The 5G Future

In November 2016, in a conference room at the Peppermill Resort in Reno, Nevada, a few hundred engineers from the world’s biggest telecommunications companies argued for three days about a problem almost no one outside the room cared about. They were members of 3GPP, the standards body responsible for writing the global cellular network’s specifications, and the agenda item was channel coding: the mathematical recipe by which a phone and a base station agree on how to scrub static out of a radio signal. The choice mattered narrowly because the recipe selected for 5G New Radio would be embedded into every chip in every phone, every car modem, every industrial sensor sold for a generation. It mattered less narrowly because that recipe would be cited in standards-essential patents whose owners could levy royalties on every unit shipped. And it mattered, in terms no one used out loud but everyone understood, because the two leading proposals had national flags attached to them.

The American proposal was a low-density parity-check code, or LDPC, a scheme invented at MIT in 1960 by a doctoral student named Robert Gallager, ignored for thirty years because the computers of the era could not run it, and rediscovered in the 1990s when they could. Qualcomm had built the modern LDPC industry, and its engineers ran the technical case inside 3GPP. The Chinese proposal was a polar code, a younger and stranger object invented in 2008 by Erdal Arıkan, a Turkish professor at Bilkent University in Ankara who had written his MIT doctorate three decades earlier under Gallager. Huawei had championed polar codes since 2010, throwing several hundred researchers and three thousand patent filings at them while the rest of the industry treated them as an academic curiosity. By Reno, the polar codes camp wanted recognition as the equal of LDPC for at least one of the two coding slots in 5G. The LDPC camp was prepared to give them nothing.

What emerged was a compromise nobody publicly admitted was a compromise. LDPC won the data channel, the high-throughput pipe that would carry video and downloads, on the argument that it scaled better at long block lengths. Polar codes won the control channel, the smaller pipe that carried the signaling traffic, on the argument that they performed better at very short blocks and very low error rates. Both sides could go home and tell their boards they had won. Both understood the result as a draw, with the Chinese now standing inside the door of the global wireless standards process for the first time. The Chinese press treated it as a national triumph. State media described polar codes as a Chinese contribution to humanity. In the engineering literature it was understood, more carefully, as the first time a Chinese-led coding scheme had ever made it into a global cellular standard. That was triumph enough.

A year and a half later, in July 2018, Ren Zhengfei flew Arıkan to Shenzhen to give him a medal. The ceremony was filmed. Ren, then in his mid-seventies, gripped Arıkan’s elbow and walked him onstage in front of a hall of Huawei scientists. In his remarks Arıkan said, accurately, that without Huawei’s vision and engineering polar codes would not have made it from the laboratory into a global standard in less than ten years. He did not say, though it was equally true, that without his theorem there would have been no Chinese flag to plant at Reno. The medal was gold. Photographs circulated on Chinese social media as evidence of a country that had, finally, learned how to convert its own science into industrial power.

A coding-theory dispute could carry that symbolic weight because of what 5G actually was. The marketing pitch was speed: peak rates of ten gigabits per second, latencies of a millisecond. The engineering reality was a more interesting bet. Every prior generation of mobile, from the analog AMPS networks of the 1980s through 4G data, had been designed around a primary use case: a phone in a human pocket. 5G was the first to be designed around the assumption that the phones would soon be a minority of users. The other users would be cars, robots, sensors, factory equipment, surgical instruments, drones. The standard’s three official service tiers reflected this: enhanced mobile broadband for humans, ultra-reliable low-latency communications for machines that could not tolerate delay, and massive machine-type communications for the long tail of cheap industrial sensors. Whoever owned the layer of pipes underneath would, in a literal sense, sit closer to the center of the next economy than any router vendor or cloud provider.

