Computing China’s future
- China is investing heavily in supercomputers for economic and military reasons
- It now competes on equal terms in the next wave of exascale and quantum technology
- Breakthroughs in these fields could give China industrial and strategic superiority
China has long had to play catch-up in most of the world’s critical industries, but one where it is no longer chasing is high-performance computing. Things move fast in the world of supercomputing and things move fast in China, meaning that the country has advanced at a lightning pace.
In 2007, China possessed only 13 of the world’s fastest 500 supercomputers. By June 2017, the TOP500 rankings list announced that this number had soared to 160, putting the Chinese just behind the United States with 168. Most impressive, however, is the fact the world’s fastest supercomputer, the Sunway TaihuLight, at the National Supercomputing Center in Wuxi, near Shanghai, runs on Chinese-made processors. Just 10 years earlier, China’s fastest computer was powered by imported technology and ranked 43rd globally.
In a fast-moving industry, however, there are already signs that supercomputers based on binary digital technology will be superseded by quantum computers. This new breed of high-performance machines has the potential to fundamentally to change humanity’s computing capabilities. Their speed and capability of handling massive amounts of data that overwhelm digital computers opens up a wide range of hitherto unthinkable possibilities and threats. The big difference for China is that it will now be at the forefront of this next technological leap, rather than being left in the dust by the more advanced countries, as it had been in the past.
The origins of China’s supercomputing reach back to the late 1980s, when the World Bank funded a project with China’s State Planning Commission to set up networking facilities across China. These centers are now based in Tianjin, Beijing, Shanghai, Shenzhen and Guangzhou.
For the authorities in Beijing, it had become clear by the early 2000s that supercomputers were indispensable if the country wanted to keep up with scientific and technological advances of the 21st century. Massive investments were made in infrastructure and human capital, as some of the world's brightest minds were recruited to help develop cutting-edge technology in China.
In 2015 Beijing launched a major new technological strategy, Made in China 2025, which aims to reduce the country’s reliance on tech imports for its future economic development. Over the 10-year period, the government set a goal of increasing the proportion of domestically produced components in Chinese output to 70 percent, from 30 percent in 2015. A crucial element of this project focuses on processing power, semiconductors and next-generation computing. Companies like Loongson Technology Corporation, a Chinese computer chipmaker, are Beijing’s biggest hope in this area, although they have a way to go before they challenge Western giants.
The Chinese desire to reduce reliance on imported chips to run its supercomputers is based on two important policy considerations: cost and national security. China spends more on importing semiconductors and other related computer components than it does on oil. In terms of national security, Beijing has long worried that the U.S. could use protectionist policies to thwart development of critical Chinese infrastructure, and this fear has only increased under Donald Trump’s presidency.
Beijing has reason to be anxious, since the U.S. placed a ban on exports of supercomputer chips to China in 2015. The Chinese responded by unveiling the TaihuLight computer, the world’s fastest, using domestically produced SW26010 processors. Yet there are lingering allegations, as with most Chinese industry, that industrial espionage and patent theft played a role in this technological success.
High performance computing plays the role of an enabling technology for a vast array of other advances
Protectionist actions are partly based on the U.S. government’s own fears that it is about to lose the top spot in high-performance computing. On some metrics, such as computer speed, China is already ahead, although the Americans maintain a very slim edge overall. A 2016 report by the U.S. National Security Agency and the Department of Energy warned that the loss of pole position in the computing supremacy race was imminent. The report also highlighted two key reasons why the U.S. and its competitors were striving to gain the upper hand. First, high performance computing plays the role of an “enabling technology” for a vast array of other advances and applications. And second, because supercomputing provides the means to wage cyberwarfare, both on attack and defense.
A viable future?
The next big focus at the top end of computing is exascale, otherwise known as “super supercomputers.” Exascale supercomputers, when developed, are binary machines capable of making at least a billion billion calculations per second, far greater than any computer currently in operation.
In January 2017, Beijing announced that China’s first exascale prototype would be complete by the end of this year, and that it would be 200 times faster than the country’s first world-leading supercomputer, Tianhe-1, which briefly took the top spot in 2010. A fully functioning exascale machine should be ready for testing in 2020 and go into service a year later. This schedule puts China narrowly ahead of its closest competitors – the U.S., France and Japan – which expect to get their own exascales running by 2022. Of course, there is no guarantee China will meet its deadline or that the quality of its technology will stack up to its rivals. But what is clear is that the Chinese are now forcing the pace, rather than struggling to catch up.
There is, however, a pretty important catch. The faster supercomputers get, the more power they require. Chips will continue to get more efficient, but there is a limit on how little energy they can use. Even at the lowest end of the consumption scale, the next generation of supercomputers will be power hogs. The unusually efficient Sunway TaihuLight already uses 15.3 megawatts of power, enough to power a small city. Exascale computers will only exacerbate this problem. There are forecasts that by 2040, the world will not produce enough energy to run exascale computers on a mass scale.
The solution to this difficulty, and to many of the world’s other intractable technological problems is quantum computing – a Holy Grail for physicists and governments alike.
Quantum computing has long been theorized and pursued by physicists with limited practical success. It would be such a major leap forward because it would allow an exponential and practically unlimited increase in processing speed. Classical binary computing allows data, or bits, to exist in two states, 0 or 1. In theory, quantum computing will allow data to be stored in quantum bits (qubits), which will allow data to exist as 0, 1, or in both states simultaneously. As the number of qubits increases, their permutations increase exponentially, allowing quantum computer speeds to increase rapidly while using less energy than binary supercomputers.
The almost incomprehensible scale of the potentially trillions of calculations a quantum computer could make every second opens worlds of possibilities. From better analyzing weather patterns to breakthrough medical advances to accelerated machine learning, humanity can wrest huge advantages from this technology. Perhaps the most significant way a high-performance quantum computer could change the world would be to facilitate an automated artificial intelligence revolution.
Quantum computing's less benevolent uses include breaking down the strongest encryption systems and developing artificial intelligence weapons
The technology can also be used in less benevolent ways. They range from breaking down even the strongest encryption systems with ease to developing an array of artificial intelligence weapons. Understandably, countries are keen to ensure they stay ahead of the curve when it comes to what could potentially become one of the biggest technological breakthroughs of the 21st century.
The advent of quantum technology is drawing closer. D-Wave Systems Inc., a Canadian company, began the commercial sale of what it describes as quantum computers in 2017. While it is widely accepted that D-Wave’s offer does use quantum technology, the company has not yet managed to take it beyond solving simple optimization problems – for example, picking the best solution from a set of fixed parameters. It remains unclear whether and when researchers will be able to develop machines capable of tackling the more difficult real-world problems theorists dream about.
Not wanting to be outdone, in May 2017 the Chinese Academy of Sciences unveiled what it called a “baby” quantum computer. While the machine was undoubtedly a major achievement, it can still only complete one type of highly complex calculation, again limiting its real-world application.
Despite the enormous hurdles that remain before quantum computing can be unleashed, China has started the race on equal footing. Given Beijing’s proven ability to make up ground fast in conventional supercomputing, it would be unwise to bet against it playing a leading role in the quantum revolution.
The national security ramifications are potentially significant. Should China achieve true quantum computing before its rivals, it may be able to bridge the military technology gap with the U.S. far faster than has been anticipated. Chinese industry would also be at the forefront of technologies that will likely define economic and political power in the 21st century. If that happens – and it is still a big “if” – it would mark a fundamental shift in the geopolitical balance, with China assuming the role played by the U.S. in the last century, when American industries and technologies reigned supreme.