The present chapter begins by discussing how the internet of the future promises to encompass large chunks of the physical world and onboard most if not all economic and social activities. The concern that Huawei and other Chinese companies would have access to the control panel of the global technological system set in motion the ongoing technology wars between China and the United States and explains the new appeal of ‘decoupling’. What is decoupling? The strategy has been applied to the most sensitive advanced technologies such as chips. The goal of geopolitical competition is to preserve exclusive control over the sources of technological power and to ensure that access to those sources is denied to a rival. The theory of world building captures this reality by stressing the existence of sovereign actors no longer constrained by a system of shared rules, architects of the world order that are nevertheless profoundly global in their outlook. Chips play a major role in the new geopolitics. If straits and islands are the gates to the oceans, microchips are the gates to the virtual. They can be compared to the basic layer of a computer operating system.

The Hack

In October 2018, Bloomberg Businessweek made the startling revelation that a tiny microchip, not much bigger than a grain of rice, had been found nested inside motherboards assembled for Elemental Technologies by Supermicro, a computer hardware company in San Jose, California. The microchip was not part of the original design for the Supermicro motherboards, and among those customers who had acquired them from Elemental were some particularly sensitive entities: the Central Intelligence Agency, the Department of Defense and even the onboard networks of navy warships.

Multiple people familiar with the matter told Bloomberg that investigators found that the chips had been inserted at factories run by manufacturing subcontractors in China. The chips allowed attackers to create a stealth doorway into any network that included the implant. One official said the exploit eventually affected almost thirty companies, including a major bank, government contractors, Amazon and Apple. Supermicro, Apple and Amazon publicly called for a retraction of the story, but Bloomberg stood by it, returning to it in 2021 with an account of new developments.¹

The code contained in the Supermicro implants was seemingly capable of granting access to anonymous computers anywhere on the internet in order to receive new code, allowing any attackers to change how the device worked. The goal was to gain full control of the target system. Hackers call this ‘god mode’: access to everything and root privileges to do everything. The chip could neutralise passwords, steal encryption keys for secure communications, block security updates and open up new pathways to the internet. In 2010, a Pentagon security team noticed that thousands of servers in unclassified networks at the Department of Defense were loaded with unauthorised instructions directing each one to secretly copy data about itself and its network and send that information to China. Analysts at the Pentagon were also worried that the implant could be a digital weapon for shutting down sensitive systems during a conflict. Keeping the discovery secret, American spies proceeded to use the implants to gather intelligence on China. The microchip became something of a double agent, a figure from Cold War spy thrillers.

Detection of a hardware hack is difficult or impossible because the source of the attack is a piece of hardware, invisible to software tools. Often the only way to detect the implant or exploit it is to use sensing techniques capable of measuring tiny variations in the device’s operation that are well below the level of noise in a standard setup. One prominent hardware hacker in Singapore with whom I discussed the Bloomberg story in 2019 compared these methods to identifying a woman among a large crowd by detecting the tiny traces of her perfume.

These extraordinarily surreptitious forms of hardware hacking were made possible by the near monopoly of the technology supply chain enjoyed by Chinese companies. When investigators tried to find the source of the attack, they had to retrace the full extent of the supply chain, traversing a labyrinthine path of suppliers, subcontractors and factories in both Taiwan and China. Supermicro, founded in 1993 by Taiwanese immigrant Charles Liang, was built to take advantage of these intricate supply chains. Many of its motherboards were manufactured in China by contractors, then assembled into servers in San Jose and elsewhere.

Back in 2018, the Bloomberg story seemed to me to announce the inception of a new era in world politics, an era in which power threatens to become absolute because its reach extends to the way the world works, or at least the virtual world of communications, information processing and computing power targeted by the Supermicro attackers. It was in some fundamental respects a stimulus for the ideas in this book.

The Huawei Files

In 2019 Huawei founder Ren Zhengfei predicted the next battle for supremacy with the United States would be over the internet of things and ‘smart factories’. Huawei had been developing chips and software for companies to connect their factory floors to the internet and hoped its deep range of products would give it a privileged position to set the standards that would eventually be adopted across the world. 5G was a stepping stone to this because high-speed connectivity is necessary for transferring bulky data from industrial devices for data analysis, and Huawei held the largest number of patents essential for global 5G standards.²

Shorthand for ‘fifth-generation’ mobile networks, 5G is the baseline standard that promises to revolutionise mobile networking technologies by enabling the connection of billions of devices and data transfer at much faster and more reliable rates. It is quickly becoming the general infrastructure upon which we conduct our daily lives, not just as technology users but as citizens and participants in society.³ Through faster connections, it is now possible to reduce latency, improve mobility and immerse users in realistic virtual environments. More and more activities are mediated by intelligent communication networks, particularly as data must be collected by all the intelligent devices organising the world around us.

As the internet grows, it already encompasses large chunks of the physical world. But even the vision promised by 5G technology may soon look outdated. By 2030 the world will be discussing the impact of 6G, and later perhaps 7G. The millisecond levels of latency that 5G offers will be replaced with 6G and microsecond latency. At the end of this process, one can easily imagine that every element of the physical world will have a digital representation, and thus that the world as a whole will have migrated to virtual reality.

A digital model is a comprehensive software representation of an individual physical object and the properties, conditions and behaviours of the real object through models and data. It allows a system to play with the requirements and calculations of each connected element in software form, and in this sense there is an obvious correlation between the concepts of a digital model and an avatar.⁴ An avatar is a virtual representation of an object – usually a person – that can behave as a substitute of the physical individual, but perhaps objects can have avatars as well. For example, a wind turbine producing renewable energy might have a digital address and be connected to a global network, the ‘internet of energy’. A power plant might no longer be a physical building but a virtual network of energy-generating units where fluctuations in the generation of renewable energy can be balanced by ramping up and down the power generation and power consumption of controllable units such as wind farms, solar parks, biomass or hydro. ‘A virtual power plant is a system of distributed energy resources’ whose virtual nature ‘comes from its lack of a central physical facility, like a traditional coal or gas plant. By generating electricity and balancing the energy load, the aggregated batteries and solar panels provide many of the functions of conventional power plants.’⁵

5G promises to make it possible to quickly and reliably collect data not only from energy systems but also from cities and roads, transportation networks and vehicles, including jet engines, so that, for example, when planes arrive at a gate, they can be tagged for predictive maintenance and management.⁶ As autonomous vehicles become a reality, 5G technology is expected to allow vehicles to relay signals directly to each other and communicate with sensors on bridges, roads and traffic lights. Other obvious use cases include telemedicine, vastly expanded by faster data exchanges and better video streaming, smart ports supporting automated cargo handling and optimising the use of shipping lanes and terminal facilities, as well as wearable communication technologies for both personal and business purposes.

