Is the US’s technology war successfully curbing China’s tech rise?
List of Critical and Emerging Technologies; Updated: 2nd September 2023
In the List of Critical and Emerging Technologies released by ASPI, a well-known Australian think tank, China takes the lead in 53 categories of cutting-edge technologies. At the same time, the USA claims the remaining 11. Notably, according to the list, China dominates in areas like advanced materials and manufacturing, energy and environment, and advanced information and communication technologies and enjoys a sizable advantage in defense, space, robotics, transportation, and quantum technology.
On the other hand, artificial intelligence technologies and biotechnology, gene technology, and vaccines emerge as the battlegrounds where these two powers fiercely compete for supremacy. With all those exciting advances, however, there are still areas where it falls behind the United States. For example, the US holds the edge when it comes to natural language processing and advanced integrated circuit design and fabrication, which are vital for technologies like ChatGPT and GPUs used in training. Similarly, in the field of life sciences, particularly genetic engineering, the US remains the leader.
Let’s start with the high-performance computing.
The current league table of supercomputers worldwide is no doubt dominated by the US, and no Chinese system feature in the top 10. Compared to a few years ago, when the Chinese systems such as Sunway and Tianhe had once claimed the top spot, it seems to be a dramatic decline. However, many parts of this can be attributed to the fact that Chinese supercomputers have ceased to submit key data regarding their system to TOP500.org, an American organization, ever since 2020, apparently due to the matter’s sensitivity. In other words, the data TOP500 has regarding Chinese supercomputers are obsolete.
So, what is the current situation of Chinese supercomputers?
On December 6, The Chinese National Supercomputing Center in Guangzhou made an official announcement about the launch of the next-generation, domestically-developed supercomputer system, “Tianhe Xingyi.” It is equipped with the state-of-the-art MT3000 processor, designed and manufactured completely in China. The system boasts an impressive 15,974,400 CPU cores, achieving a remarkable peak performance of 620 petaflops, putting it ahead of the Japanese Fugaku system, which ranked 4th in the TOP500 list. It’s not bad for a totally homegrown system.
However, Tianhe is not the most powerful supercomputer in China. Remember the Sunway I mentioned earlier? Its latest version is Sunway Ocean Light, incorporated with domestically developed “SW26010-Pro” many-core processors.
The supercomputer stands out with an impressive configuration of over 100,000 nodes, housing over 41 million cores distributed across approximately 105 cabinets. It achieves a remarkable peak performance of 1,500 petaflops, placing it the second most powerful supercomputer globally, trailing only the U.S.-based Frontier supercomputer. According to South China Morning Post reports, it also outperformed leading supercomputers, including the Frontier, in computing efficiency. In an interview with the post, the Chinese scientist said, “The processor is not new. It has been used in China’s supercomputing systems for the past two or three years, but the public, especially the Western world, only just came to know it.”
What is most remarkable about Sunway and Tianhe is that their predecessors relied on Intel’s processors. However, after years of dedicated research and development, they are now fully homegrown, immune to US sanctions.
Next, let’s move on to the next area where the US holds sway as the primary force: AI.
ChatGPT from OpenAI has started a revolution and pretty much changed the way we work and study. By comparison, Chinese AI models haven’t gotten such a spotlight and are missing from the headlines worldwide. This may leave us wondering about the current status of Chinese AI. To find it out, we can again look at international rankings. One of these is released by Hugging Face, the biggest community for open-source AI models. As per the Open LLM Leaderboard, the Qwen-72B model from China’s Alibaba tops the list, beating Meta’s more popular Llama model. This is the ranking for open-source models, but what about the close-source ones such as GPT-4?
Luckily, we found another leaderboard compiled by Stanford University, the AlpacaEval, which put together all the closed-sourced or open-sourced models. The list shows the Chinese-made Yi 34B Chat ranks just behind GPT-4 and the GPT-4 turbo, in third place, comfortably above ChatGPT. Given that ChatGPT was unveiled exactly one year ago, we can safely conclude that Chinese models are less than one year behind their US counterparts.
