A new policy paper by an A-list of academics and former officials close to the Biden camp proposes cooperation rather than confrontation with China in a wide range of fields, including a role for Huawei in the global buildout of fifth-generation mobile broadband. Entitled “Meeting the China Challenge: A New American Strategy for Technology Competition,” the November 16 report may signal a major change in US policy under the incoming administration.
Most of the document is unremarkable, indeed modest in its recommendations. It proposes for example to increase the National Science Foundation’s annual funding for artificial intelligence (AI) to $1.8 billion from $1 billion. By contrast, the city of Shanghai alone plans to spend 270 billion yuan ($40 billion) during the next five years to support AI and the Internet of Things; the National People’s Congress in May proposed a $1.4-trillion technology budget. China fought the Covid-19 pandemic by linking its vast network of sensors to AI servers that helped to anticipate surges in infection and direct forensic testing to potential hotspots.
But the report’s authors, who include Professor Susan Shirk, a former State Department official under the Clinton Administration, former Deputy Secretary of Defense Bob Work and former US trade representative Charlene Barshefsky, depart radically from the Trump Administration by declaring that the US has to learn to live with Huawei and other Chinese telecom equipment providers. They write:
The United States should not attempt to win a race between Huawei and a new American national champion. Instead, the United States should adopt a forward-looking strategy to enable a variety of new entrants to enter the 5G innovation space successfully…
That is a remarkable departure from previous American thinking. Even if the “variety of new entrants” appears, the United States will abandon any hope of world leadership in telecommunications equipment.
Huawei’s prominence in the global technology market presents a special challenge. The risks associated with Huawei can justify a ban on some products by some countries, but total global exclusion of Huawei is not feasible – nor is Huawei the only risk. Instead, the United States should pursue a layered approach to risk mitigation that maximizes network reliability and security and manages the espionage and sabotage risks in the applications and end-user devices that interact deeply with these networks. While a ban on Huawei is feasible in some key countries, especially allies and partners, this is a global networking challenge that requires multifaceted solutions. Considering that Chinese components, user terminals, and software will be intermixed among the billions of connected end users of 5G globally, a total global market ban on Huawei and other Chinese suppliers is not practical.
Shirk and her colleagues do not propose to let Huawei or ZTE build out America’s 5G networks, to be sure, but they eschew Secretary of State Pompeo’s “clean networks” plan. The notion that the United States can insulate its own networks when billions of 5G customers will use Chinese components is absurd, as the authors argue.
A “layered approach to risk mitigation” is exactly what Huawei has advocated for years in response to American charges that its networks would allow Chinese intelligence services to scoop up all the data that flows through them. The Chinese firm avers that it has no access to customer data except when national police agencies require it, and has proposed that governments certify independent research labs to determine whether data is secure. The United States declined the offer.
Rather than depend on the major 5G network providers – Huawei, Ericsson, Nokia, or recent entrant Samsung – the “China Challenge” supports a Silicon Valley pet project,
“O-RAN” (open radio access network), “mixing and matching offerings from various hardware and service providers.” Instead of dedicated chips to process and route signals, O-RAN uses off-the-shelf components. A related approach, or “virtualized radio network” (“Vran”) routes calls through the computing Cloud. Among the major telecom providers, only Nokia has endorsed ORAN and Vran. Ericsson and Huawei argue that the software-based approach is untested, unwieldy and unstable.
Substitution software (routing signals through rented computing capacity in the Cloud) for Huawei’s or Ericsson’s base stations might increase costs, according to experts at both companies. Writing and testing billions of lines of code, rather than embedding the instructions in dedicated chips, could take years and involve long delays and unforeseen expenditures. Since ORAN and Vran are untested, it’s hard to guess in advance whether the approach offers benefits.
America’s tech industry writes code instead of producing physical chips, and it’s understandable why it would prefer subsidies for software rather than hardware. Outside the US, 5G will be built out with dedicated boxes from Huawei or its competitors, and the “China Challenge” authors accept this as a fact of life.