As a consequence, everyday things in our surroundings will become connected and programmable. Physical objects would be no more than ‘peripherals’, hardware elements that collect sensor data and submit the collected data to the broader communications ecosystem. Actuation or control from the cloud to the peripherals is equally important. These capabilities promise to make it possible to command and control everyday objects all over the planet from the comfort of a virtual programming environment.⁷ ‘Once we get there,’ Bill Wasick suggested ten years ago, ‘that system will transform the world of everyday objects into a designable environment, a playground for coders and engineers. It will change the whole way we think about the division between the virtual and the physical.’⁸

Conceptually, it makes sense to speak of 6G as the advent of a programmable world, and 7G might introduce the ability to rigorously simulate and predict future events. ‘To realise future networks and operate a large number of versatile services without accelerating cost and complexity, the level of network intelligence must be raised.’ Such an approach implies that the system becomes more and more autonomous, adjusting to its environment, constantly observing and learning from previous actions.⁹ All these future capacities will be based on the kind of ‘digital representation’ that is already available today. Nokia and Bosch, for example, see the transition from 5G to 6G as eliminating the need for objects to be connected in order to be tracked, ‘which will allow 6G signals to function similarly to radar, giving users an awareness of their surroundings beyond their traditional senses’. Peter Vetter, President of Bell Labs Core Research at Nokia, suggests the following way to conceptualise the future 6G: ‘In the next decade, 6G will be capable of sensing all objects in their coverage areas regardless of whether they contain active radios. We are creating networks that will endow humans with a digital sixth sense.’¹⁰ InterDigital and Omdia believe that 6G, while still a work in progress with an expected 2030 launch date, will be capable of providing superior user experiences and will support new sensory interfaces. Sensory 6G describes the human ability to experience digital services with an expanded range of senses using communication networks combined with optimised devices, applications and content.¹¹

As the number of connected devices and the amount of data explode, a greater share of total global economic output will come to rely on global data networks. More and more of the physical world will be controlled by a communication network superimposed on every economic and social process. ‘The vulnerability of companies, industries, cities, and even entire countries to disruptive cyberattacks or network outages will grow accordingly.’¹² With the advent of 5G, practically everything with a chip will be tethered to a wireless network. The 5G infrastructure will intertwine factories, power plants, airports, hospitals and government agencies. If it comes under a broad, sustained attack, one commentator alarmingly argues, it ‘would mean a total collapse of society’.¹³ The attack surface for exploitation attacks is growing as well.

Unsurprisingly, the sudden realisation that a Chinese company with ties to the Chinese Communist Party was leading the next telecommunications revolution sent shock waves through the national security apparatus in Washington. A memo produced by a senior National Security Council official in 2018 explained that ‘whoever leads in technology and market share for 5G deployment will have a tremendous advantage towards ushering in the Massive Internet of Things, machine learning, artificial intelligence, and thus the commanding heights of the information domain’.¹⁴ ‘Massive Internet of Things’ may be a Trumpian term, although one not as extreme as ‘Tremendous Internet of Things’, even though ‘tremendous’ is a term also used in this memo.

The emergence of Huawei as a global leader in 5G was received not as a mere economic challenge but as a national security threat. By deploying its most successful company worldwide, the Chinese state would not just be leaping to the frontier of technological innovation; it would place itself in a position to build the virtual worlds of the future and, as a result, to effectively control how they work or may work. ‘Telecommunications is both a homeland security and a national security issue because the entire economy depends on a shared, interoperable infrastructure, and because many national security communications ride on top of commercial infrastructure both in the United States and abroad.’¹⁵ On top of that, as we just saw, 5G is a new form of telecommunications network capable of onboarding most if not all economic and social activities. The concern was that Huawei, as it built the internet of the future, would have access to its control panel for the purposes of blackmail, influence, sabotage or intelligence and data collection. The virtual world is not neutral. It is endowed with certain qualities and purposes by its creators. American analysts and policymakers were openly afraid that, were nothing to be done, we might soon be living in a Chinese world, a world built and controlled by China.

European providers used Nokia and Ericsson equipment but were increasingly turning to Huawei due to its better prices and ‘advanced’ capabilities that could be readily deployed. The Dutch telecommunications carrier in April 2019 cited a 60 per cent price advantage when choosing Huawei. By 2019 Huawei seemed to be on the verge of conquering Europe.

In November 2017 the National Cyber Security Centre in the United Kingdom publicly stated that they had seen evidence of Russian attacks against British telecoms networks. The targeted networks did not contain Russian equipment but were affected by architectural weaknesses that the attackers were able to exploit. The need to mount cyber operations might disappear if states control the very networks they intend to hack. Then the bug becomes a feature, and network equipment acts as a backdoor by granting vendors access to all of the data, information and systems communicated or provisioned over the network. Those who control the system may also deploy ransomware and lock down factories, power plants or even military assets. In 2019 Norwegian aluminium manufacturer Norsk Hydro suffered an attack that forced the company to switch some operations to manual mode.¹⁶

Traditionally, cryptography has been seen as defensive in nature. Even in cryptocurrencies, cryptographic keys are used to secure the network. In reality, cryptography can just as easily be used offensively, as ransomware actions have decisively proved. Adam Young and Moti Jung called this concept ‘cryptovirology’.¹⁷

Cryptovirology, to stick with the term, has so far been applied to criminal activity aimed at extorting sums of money from its victims. The next step is for cryptovirology to be used for geopolitical purposes, often in situations short of military conflict. One interesting question is whether cryptovirology has already been used in a geopolitical context and we simply have not heard about it. There are good reasons why its use might have been kept secret, so no one can state for certain that this is an area lying entirely in the future.

A representative of a Russian ransomware group – responsible for the breach at the software company Kaseya, among other attacks – said in an interview that several of its affiliates have access to a ballistic missile launch system, a cruiser, a nuclear power plant and a weapons factory.¹⁸ He added that none of them was seriously considering taking over these assets: it was simply not profitable. In the same interview, the group representative showed an acute awareness of the geopolitical situation, arguing at one point that ‘with the current geopolitical relationships, everything is very beneficial for us even without any interference’. The Kremlin will give it carte blanche in exchange for leaving Russian companies and assets alone; or rather, leaving Russia alone and creating all kinds of havoc for its main adversaries.

Cryptovirology offers one great advantage for geopolitical actors. As Jenny Jun puts it, it removes the trade-off between the need to demonstrate the capability to carry out an attack and the need to maintain a covert presence until the final payload is dropped.¹⁹ When North Korea threatened a cyber-attack against South Korea’s power plants unless its demands were met, the articulation of this threat prompted a lockdown on the plants as well as security monitoring and drills. Ransomware, however, bypasses this trade-off entirely because the attackers only need to stay hidden up to the point of executing the encryption.