This was the geopolitical pitch both Washington and Beijing made to themselves in the second half of the 2010s, and they made it with a particular history in mind. There had been an earlier era, in living memory, when the company that owned the world’s telecom equipment was American. Through the 1990s, AT&T’s manufacturing arm Lucent Technologies and its Canadian neighbor Nortel Networks built the backbone of the global telephone system. Cisco built the routers that connected the new internet to it. Lucent at its 1999 peak was worth more than $250 billion on the Nasdaq. Nortel briefly accounted for a third of the value of the entire Toronto Stock Exchange. Their engineers ran the standards bodies. Their patent portfolios were assumed unassailable. Then the dot-com bubble burst, the carriers stopped buying, and both companies, having grown by acquisition during the boom, discovered they had bought too much, hired too many, and built too little of what came next. From 2000 to 2002, Lucent’s revenue fell seventy percent; Nortel’s fell sixty-five. Lucent merged with Alcatel in 2006 in a transaction politely called a merger and, in industrial terms, an absorption. Nortel filed for bankruptcy in January 2009 and was carved up at auction. Cisco survived because its core business was enterprise data networking, but it never tried to take the equipment slots Lucent and Nortel had vacated. By the early 2010s the world’s tier-one mobile equipment vendors were a European duopoly, Ericsson and Nokia, with the Chinese newcomers Huawei and ZTE growing fast underneath them. There was no longer an American company in the field at all.

That was the structural condition into which 5G arrived. The country that had built the modern telecommunications equipment industry no longer had a domestic champion in it. The country that had spent a quarter-century catching up did, two of them, and one had developed a habit of showing up at standards meetings with so many engineers and so many proposals that it could effectively set the agenda. By the time the 3GPP working groups were finalizing the Release 15 specifications that defined the first wave of 5G in 2017 and 2018, Huawei was the largest single contributor of technical proposals across multiple working groups, from polar codes at the physical layer through proposals for scheduling, beam management, and signaling. All of them, accepted or rejected, generated patent filings. By 2019, third-party studies that counted declared 5G standards-essential patents put Huawei at the top of the table, with shares variously estimated in the mid-teens of percentage points; Nokia, Samsung, LG, Qualcomm, ZTE, and Ericsson clustered behind. The order shifted with methodology. The headline did not. For the first time in the modern history of mobile telecommunications, a Chinese company sat at or near the top of the patent league for a global wireless standard.

Inside Washington, this realization arrived with a delay. Through the Obama administration, 5G had been treated mainly as a spectrum question, a piece of engineering policy entrusted to the FCC. The China dimension surfaced episodically. The 2012 House Intelligence Committee report against using Huawei or ZTE in American networks was treated by most carriers as a technicality rather than a directive. American carriers had mostly not bought Huawei anyway, partly because Huawei had never penetrated the major operators, and partly because Washington had pressured Sprint in 2010 to exclude Huawei from a major procurement decision in a way that signaled future blocks. What changed in 2018 and 2019 was that the question stopped being about which equipment American carriers would buy and started being about which equipment everyone else would buy.

The trigger was a Pentagon-adjacent argument that began circulating in policy circles in 2018. Its sharpest version, articulated by the historian Niall Ferguson in newspaper columns and by the analyst James Mulvenon in Defense Department briefings, ran roughly as follows. Whoever supplied the 5G base stations would have privileged access to the data crossing them and to the firmware running underneath. In a benign world, that access would be limited by encryption, supply-chain audits, and contracts. In a less benign world, where the supplying company was domiciled in a state whose 2017 National Intelligence Law obliged Chinese citizens and organizations to support the work of state intelligence services, the privileged access amounted to a structural intelligence advantage for that state. Even if Huawei had no current intent to assist Beijing, the option existed. The option, in security terms, was the threat. Allied networks built on Huawei equipment would, by 2030, carry the data of allied governments, militaries, and citizens through pipes whose ultimate vendor sat under the legal authority of a strategic competitor. There was no precedent in Western telecommunications policy for accepting that. There was precedent for refusing it: the United States had imposed legal restrictions on foreign ownership of American telecommunications infrastructure for almost a century, on exactly that reasoning.