AI models are trained on GPUs, which makes it another battleground of the ongoing China-US tech war. By restricting the sale of the most advanced INVIDIA GPUs to China, the US government aims to slow down China’s AI development. But will it succeed? How far behind are the GPUs made by Chinese manufacturers? Unfortunately, we don’t have another ranking here to show you, and the information regarding this is sometimes contradictory. Nonetheless, we can hear what the people in China’s AI community say about this.
In August, Liu Qingfeng, the chairman of Chinese AI giant iFlytek, commended Huawei for manufacturing a GPU that he claimed was “essentially equivalent to Nvidia’s A100.” He further stated that iFlytek was collaborating with Huawei to develop hardware.
It’s believed that the hardware in question was powered by Ascend 910B, a previously undisclosed model. During iFlyTek’s recent earnings call, Senior Vice President Jiang Tao reiterated that the capabilities of Ascend 910B were “on par with Nvidia’s A100.”
Like iFlytek, Baidu also ordered 1,600 Huawei 910B chips to deploy 200 servers during August. Analysts and sources note that while the 910B chips exhibit raw computing power comparable to Nvidia’s offerings, they still lag in performance. Nevertheless, they are considered the most advanced domestic option available within China.
Next, as we go through the ASPI’s List.
We can see China is trailing the US in genetic engineering, vaccines, and nuclear medicine.
For genetic engineering, one of the most important technologies potentially transforming life science is CRISPR, which can be used to selectively modify the DNA of living organisms and edit DNA like a precise pair of molecular scissors. This has huge potential for treating diseases and agriculture. As of February 2021, the number of global CRISPR technology-related patent families has exceeded 4,000. China holds the highest number of CRISPR patents, with over 2,000 related patents, followed by the United States. These two countries are the main contenders in the competitive market for CRISPR technology.
Regarding patent flow, the United States stands out as the largest contributor to CRISPR-related patents, with a substantial number of patent filings in various countries, highlighting a notable trend of technology export. Approximately 12% of CRISPR technology applications from the United States are directed to the World Intellectual Property Organization or the European Patent Office, enabling protection across multiple nations.
On the other hand, China takes the lead as the primary recipient of patent filings, positioning itself as a significant global technology importer. Other countries exhibit a considerable presence in terms of patent filings within China. It is worth noting that approximately 90% of the patents filed in China originate domestically.
In a word, most of these Chinese patents on CRISPR have never crossed the border, while the American ones travel widely. That may explain why the US CRISPR technologies have a larger influence globally. Patent numbers alone aren’t the sole indicator of technological leadership.
In terms of vaccines, during the COVID pandemic, US companies have pioneered mRNA technology. This newer technology instructs cells to produce certain viral proteins to trigger an immune response. China initially opted for the trusted and proven inactivated vaccine, which uses weakened or killed virus particles. This choice likely stemmed from factors like existing manufacturing capacity and familiarity with the technology. mRNA vaccines typically elicit a stronger immune response but may have more side effects. In contrast, inactivated vaccines generally carry a lower risk of side effects but might offer slightly lower protection. And not until mid-2023 did Chinese firms launch their first mRNA vaccine for COVID.
Moreover, if we can look at the overall academic production, that is, the research papers published in top journals in a certain field, life science represents one area in which China has to make a serious effort to catch up. The Nature Index is a database from the prestigious scientific journal Nature that tracks the affiliations of authors and institutions and their contributions to research articles. It also ranks countries and institutions by the number of scientific papers published in leading journals. In both biological science and health science, China comes at the 2nd place right behind the US, but a distant second. Therefore, a lot of work for China to do here.
Quantum computing is another area where the list puts the US ahead of China.