If an adversary state is able to acquire control over critical assets, it could be in a position to ask for significant concessions in return for releasing them. Or it could just use cryptovirology to paralyse any responses. Imagine half of a nation’s navy is suddenly unable to move because an attacker has encrypted the software and data needed to operate it, but this attacker will provide the encryption key only if that nation withdraws from a given military alliance … Too fanciful? Then imagine instead that a nation’s air force is paralysed on the eve of a ground invasion.

As Christopher Ford, Assistant Secretary of State for International Security and Nonproliferation, wryly put it in a speech in 2019, he could only wish it were true that the distress and anger in Washington over China’s strategic goals ‘related solely to matter of dollars and cents’. In fact, as he saw it, Chinese officials had made it clear that in some sense their target was the ‘operating system’ of the world order. ‘Ultimately, China seems to think that it really can reorder the world,’ Ford added, quoting what a Chinese ambassador exclaimed some years ago during negotiations over joining the World Trade Organization: ‘We know we have to play the game your way now, but in ten years we will set the rules!’ Interestingly, even though China wanted to set the rules of the new game, it seemed to have grown in its capacity to do so while playing under the previous set of rules. What was the lesson here? According to Ford, the game had moved up a level. In order to keep on winning, the United States should use its power as the incumbent to shape and reshape the rules as circumstances changed. That is what the Trump administration was planning to do by reforming national security export control rules and recalibrating export control policy and closing ‘loopholes in our traditional methods of screening foreign investments in the United States for national security implications’, among many other examples.²⁰

Suddenly, in or about 2019, the rules changed. If you thought technological development and economic competition were subject to the rules of the marketplace, what happened after 2019 came as a surprise. Decision-makers in Washington did not see the economic game as something governed by natural, more or less permanent rules long defended by free market economists. The rules were programmable, and those with root access to the code would be foolish not to set them in ways that would bring about favourable outcomes. Contrary to what two authors have recently argued, this was not a question of weaponising interdependence patterns arising naturally between nations. These patterns are built according to certain models. The ‘asymmetric networks that make up much of the structure of a globalised world’ are indeed ‘constructed as tools of statecraft’.²¹

In a way, of course, the question is how to reach the deepest layer of the system in order to reprogram its most basic operating rules. Here Washington had some determinate advantages. Huawei could be deprived of the advanced chips it had been acquiring in global markets because those chips were still dependent on American technology. It could also be deprived of those valuable European markets it needed to succeed because European nations needed their telecommunication networks to remain compatible with the secure transport of American intelligence. ‘Drop Huawei or See Intelligence Sharing Pared Back,’ Ambassador Richard Grenell told Germany.²² The rules ceased to apply to Huawei – an exception was introduced to the rules of the game – and governments such as the United Kingdom were even able to point out that they could no longer rely on a player that had for all practical purposes become an outcast. The National Cyber Security Centre was blunt in its assessment that, given the ongoing – and escalating – measures against Huawei, and notwithstanding that British authorities themselves had nothing against the firm, it would not be sustainable or advisable for British operators to continue to rely on Huawei for the operation of their networks. This was due to the likelihood that a future and predictably brusque measure from Washington could seriously inhibit or impair this essential support at short notice, and in doing so disrupt network services.²³

Second, and to avoid a repetition of the Huawei story, new rules would be established for the future, organised around open interoperable standards. Companies making equipment based on open standards allow wireless carriers to mix and match the antennas with different hardware and centralised electronics. The basic concept aims to replace traditional network proprietary technology with software-driven technology that will run on any piece of hardware. ‘We may be able to increase security, reduce our exposure to any single foreign vendor, lower costs and push the equipment market to where the United States is uniquely skilled: in software,’ the Federal Communications Commission Acting Chairwoman said in March 2021.²⁴ If that did not work, the United States might be prepared, as Attorney General Bill Barr said in 2020, to buy one of the only two viable rivals to Huawei: Ericsson of Sweden or Nokia of Finland. Andrew Ross Sorkin of the New York Times commented: ‘The idea of the Trump administration buying a major telecom is a curious proposal from an administration that is currently decrying the rise of socialism.’²⁵ That was to miss the point. The administration had by then made it amply clear that it did not subscribe to any notion of rules divorced from particular outcomes.

After Huawei was added to the Entity List in 2019, reports emerged that many American companies were able to use foreign subsidiaries and affiliates to continue exporting to Huawei under these rules. A new rule approved in May 2020 prevents any company from selling to Huawei without a licence if the product they are selling has been developed or produced with tools sourced from the United States. In August 2020 the Trump administration expanded those export restrictions to include chipmakers based anywhere in the world that use American software or equipment to produce semiconductors, an addition that analysts argued would deliver a mortal blow to Huawei. In January 2023, the Biden administration announced it would stop providing companies with licences to sell to Huawei.

The rules had changed, but more importantly they seemed to be breaking apart, giving way to two internets, two technology spheres, two separate spheres of influence governed by two sets of principles and values. A new Iron Curtain was descending upon the world.

What made the Cold War unique was the existence of nuclear weapons, which rendered a final war impossible by integrating the two contenders within the same technological system. Things have continued to progress in the same direction. During the Cold War it was still possible to delimit two geographic areas under the control of Washington and Moscow. You could say the natural element was still present. Neutral space made something like the separation between these two spheres logically conceivable. Today, we inhabit a fully developed technological system. Can Washington and Beijing break it apart into two spheres? There is no natural environment here. If one state refrains from creating its own environment, then it will be created by another state. Space is political, or rather geopolitical. It has disappeared as a neutral, physical medium.

Imagine the crew of a ship is divided into two groups battling for control of the ship. Can they divide the ship into two separate spheres? Of course not. The conflict is about who will succeed in taking control over the ship as a whole, the ship as a technological system …

The Art of Decoupling

Confusion continues to reign on the issue of decoupling. On the one hand, authorities in Washington and Beijing regularly announce new initiatives to limit the free flow of capital and technology between their two countries, and many other state actors are not far behind. Economic connections with Russia are being severed, but in this case the root cause is less what one might call decoupling than conflict: during conflict and war, economic and social links are severed for the very simple reason that they cannot take place under conditions of radical insecurity. Connections also break down inside countries during civil wars …

The model of the Cold War is often in the background: again, the underlying theory here seems a simple one. The conflict between the West and the Soviet Union during the Cold War made most exchanges radically insecure and so they were discontinued. Conflict, not decoupling. There were always exceptions, when state authorities on both sides were able to carve out special areas under some sort of mutual protection, and there was even a deepening of economic exchanges after 1975, particularly with the Eastern Bloc. We take up this question in Chapter Three. But we also talk of decoupling in cases of much deeper integration, which is where the concept becomes somewhat paradoxical. There is as of now no Cold War between China and the United States. A Cold War may yet develop, and in that case many of the existing economic and social links would disappear. The current situation, however, is peculiar. In some areas, such as microchips, integration is being reduced. In others it is still flourishing. In the first three months of 2024, the United States exported to China crude oil, soybeans, integrated circuits and aircraft parts for a total of $40 billion, and it imported computers, telephones, electric batteries and motor vehicles for about $110 billion. Over the first half of 2024, China’s exports to the United States were up by 1.5 per cent from the previous year. American investment banks are still rushing into China: Citi is aiming to launch its wholly owned China investment banking unit by the end of 2024, having applied for a brokerage business licence in late 2021 as part of its push to ramp up its presence in China.²⁶ The two economies remain highly dependent on each other, and no one can see how they could safely be broken apart.

