

China’s 300 exa-FLOP target represents a shrewd benchmark in the ongoing tech race. Yet, the U.S. remains proactive, unveiling fresh initiatives like the CHIPS and Science and Infrastructure Acts and imposing new export sanctions on the most sensitive semiconductor technologies. As both nations jostle for technological supremacy, it’s evident that the challenge posed by China to U.S. leadership is intensifying, especially as the former takes determined strides towards its ambitious goals.

The rate of technological advance in this arena can’t be ignored; the potential to improve the ability to analyze large data sets at much higher speeds has applications across both the U.S. federal market and commercial markets from medicine to weapons systems design, crime prevention to customer service for veterans and taxpayers.