The first government to act on the argument, before it had even fully crystallized, was Australia’s. In Canberra, Prime Minister Malcolm Turnbull, a former technology investor, took the question personally in 2018. He instructed Mike Burgess, the head of the Australian Signals Directorate, to study whether the risks of a Chinese-built 5G network could be technically mitigated. Burgess’s team spent more than eight months on the question and came back with the answer that they could not. The specific concern, as Turnbull would later describe it, was less about espionage than about availability: a Chinese 5G network, in a serious crisis, could be told to stop working, and Australia would have no way of preventing that. On August 22, 2018, in one of the last acts of his prime ministership, Turnbull’s government issued the guidance that effectively excluded Huawei and ZTE from Australia’s 5G rollout. It was the first ban by any Western government and it stuck. Beijing added it to the list of grievances it would later present to Canberra under the heading of fourteen complaints. Japan, without making a formal public ban, began quietly excluding Chinese vendors from procurement decisions later that year.

The British case unfolded more slowly and more painfully. The United Kingdom had operated since 2010 a peculiar piece of bureaucratic machinery called the Huawei Cyber Security Evaluation Centre, in Banbury. Inside it, British engineers vetted by GCHQ inspected Huawei’s source code line by line, looking for backdoors. The arrangement was unique in the world. It produced annual reports that grew steadily more critical of Huawei’s software quality, but year after year concluded that the residual risk could be managed. On that basis, in January 2020, Boris Johnson’s government announced that Huawei would be allowed to supply up to thirty-five percent of Britain’s 5G access network, excluded from the sensitive core but permitted in the periphery. The decision angered Washington and split Johnson’s own Conservative caucus. Then, in May 2020, the United States sharpened its export controls to cut Huawei off from chips made with American manufacturing equipment, which meant within a year Huawei would no longer be able to obtain the leading-edge silicon its 5G base stations relied on. Britain’s National Cyber Security Centre reopened its assessment. It concluded that Huawei’s forced shift to whatever chips it could still source would itself become a security problem, because the new components would be untested and outside the audit framework Banbury had built. On July 14, 2020, the National Security Council met under Johnson’s chairmanship and reversed the January decision. Huawei would be banned from new 5G equipment purchases after the end of 2020 and removed from the British network entirely by 2027. The Conservative MPs who had been agitating for a tougher line celebrated. The British carriers who had built their rollouts around Huawei estimated the cost of the swap-out at roughly two billion pounds and a delay of two to three years.

Three weeks after the British reversal, on August 5, 2020, Mike Pompeo stood at a State Department lectern in Washington and announced what he called the expansion of the Clean Network program. The program had originally consisted of a single pillar, ensuring that Chinese carriers were not connected to American networks. Pompeo now stretched it to cover five further pillars: clean carriers, clean stores, clean apps, clean cloud, and clean cables. Translated out of slogan form, the Clean Network was an attempt to construct, in one diplomatic document, a list of categories of digital infrastructure in which the United States was prepared to refuse Chinese vendors and to ask its allies to refuse them too. Under Secretary Keith Krach took the program around the world. By the end of 2020 the State Department was claiming more than fifty endorsing countries, including twenty-seven of thirty NATO members and twenty-six of twenty-seven EU members, although the endorsements varied wildly in their bindingness. The Clean Network was the moment at which American foreign policy began to treat the architecture of global wireless networks as a national-security concern of the same order as nuclear nonproliferation or basing rights.