What is most revealing, however, is that the United States may have remained just as dependent on supply chain links to China, even as its direct engagement with the country was decreasing. Because the main alternatives to China such as Vietnam or Mexico have seen their import links with China step up progressively over time, indirect links remain intact and may even have intensified. For Vietnam, goods from China were 9 per cent of its total imports in 1994, and this has surged to 26 per cent in 2010 and approximately 40 per cent by 2022; the main items that China ships to Vietnam include integrated circuits, telephone sets and textiles. To the extent, then, that Chinese firms increase their exports of parts and components to these locations, where they are assembled into final goods and shipped to the United States, China would ‘continue to be a relevant player in the upstream stages of US supply chains’.²⁷ Another paper considered network firm distance and concluded: ‘The lengthening of the distance between suppliers in China and customers in the United States suggests that firms from other jurisdictions have interposed themselves in the supply chains from China to the United States.’²⁸

One can only make sense of the paradox of ‘decoupling with integration’ by drawing a distinction between two levels of action: the geopolitical and the economic. At the geopolitical level, actors try to concentrate power and will build all kinds of barriers to prevent their own leverage from flowing elsewhere. Consider the example of the most advanced chips: since they are a source of geopolitical power, the priority becomes to prevent your geopolitical rivals from having access to them or, at least, from having exclusive access to them. In certain circumstances, energy might also rise to the geopolitical level. At the same time, integration is encouraged at the economic level: there are no questions of power when it comes to the production of footwear, toys or chemicals. Whether Chinese companies win contracts to build a road somewhere in the world comes under the rules of economic competition, but prima facie there is no geopolitical question.

Decoupling is a strategic shift whereby states switch their focus from economic growth to economic control.²⁹ In the United States, decoupling from China emerged as a strategic goal with the growing concern about the security of supply chains and the struggle for dominance over key technologies. In 2017 Peter Navarro, then Director of the Trade Council in the Trump administration, argued in a public speech that global supply chains endangered national security. ‘Suppose that it is not a benign ally buying up our companies, our technologies, our farmland and our food supply chain, and ultimately controlling much of our defence industrial base. Rather it is a rapidly militarising strategic rival intent on hegemony in Asia and perhaps world hegemony.’ The administration’s policy would seek ‘to reclaim all of the supply chain and manufacturing capabilities that would otherwise exist if the playing field were level’. It was the opening salvo for the decoupling programme.³⁰

The first mention of ‘decoupling’ with this meaning that I was able to find is from August 2018, reflecting what by then was becoming a popular debate in the Washington think tank world.³¹ In 2019, speaking in Beijing, Henry Paulson argued that the notion of decoupling expressed the replacement of metaphors of economic competition by metaphors of military competition. An economic Iron Curtain was falling upon the world and on competition moving outside the realm of existing rules. But Paulson also warned that forging an exclusive technology bloc might be counterproductive. If the point of decoupling was to contain Chinese power, an Iron Curtain would reduce America’s field of action. Decoupling China from its financial markets could in time threaten US leadership in finance, just as sequestering technology inside its borders would make it more difficult for the United States to set global technology standards. If decoupling is a strategy aimed at global leadership, it must be made compatible with globalisation. Therein lay the paradox. In his speech, Paulson seemed attracted to the concept of ‘strategic decoupling’ – that is, a type of decoupling limited to the ‘most sensitive advanced technologies vital to national security, our communication backbones, and Internet governance’. Elsewhere, globalisation might be able to survive.³²

Somewhat similarly, European Commission President Ursula von der Leyen stressed in a 2023 speech that the European Union does not intend to cut economic, societal, political or scientific ties with China. Rather, the focus should be on reducing or eliminating the risks created by economic dependence, ‘where trade and investment poses risks to our economic and national security’. To achieve this goal it will be necessary to acquire more independence and diversity when it comes to the key inputs needed for economic competitiveness and the energy transition. Von der Leyen noted that the European Union relies on China ‘for 98% of our rare earth supply, 93% of our magnesium and 97% of our lithium’. Simultaneously, the European Union needs to ensure that European ‘capital, expertise and knowledge are not used to enhance the military and intelligence capabilities of those who are also systemic rivals’, such as China. ‘So we have to look at where there are gaps in our toolbox which allow the leakage of emerging and sensitive technologies through investments in other countries.’³³

In China too the emphasis is on the return of radical insecurity to global politics. The decoupling school in China ‘insists that greater interdependence always leads to greater economic security risks’.³⁴ Already in 2018, two Chinese scholars, Wang You and Dingding Chen, argued that decoupling follows from ‘the rise of strategic mutual suspicion’. In the thirty years after the establishment of diplomatic relations between China and the United States, the interests of the two countries in economic and trade cooperation had been expanding and their interdependence increasing. But then things changed. At some point, the United States stopped expecting China to ‘be more like itself’ – the theory of convergence – while China opened up a path for the development of its own great power ambitions. In this new context, the authors thought that not only would the economic links between the two countries be reduced, but also political relations would be strained and strategic cooperation weakened or destroyed.³⁵ At the heart of the process lay the political differences and ideological confrontation between the two countries, which became increasingly marked after 2015, as the theory of convergence was gradually abandoned.

Ultimately, the meaning of decoupling is connected to the role of rules in global politics. If a set of rules is generally shared and accepted, states can focus on competing under those rules. It is very different when rules are openly contested. In that case, states will focus on acquiring and preserving the greatest measure of control in setting the rules governing the global system.

It is easy to imagine a situation in which the system of rules breaks down. Decoupling would then take the form of a radical retreat from globalisation. Imagine, however, that states accept the existence of common rules but disagree on what those rules should be. In such a case, global trade tends to bifurcate. Some goods and resources may still be exchanged on the global marketplace, but strategic resources will be carefully guarded and jealously kept. Strategic resources are those granting their holders the power to shape global rules, the rules governing the trade and exchange of everything else. ‘For example, Black Crows, a French maker of freestyle skis, sells some but not many skis in China and has minimal upstream activities there. Because skis are not a targeted segment, as long as its products are superior to those of Chinese rivals, it may be left alone to grow.’³⁶ Strategic industries will fare very differently.

