China's new LineShine supercomputer has become the fastest in the world, overtaking the U.S.-based system El Capitan [1].

This shift in computing dominance signals a significant change in the global race for artificial intelligence and high-performance processing. The ability to process massive datasets more quickly than any other nation provides a strategic advantage in scientific research and military applications.

Located in Shenzhen, the LineShine system reached a performance level of 2.198 exaflops [1]. This represents 2.198 × 10¹⁸ calculations per second [1]. The achievement officially moves the machine to the rank of No. 1 globally [1].

This displacement pushes the El Capitan system, operated by the Lawrence Livermore National Laboratory in the U.S., to the second position [1]. The transition marks a pivotal moment in the competition between the two nations to lead in supercomputing hardware and software architecture.

Officials said the machine is intended to accelerate the development of artificial-intelligence models [2]. The system is also designed to handle other high-performance computing workloads that require extreme processing power [2]. These workloads often include complex climate modeling, genomic research, and cryptographic analysis.

While the U.S. has previously held the top spot, the emergence of LineShine demonstrates China's growing capacity to develop indigenous high-end computing infrastructure. The system's deployment in Shenzhen underscores the city's role as a central hub for Chinese technological innovation.

“LineShine reached a performance level of 2.198 exaflops.”

The ascent of LineShine to the top of the global rankings reflects a narrowing gap, or a complete reversal, in raw computing power between China and the U.S. Because supercomputing is the primary engine for training the next generation of large-scale AI models, this hardware advantage could allow China to iterate on AI capabilities faster than its competitors, potentially bypassing current U.S. leads in software and algorithmic development.