The country that gave the campaign its hardest test was Germany. Angela Merkel’s chancellery was visibly reluctant to do what the British had done. Deutsche Telekom had been a major Huawei buyer for years, and the German automotive industry, which depended on access to the Chinese market, lobbied hard against any decision that would invite Beijing’s retaliation. Merkel’s preferred policy was security-by-process: not banning vendors by name but imposing certification requirements that any vendor would have to meet to sell into critical German infrastructure. The Bundestag, including parts of her own Christian Democratic Union, pushed back. The wrangling went on through 2020 and 2021, ending in legislation that did not name Huawei but did impose a political-trustworthiness test Huawei was widely understood to be unable to pass on its merits. German operators continued to carry Huawei in existing 4G networks, gradually wound down new procurements, and conducted the transition with the embarrassed deliberateness of a country that had decided, without quite admitting it, that an American argument it disliked was an argument it could no longer ignore. France took a parallel route, formally permitting Huawei in non-core elements but issuing licenses so short that Orange and Bouygues planned around the assumption that Huawei would be gone within a decade. Italy, the Netherlands, and Spain settled into roughly the same position.

The Western strategy, by 2021, was less coherent than the Clean Network’s pillars suggested. The United States had no domestic 5G equipment vendor. It could not credibly tell allies to buy American. It could only tell them to buy European, which favored Ericsson and Nokia, neither of which had the scale or pricing power to fully replace Huawei in markets where Huawei was already deployed. To address that gap, American policymakers seized on a movement that had begun for unrelated technical reasons inside the carrier industry: Open RAN. The idea was to disaggregate the radio-access network into modular components defined by open interfaces, so that a carrier could buy radio units from one vendor, baseband processing from another, and software from a third. The O-RAN Alliance, founded in February 2018 by AT&T, China Mobile, Deutsche Telekom, NTT Docomo, and Orange, became the standards venue. By 2020 both the Trump and Biden administrations were promoting Open RAN as the architectural counter to Huawei: instead of replacing one vertically integrated Chinese vendor with one vertically integrated European vendor, allies would buy from a constellation of smaller suppliers, many of them American startups, glued together by open standards. The most concrete demonstration was Rakuten Mobile, the Japanese e-commerce conglomerate’s mobile network arm, which built a fully virtualized 5G network in Japan from scratch in 2019 and 2020 on commodity servers. The FCC and DoD began funding domestic Open RAN initiatives in 2021 and briefing allied governments on its merits. Whether Open RAN would mature into a real second source or remain mostly a slide deck was, by the early 2020s, still an open question. The strategic intent was an attempt to fragment the dependency that had made Huawei the default choice for so much of the world.

Viewed against the longer history of telecommunications standards, the absence of any prior precedent was striking. The transitions from 2G to 3G to 4G had been industrial and commercial fights, occasionally with industrial-policy overtones, never with national-security overtones at the level Pompeo’s State Department had now made routine. American officials had not flown to Berlin or Tokyo or New Delhi in the 2000s to ask other governments to choose between Ericsson and Nokia on the grounds that one of them was a strategic risk. There had been no clean network for 4G. The 5G fight established, in the working assumptions of Western foreign ministries, that the question of who built a country’s wireless network was, like the question of who supplied its fighter jets, a question of allegiance.

The standards meetings did not stop. The same engineers who had argued over channel coding in Reno were soon arguing in Stockholm and Hangzhou over the contours of 6G. The American campaign had carved Huawei out of the buyer side in much of the West. It had not carved it out of the standards side. Huawei engineers continued to attend 3GPP meetings, file patent declarations, and propose new ideas. So did Ericsson and Nokia and Samsung. So did a smaller cohort of American firms, since the 5G fight had reminded Washington that it had no domestic champion to send. Polar codes were now embedded in the firmware of every 5G phone in the world, including phones manufactured under American sanctions in Chinese factories that no longer accepted the chips America had once exported freely. Arıkan, semi-retired in Ankara, occasionally gave lectures on what came next.

A wireless network was no longer a piece of telecommunications equipment. It was a layer of national infrastructure dense enough with sensors, intelligence, and decision-making to count as part of the state apparatus. The United States, having watched its own equipment industry liquidate itself in the early 2000s, had decided to wage a global campaign over who supplied that layer to its allies, despite having no domestic supplier of its own. It was the first wireless standards battle the country had ever tried to win on national-security grounds, and it was being waged on a battlefield where the country could no longer field a team.