The term ‘decoupling’ would seem to refer to the process by which certain states are excluded from access to sensitive areas of the global system, areas one might call the deepest layer of the global operating system. While ‘the spatial logic of the Cold War was characterised by control over territory’, the ‘spatial logic of the Second Cold War is altogether different and geared towards control over networks and their structures’. The strategic goal – a goal shared by the United States and China – is to exert control over those global networks, conduct the movement of flows throughout networks, exclude rivals and reduce vulnerability to foreign interference.³⁷ There is no reason for the affirmation of different states qua sovereign actors to be regarded as a move away from globalisation. It is just that being global will translate into the ambition to shape the global rules rather than the act of following a set of naturally given rules. What we see is the emergence of global industrial policy, with the concentration of power in the hands of state actors that expect to use this power to shape the rules of the global economic system. The model of how countries work within their borders has now been replicated at the global level.

After the Global Financial Crisis there were many attempts to think about a deglobalisation scenario and what it could mean for the global economy. As Evan Hillebrand put it, ‘What happens to the global system if tariffs rise substantially, if international investment capital and aid flows are reduced, and migration diminishes? Is economic growth increased or decreased? Does poverty rise or fall?’³⁸ His model estimated large falls in world economic growth because trade flows are linked to efficiency gains in production, but the results were sufficiently ambiguous to predict, for example, that poor countries might be able to increase the relative size of their domestic manufacturing industry and shift the relative wage structure in such a way that would increase overall equality. Be that as it may, this debate was about a generalised move away from globalisation, where every state would reduce global flows of goods, people and capital with every other state.

The debate around decoupling in recent years is fundamentally different. In a way, it questions much more radically the foundations of the global order. The deglobalisation debate presupposed that states rather than markets would order the basic framework of global relations, simultaneously creating the space for their internal affairs and the need to respond to democratic aspirations at home. A kind of social democracy or social market economy might be accepted more or less universally. Today, it is much less clear that such an agreement can take place. With the sense that the global order can be reshaped or redesigned according to often contradictory ideas or plans, there is also the recognition that certain economic and political areas or resources are definitionally outside the purview of common rules because they concentrate the power to set the rules. At this point in the argument it is easier to see why sovereignty might become an imperative. Where no common rule can reach, there can be no system of exchange. We may expect dynamics of competition to prevail when insecurity prevails, but note that insecurity here has an ontological character: it is not possible to subject the power to set the rules to rules that arise from that power’s own activity.

There is a lot in the contemporary debate on decoupling that reminds one of the old struggle for key fortresses or chokepoints in global trade routes. The Portuguese merchant Tomes Pires said in the sixteenth century that ‘he who controls Malacca has his hands on the throat of Venice’. The capture of Malacca by Portugal in 1511 was an important step in the drive to divert the spice trade from the Red Sea and the Persian Gulf to the route round the Cape of Good Hope leading to Lisbon. Goa in India and Hormuz at the entrance to the Persian Gulf were other strongholds in the Portuguese crown’s grand design of controlling the exits to the Indian Ocean. As Afonso de Albuquerque, Governor of Portuguese India at that time, put it: ‘I hold it very certain that if we take this trade of Malacca out of Muslim hands, Cairo and Mecca are entirely ruined and to Venice will no spices be conveyed except that which her merchants go and buy in Portugal.’³⁹

For as long as power politics remains concerned with gaining control over physical territory, geography will be among its core subjects. Which regions, cities and fortresses offer the best prospects for acquiring control over industry or territory? Which chokepoints can grant a state control over the main sea lanes? There must be a science or at least a certain number of principles guiding the statesperson or the general in their decisions. The great theorist of sea power and geopolitics Alfred Thayer Mahan wrote at length about the geographical positions of different countries, their advantages and disadvantages and the search for those critical global chokepoints that have for centuries functioned as force multipliers. Gibraltar, Malta, Corsica – these geographic positions may well have changed the balance of power in Europe. Mahan liked to quote the aphorism by Napoleon: ‘War is a business of positions.’ In his time, however, the focus had already shifted, rather intriguingly, to the ways in which the globe could be remade. Once that happened, the key sources of power started to move from the geographical to the technological plane.

After discussing some of the chokepoints already in existence, Mahan turns to a future or potential one: the Panama Canal, whose construction was to begin a little over a decade after the publication of his The Influence of Sea Power Upon History. If a canal be made, Mahan wrote in his book, ‘and fulfil the hopes of its builders, the Caribbean will be changed from a terminus, and place of local traffic, or at best a broken and imperfect line of travel, as it now is, into one of the great highways of the world’. At this point, the geopolitical question is less about how to control the world in its current form than how to build a new world that more significantly furthers our interests. In the Panama Isthmus, nature had so to speak built a great connecting station between two great seas. The control over these ‘decisive regions’, as Mahan calls them, inevitably becomes an object of competition and conflict between states, who know how their acquisition can exert a vital influence upon global trade. As Mahan saw it, a seafaring canal across the Isthmus would help bring about a reorientation of global trading power from Britain to the United States. Before the canal, New York and Liverpool were about equidistant from the west coast of North America, from Mexico to British Columbia. After the canal, the line of equidistance would pass through Shanghai and Melbourne, assuming British ships would use the Suez. And that a canal would be built there could be in no doubt, given technological progress and economic demands. For Mahan, the question was which state would succeed in controlling it. After the canal, it would become much more difficult for the United States to remain aloof from geopolitics: ‘The position of the United States with reference to this route will resemble that of England to the Channel, and of the Mediterranean countries to the Suez route.’⁴⁰ The case of the canal thus illustrates the central thesis of this book: one of the main reasons to build a new world is that those who build it will be in a privileged position to rule it.

As Xi Jinping put it in a speech in April 2016, internet core technology is the greatest ‘vital gate’, and the prospect that such core technology would be controlled by others is the greatest hidden danger. To be dependent on the outside world for core components, if the ‘vital gate’ of the supply chain is grasped by others, is like building a house on a shaky foundation. And he went on: ‘Having determination means that we must establish an ambition of fighting indomitably and storming strongholds in defiance of difficulties, unwaveringly implement the strategy of innovation driving development, direct ever more human power, material resources and financial input into core technology research and development, gather crack forces, and make strategic arrangements.’ Strategic thinking means to ‘closely focus on climbing the strategic commanding heights’. The metaphors Xi appealed to were openly military, stressing the importance of passes, gates and strongholds: ‘We must assault the fortifications of core technology research and development well, we must not only call forth the assault, we must also sound the call for assembly, which means that we must concentrate the most powerful forces to act together, compose shock brigades and special forces to storm the passes.’⁴¹