Shirk et. al. propose to keep some of the Trump Administration’s restrictions on technology sales to China, but to allow exports of finished chips for “civilian use”:
The United States should enhance its capabilities by funding R&D in advanced semiconductor capabilities, including manufacturing equipment and by providing incentives for the construction of state-of-the-art semiconductor manufacturing facilities in the United States. In addition, the United States and its allies should impose strict export controls on the sale of semiconductor manufacturing equipment for advanced chips to all Chinese companies, private as well as state-owned, while continuing to allow the sale of finished chips to Chinese companies for civilian uses.
Sooner or later (and probably sooner) China will de-Americanize the production processes for chip fabrication, and the total impact of export controls on equipment will be to impose a modest amount of overhead on the investment side of the Chinese economy. The authors no doubt know this, but think it impolitic to make too many concessions to China. The concession they propose to make, namely the sale of finished chips to China, will make it easier for Chinese equipment producers to build out 5G networks globally.
One might as well state the obvious, however that might grate on American sensibilities. If America is not prepared to create its own national champion in telecommunications equipment, then the next best thing it could do would be to allow Huawei to compete freely in America’s domestic market, under suitable security review, to be sure. If the US doesn’t want to lead, it might as well take advantage of superior (and lower cost) Chinese gear.
Technologically, the security challenges probably are manageable.
Within a year or two quantum-key cryptography will provide end-to-end data security outside the reach of any observer. The real danger to data security comes not from 5G base stations but from chip fabrication itself: University of Michigan researchers demonstrated that it is easy to implant an undetectable “back door” among the billions of transistors contained in a thumbnail-sized chip. There is no way to prevent this (except by fabricating all chips in secure facilities). Even worse, there is no way to prevent an employee at a telecom provider from opening an attachment to a phishing email and letting malware loose in a network.
The most important means of mitigating network security risks is artificial intelligence analysis of signal patterns. Any back door transmitting bootleg data will show up as an anomaly in normal chip activity, and AI surveillance programs do a good show of detecting such anomalies.
What the Trump Administration never understood, and the authors of “The China Challenge” do not quite grasp, is that 5G broadband by itself is a cost, not a profit center. It is like the railroads of the 19th century, which enabled the First Industrial Revolution without making any money for their investors. Huawei and Ericsson accept a return on equity in the single digits because telecom infrastructure is inherently a low-margin business. All the real economic benefit comes from the businesses that 5G makes possible: Smart cities, flexible manufacturing, telemedicine, pollution controls, autonomous vehicles and so forth.
If the United States doesn’t want to take leadership in 5G equipment, it should abandon the field to China, buy cheap Chinese equipment and build out 5G infrastructure quickly. The US should focus its efforts on leadership in 5G applications, where the real money is made. Google and Microsoft are still the best computing companies in the world, and could position themselves to lead in these fields. Whether China can harvest the benefits of its 5G investment in time to justify the estimated $280 billion cost of installing the network during the next five years remains to be seen. That is the question that the United States should be asking of its own tech companies.
As matters stand, the US will have only 10,000 base stations installed by the end of 2020. China has 700,000 and plans for 10 million. If the US pulls back and spends the next five years tinkering with ORAN and Vran, China will establish its dominance in 5G applications. It already has the home field advantage of the world’s largest 5G network, and it also has access to vast data resources, especially in the medical arena, which provide the fuel for artificial intelligence.
I don’t think the US should cede the field to China in 5G. But if it does, it should focus on the new technologies that 5G makes possible. The Sino-American tech war is at a turning point that reminds me of Russia’s position vs. Japan in 1904-1905. After the Battle of Port Arthur, Russia should have drawn the conclusion that it needed a new navy to defeat Japan’s more modern British- and German-built battleships. Instead it sent its old navy, the Baltic Fleet, to engage Japan, and lost most of it at the Battle of Tsushima Strait in 1905. The United States doesn’t need an escalation with China with its existing configuration, but rather a new tech industry.