The strategic value of quantum computing is comparable to that of AI, with potentially far-ranging applications from scientific research to the military. Here again, Chinese scientists have made exciting breakthroughs during 2023. In October, the Institute of Quantum Information and Quantum Technology Innovation at the Chinese Academy of Sciences introduced a cutting-edge quantum computing prototype, “Jiuzhang III”, with 255 photons. The processing speed of “Jiuzhang III” for Gaussian boson sampling has skyrocketed by an astonishing one million times compared to its previous iteration, “Jiuzhang II.” To put this tremendous leap into perspective, the highest complexity sample handled by “Jiuzhang III” within a mere one-millionth of a second would demand over 20 billion years of computation time from the current fastest supercomputer, “Frontier,” that is, faster by a staggering factor of 1016 (ten quadrillion times). This achievement has not only set a new world record in the field of photonic quantum information technology but also demonstrated the remarkable capabilities of quantum computing.
Small satellites and space launch systems are emerging technologies that have the potential to revolutionize the access and utilization of space.
Small satellites, also known as CubeSats or nanosatellites, have masses ranging from 1 to 500 kilograms. They can perform various functions such as remote sensing, communications, scientific experiments, and technology demonstrations. Space launch systems are rockets that deliver payloads, such as small satellites, into orbit. They can be classified into ground-launched, air-launched, or sea-launched systems, depending on the launch platform.
The US and China are two of the leading countries in the development and deployment of small satellites and space launch systems. According to a Center for Strategic and International Studies report, the US has been at the forefront of these innovations, accumulating 1,763 payloads launched between 1992 and 2019. Russia’s 840 payloads and China’s 480 payloads round out the top three. The US has also pioneered air-launched systems, such as Virgin Orbit’s LauncherOne, which can provide flexible and economical options for launching small satellites. However, China is also working on a design for a reusable air-launched orbital rocket, which could offer a competitive advantage in the future. China has also demonstrated its capability to launch satellites from the sea using its Long March 11 rocket, which can reduce the constraints of land-based launch sites. Both countries are also investing in new-generation launch vehicles, such as the US Space Launch System (SLS) and the Long March 9 in China, which could enable more ambitious missions to the moon and beyond.
In launching small satellites from space, China has a unique advantage in its Tiangong Space Station. A more flexible approach to launching micro and nano satellites has been achieved through the successful implementation of the space station project and the Tianzhou series cargo spacecraft. This technological innovation breakthrough paves the way for new possibilities in global space exploration. Through extensive testing and validation, China has successfully deployed CubeSats from the space station, achieving a remarkable on-orbit storage duration of over six months. The development and application of this novel method for launching micro and nano satellites bear significant practical value.
Firstly, it substantially enhances satellite launch efficiency. By leveraging the space station as a launch platform, scientists can release satellites as required, eliminating the need to wait for suitable rocket launch opportunities. This shortens the satellite development cycle and enables the potential for prompt satellite deployment in emergencies.
Secondly, this innovative launch approach significantly reduces the cost of satellite launches. Unlike traditional methods that rely on large rocket payloads or the carpooling of multiple satellites on small rockets, launching from the Tiangong Space Station offers a more cost-effective solution. This is achieved by sharing launch costs with other missions, such as regular resupply missions to the station, allowing satellites to ‘hitch a ride’ and thereby reducing the cost for each payload. Additionally, the flexibility in launch timing from the space station avoids the costs associated with delays and holding costs on the ground. Moreover, as space operations become more regular, economies of scale can be realized, lowering the costs per satellite. This flexible launch mechanism, therefore, not only enhances efficiency but also opens up space exploration to a broader range of research institutions and individuals by making it more financially accessible.
This comparative analysis of Chinese and American technological advancements reveals a landscape of intense competition and rapid innovation. While China demonstrates formidable strength in several areas, it still faces challenges in others, particularly in achieving global influence in certain technologies. As both nations continue to push the boundaries of scientific and technological capabilities, the world watches with keen interest to see how this rivalry will shape the future of global technology.