If straits and islands are the gates to the oceans, microchips are the gates to the virtual. In their countless billions or trillions, ‘they are the brain cells driving the modern information age’. ‘It is as if individual grains of sand had been granted the power of thought and memory.’⁴² The critical operation takes place in the transistor: open or closed, it provides an input or output for logical thinking, since binary logic can easily turn an inert system of consecutive switches into a computing tool. Already in 1961, Harvey Cragon at Texas Instruments built a computer made from chips. About the size of a transistor radio, it relied entirely on electronic pulses, manipulating them to perform logical operations. The pulses represented binary numbers, a string of zeros and ones corresponding to the two possible states of transistors – on and off – directing current to other transistors down the line at the speed of light or something approaching it.⁴³ ‘With integrated circuitry, the neat patterns of Boolean logic could be mapped directly onto the surface of a silicon chip; an entire addition circuit would now take up less space and consume less power than a single transistor did in the days of discrete components. With the advent of the chip, the digital computer had finally become as elegant in practice as it was on paper.’⁴⁴ One of the first uses of integrated circuits was the modern microwave with its sequential cooking functions: turning the power on and off repeatedly in an intelligent or programmable way. When Amana unveiled the first microelectronically controlled oven in 1975, the audience was mesmerised: an oven with a brain.⁴⁵ Then came chips that tune themselves to any incoming input frequency, thus functioning as a variable frequency generator for mobile communications, including mobile phones. The synthesiser uses division, multiplication and mixing operations to generate the desired output.

Assemble these artificial brain cells and a new world takes shape: with an objective or external existence, they nonetheless mimic the workings of the human mind. The virtual world aims to resemble the physical while being subject to rules of our choosing and remaining a creature of our desires. The goal is the injection of mind at the level of the structure of matter.

Much or most of our lives now takes place in this virtual world. The writer Jeffrey Zygmont asks us to consider a traditional couple living somewhere in, say, Niagara Falls. They are suspicious of technology but addicted to baseball, so they let cable television enter their safe haven. But cable television, of course, involves streams of digitally encoded signals beamed by a satellite, and a satellite contains millions and millions of electronic parts, themselves designed and constructed by variations of the same microcircuits. Each little pixel of the footage is represented by endless strings of binary numbers processed at close to the speed of light by the latest generation of microchips. The fundamental technology is not computers or television but the essential component that accounts for their visible presence: the microchip.

Zygmont fails to note one important element of our historical moment: the virtual world he describes is now embedded in the physical one. The television set is a physical object, but one created to exist within a virtual world of electric signals. Microchips are so small – so infinitesimally small – that they can provide the invisible foundation for a world built anew. They have become so ‘penetrating’, as Zygmont puts it, that they guide our lives and activities without ever being noticed. When you drive to the shopping mall, the idea of driving is built into the decision of going to the mall, and the microelectronic controller that makes the car engine run is built into that original decision.⁴⁶

The ontological status of objects such as the television set or the digitally wired automobile remains intriguing. Their existence in the physical world is indisputable, but that existence is strictly subordinate to the action taking place in the virtual world. People want a television set not as a physical object but as a portal to the virtual dimension where, for example, a baseball game is being recreated as a stream of electric signals.

The virtual ultimately requires the existence of the physical world as a base reality against which virtual worlds – the plural is important – can be measured and understood. It seems that reality is a childish belief we have outgrown, but also one worth keeping as a warning against the old error: one uses reality only to deny every claim to it. If something is not real, that does not mean that something else is. Reality today has been stripped of content and subsists only as an empty form, but reality as an empty form still performs the important role of preventing a virtual world from assuming the place of the real one.

Already in the didactic poem of Parmenides, philosophy begins by declaring its topic to be the real but, humorously, has almost nothing to say about it, while a lot more could probably have been said about the unreal.⁴⁷ In the history of metaphysics, reality was never able to rise to the level of a positive concept, but its role was nonetheless essential as the wedge with which to apply the tool of criticism to the many positive concepts being advanced as foundational. Reality served above all to maintain the modern pluralism of worlds, what Nelson Goodman called ‘a diversity of right and even conflicting versions or worlds in the making’. The world interpreted as a creation or a construct must be distinguished from something like the real world, but the import of this distinction is simply this: the world as a construct can be created and destroyed and replaced by another construct. Goodman opposes the idea of reduction to a common base, to the physical or anything else. The concept of creation can refer only to the void preceding that creation.⁴⁸

In terms of a simple scheme, one could say: there is at first an effort, in a kind of prehistory of humanity, to build a natural platform for virtual worlds, an empty screen or portal, but all this is but a preliminary stage to the appearance of the metaverse, when the empty screen comes alive with living images, the work of the imagination. At that point the geopolitical struggle to control the portals granting us access to the virtual will inevitably begin.

We shall consider in Chapter Four the sources of energy needed to power these new digital worlds, for energy is just as much a building block of the virtual as the integrated circuit. But for now it is important to see how the microchip grants its creator an almost unlimited power over the world.

Skyscrapers of the Infinitesimal

In 1965 Gordon Moore from Fairchild Semiconductor, the legendary microchip pioneer, was asked to speculate on the future of integrated circuits. He looked around the lab, where engineers were starting to place roughly sixty components on a chip – a number that had doubled approximately every year since the origins of the planar chip design in 1959. He then made an astounding prediction: complexity and miniaturisation would keep doubling every year, so that, in a decade, a chip would have 60,000 components, even as prices and costs continued to plummet. And so it came to pass, more or less, with the prediction for 1975 being remarkably close to the mark. Today, the most advanced chips have complexity counts above 100 billion.

It should come as no surprise to learn that we may now be close to the end of this process. But what is the meaning and significance of this limit? Where have we quietly been heading in the mad dash that Moore first described?

The latest microchips can perform tens of trillions of calculations per second and build perfect simulations of physical stores, where shoppers can grab objects from virtual shelves and get billed through their mobile devices as they leave. We are trying to squeeze every possible ounce of computing power out of semiconducting materials, seemingly anxious that the reservoir will run dry before we reach the goal. But what is the goal, if not a smaller and more powerful smartphone?

Here is what the final stretch might look like: large virtual environments rendered in real time, with data transferred at super-low latencies for hundreds of millions or billions of users sharing the same persistent life-world. The immersive experience of your favourite games and platforms was powered by the last generation of microchip breakthroughs. Truly persistent and immersive computing, at scale and accessible by billions of humans in real time, will require even more – perhaps a thousand times the computational efficiency available today. But as Raja Koduri, formerly of Intel, puts it, ‘the dream of providing a petaflop of computer power and a petabyte of data within a millisecond of every human on the planet is within our reach’.⁴⁹

As Moore noted in later reflections, materials are made of atoms, so the process should reach an end when we start manufacturing transistors the size of a single atom. That has already happened: scientists at Tsinghua University in Beijing announced in 2022 that they had built a graphene transistor gate with a length of 0.34 nanometres, roughly the size of a single carbon atom.

Drawing features the size of an atom or a few atoms is a miracle of the transcendental. What tool could you even use? Would it have to be as small as the features you want to draw? In reality, of course, you draw them with light, not with solid materials, in a process that some compare with black magic.

With the lithographic method, the minimum feature size is constrained by the minimum wavelength that can be applied. To produce the most advanced chips available today, you vaporise droplets of molten tin with two successive laser blasts. The resulting plasma emits extreme ultraviolet radiation, with a wavelength of only 13.5 nanometres. Various tricks and tweaks like shooting light through water or through multiple ‘masks’ enable us to build transistors even smaller than the wavelength of the light used to pattern them. The clean rooms in which these transistors are drawn are illuminated with a special yellow light that contains no ultraviolet radiation. With this kind of light, it is no longer possible to rely on lenses, so mirrors are instead deployed to guide the light to a silicon wafer and draw transistors with features measuring 5 nanometres or less, the size of just a few atoms. By comparison, a fingernail grows, on average, 1 nanometre per second. Zeiss, the company responsible for these mirrors and sensors, claims they are so precise that they could be used to aim a laser to hit a golf ball as far away as the moon.

After a layer of transistors has been drawn, many other layers need to be superimposed. Semiconductor chips are the skyscrapers of the infinitesimal. The more layers, the more complex and powerful the chip. The most advanced designs can have hundreds of layers, and they all need to align with nanometre precision. The ratio between the height and surface area of a sixty-four-layer memory chip is equivalent to three times the height-to-surface area ratio of the Burj Khalifa in Dubai. And we are already producing chips with 256 layers.

The point of miniaturisation is less about size than power. Electrons must move through the semiconductor material as fast as possible. The smaller the distance, the faster and more powerful the chip can be. Shrinking the transistors by half on a chip of a given size yields approximately four times the computing power because both the number and speed of the transistors have been increased. Approaching the limit of miniaturisation thus means approaching the limit of how the universe works: a construct built from the smallest, most elementary particles possible; rather than atoms, transistors the size of atoms.

The Intel 4004, at the size of a fingernail, delivered the same computing power as the first electronic computer built in 1946, which filled an entire room. It was the first programmable processor on the market, following software instructions to perform many different functions on various devices. Released in 1971, it was built of approximately 2,300 transistors – a paltry sum by more recent standards, but the chip made it clear that these powerful microscopic machines would soon have the capacity to construct worlds of their own.

Chips are strange creations. If space exploration represents the human search for the infinitely large, chips show us what Blaise Pascal called the ‘abyss of nothingness’, the vanishingly small. ‘Let him see therein an infinity of worlds, each of which has its firmament, its planets, its earth.’⁵⁰ And its skyscrapers.

From Ukraine to Taiwan, there are now two different levels of conflict. On one level is the war happening on the streets of Ukrainian cities. On the other is the fight to control the skyscrapers hidden within our vanishingly small chips. For all the human tragedy of the former, I cannot shake off the conviction that the future of the world order will be decided by the latter. It is the latter, after all, that will decide what weapons you can bring to the battles happening in the human world.

When the war against Ukraine started, one of the main goals for the Western coalition was to deprive Russia of access to the kind of chips it needs to build aircraft and missiles or to modernise its economy. It was vital that Taiwan joined the sanctions, which it did. China quickly became an alternative supplier for Russia, but China is not yet capable of producing the most advanced chips. ‘The fact that Russia faced shortages of guided cruise missiles within several weeks of attacking Ukraine is also partly due to the sorry state of its semiconductor industry.’⁵¹ By contrast, Ukraine has received huge stockpiles of guided munitions from the West, such as Javelin anti-tank missiles, and each missile contains upward of 200 microchips. The message from the West was clear: were China to invade Taiwan, it would be similarly excluded from access to semiconductors, and the consequences might be catastrophic for an economy with such a voracious appetite for chips.

One obvious complication is that advanced chips are predominantly manufactured in Taiwan. A war in Taiwan might deprive everyone, not just China, of the latest generation of chips, triggering a worldwide industrial depression: ‘These days, when we look five years out we hope to be building 5G networks and metaverses, but if Taiwan were taken offline we might find ourselves struggling to acquire dishwashers.’⁵² Two geopolitical analysts argued in 2021 that the best way to stop a Chinese invasion of Taiwan would be for its government to announce in advance that the Taiwanese semiconductor industry would be destroyed in the initial moments of an invasion.⁵³ That seems unnecessary. There is no danger that Chinese forces will simply seize – intact or almost intact – the industry on the island, which depends on high-precision machinery, deep and delicate global networks and large inputs of human capital. There may be other safety measures too: in May 2024, it was made public that ASML, the company producing the world’s most advanced chipmaking machines, had assured the Dutch government that the company could remotely disable its machines in Taiwan in case of a Chinese invasion. The procedure would act as a kill switch.⁵⁴

But imagine China imposed a blockade on the island, using its navy to force customs checks on ships sailing out of its ports and demanding that Taiwan Semiconductor Manufacturer Company (TSMC) restart chip production for Huawei and other Chinese technology companies. Chen Wenling, chief economist at the China Center for International Economic Exchanges, argued in May 2022 that China ‘should seize TSMC’ as a way to reconstruct the Chinese semiconductor supply chain, battered by export restrictions and threatened by the looming transfer of TSMC chip fabrication plants, or fabs, to the United States.⁵⁵ As Chris Miller puts it, in this indirect way ‘Beijing could conceivably gain influence or control over the only fabs with the technological capability and production capacity to churn out the chips we depend on.’⁵⁶ There would be an American response, of course, but at this point events become more difficult to predict. New export restrictions would be imposed on China and also, perhaps, on Taiwan. The two blocs would be entering a critical phase in which control over the fundamental gateways of the world system would be decided.

I believe it was exactly to try and pre-empt such a perilous clash – one with a very uncertain outcome – that Washington approved a set of draconian measures against the Chinese semiconductor industry in October 2022. The ultimate result of a chess game is decided a few moves ahead. Or, as Zoltan Pozsar cleverly puts it, the United States is trying to ‘invert time’ using technology sanctions. Like in the movie Tenet, ‘inversion’ is being used to shape future outcomes before they happen, ‘one technology sanction at a time’.⁵⁷

Ensuring China was no longer dependent on overseas chip manufacturers became a priority for Beijing after Washington denied Huawei access to American software and hardware components. Much of this remains top secret, but a rare report from Nikkei Asia in 2021 detailed the creation in China of a group of industry insiders and political officials whose mission was to review all components needed to develop native semiconductor capacities and create a new network of technological doppelgängers.⁵⁸

Over the summer of 2022, China’s largest foundry, Semiconductor Manufacturing International Corporation (SMIC), took everyone by surprise when it became known that it had produced a 7-nanometre chip. SMIC was not expected to proceed so quickly to 7-nanometre manufacturing because it had been denied access to the extreme ultraviolet lithography equipment from Dutch monopoly supplier ASML. In 2019, the Trump administration mounted an extensive campaign to block the sale of Dutch chip manufacturing technology to China. The effort began in 2018, after the Dutch government gave ASML a licence to sell its most advanced machine to China. On 18 July 2019, Deputy National Security Advisor Charles Kupperman raised the issue with Dutch officials during the visit of Dutch Prime Minister Mark Rutte, who was given a classified intelligence report. Shortly after the White House visit, the Dutch government decided not to renew the export licence, and the $150 million machine was never delivered to SMIC.⁵⁹

And yet, two years later, the Chinese giant was making impressive progress. The news seems to have spurred the United States into action. How do you stop China from developing high-end chips? It was a notable achievement for SMIC to take only two years to reach 7-nanometre chips from 14-nanometre chips, and in some obvious respects Western sanctions had furthered rather than hindered this process. Imposing export controls on advanced computing semiconductor chips creates a captive market for Chinese manufacturers. Revenues that formerly flowed to US chipmakers would have no option but to go to Chinese companies, whose business model would become a lot more attractive. Increasingly deprived of access to chips manufactured abroad, Tencent, Alibaba and Baidu might become the domestic chip industry’s best customers, potentially capable of bootstrapping new advanced capacities into existence. Absent geopolitical imperatives, there would be no reason for Chinese industry to try to compete with TSMC or ASML or any of the other specialised parts of the supply chain. It would always be easier to buy instead of build. Allowing Chinese firms the option to work with superior foreign partners meant local companies along the full value chain would struggle to get off the ground. Paradoxically, perhaps, allowing China access to greater capabilities in the present could help maintain China’s dependence on Western technology in the future.

Thus, on 7 October 2022, the Department of Commerce tried a different strategy. It targeted not chips but the whole semiconductor value chain.

Specifically, the new export controls included:

1. (1) Certain advanced and high-performance computing chips and computer commodities that contain such chips

2. (2) Certain semiconductor manufacturing equipment and related items

3. (3) Items to develop or produce semiconductor manufacturing equipment and related items.

Chinese companies were denied access to every item in the semiconductor value chain, including chip design software. With chips now containing more than a hundred billion features, they cannot be designed by hand. Electronic design automation software is a critical segment of the value chain, as valuable as lithography machines.

The goal is to preserve exclusive control over a specific source of technological power and to ensure that access to that source of power is denied to a rival through a system akin to concentric fortifications. One author speaks of ‘four interlocking chokeholds’⁶⁰: chips, manufacturing equipment, design software and components for manufacturing equipment.

When National Security Advisor Jake Sullivan set out to explain what he called ‘carefully tailored restrictions on semiconductor technology exports’ to China, he used the term ‘small yard, high fence’, which the new restrictions would implement: ‘Chokepoints for foundational technologies have to be inside that yard, and the fence has to be high because our strategic competitors should not be able to exploit American and allied technologies to undermine American and allied security.’⁶¹

The United States does not want to cut off broad economic links with China, but it will erect high barriers around technologies that it thinks are strategically important. Hence the small yard: restrictions are tailored to address specific high-performance computing and advanced semiconductor capacities but not necessarily committed to advancing a broad decoupling agenda. The measures announced on 7 October were designed not to affect the development, production or use of products at the trailing edge or mature node semiconductors, reserved primarily for consumer applications like mobile phones or washing machines. There are two large markets where older chip technologies and legacy semiconductor designs remain competitive: (1) chips for consumer devices that do not require the latest technologies; and (2) chips for safety-critical systems such as aerospace, automotive and infrastructure systems that are reluctant to change designs that have already made it through lengthy and complex system tests.⁶²

By only targeting chips with very high interconnect speeds, the Biden administration seemed to want to ensure that China would be able to buy parallel computer chips that are optimised to work in individual computers, such as those that are included in video game consoles. There are good reasons to be restrained: trailing edge factories still use a lot of equipment from American suppliers. In many cases, China is responsible for around a third of their revenue. Cutting off trailing edge fabs might create shortages in many areas important for American consumers, while the same American companies that have acquired dominant positions would have their revenue streams seriously impaired. Even with all these qualifications, export controls have significantly restricted the customer base for some American firms. A study published by the Federal Reserve Bank of New York in April 2024 found that a negative stock market reaction occurred immediately after the export control announcement and was economically significant, representing a 2.5 per cent abnormal decline in stock prices. The study estimates that export controls against Chinese companies cost the average affected American supplier $857 million in lost market capitalisation, with total losses across all the suppliers of $130 billion. The inability of affected suppliers to quickly find alternative customers outside China helps explain the result, with the authors concluding: ‘The expected benefits of export controls ought to be carefully weighed against the costs we estimate.’⁶³

This fundamental ambiguity should not be lost from view: the United States is trying to choke certain Chinese companies and capacities, but on the other hand, it is also hoping to continue selling its products to the Chinese market. Lu Feng, a Chinese political economist, calls it the ‘growing tension between choking China and selling to China’.⁶⁴

How are we to understand this tension? Chips are not a natural resource. They are not the new oil. Rather, they form the most basic layer in world building. They can be compared to the basic layer of a computer operating system. Take the metaphor seriously – as it is not just a metaphor – and two conclusions follow. First, the United States, as the original programmer of the world game, has root access to the code and thus the ability to rewrite it as it wishes. But the point is of course to continue running the program upon which its power and prerogatives ultimately rest, and thus broad economic links with China need to be preserved. Second, the more the United States uses its programming power to shape the global environment in its favour, the more will Chinese authorities attempt to hack the operating system, taking over its processes with a view to assuming control over the rules. Having control over chip production is no more than a simplified way to speak of having control over the source code governing the world game.

The reason for obtaining root access is to overcome the limits imposed on the system by its normal operations. As a root user of the global system, one has the ability to change any rules and settings, to create specific exceptions for every rule, to change the current game state or to execute operations that are inaccessible to normal users. Potentially, one could remove the entire operating system of the world game and replace it with another. The root account can also be called the superpower. Basically, the root account can do whatever it chooses.

Now how does the source code work and how can it be hacked? It is difficult to know of course. If it were easy, China would have done it a long time ago. But one can test and experiment with different payloads in the exploitation of certain system vulnerabilities. Lu Feng suggests a possible hack. The point is not to control individual technologies such as advanced chips but to seize control over the highly complex system of rules governing chip production. Advanced chips will then follow more or less automatically. Looking at chip production as a whole, the basic change to the system that China should be trying to introduce is, according to Lu, the development of a base of capabilities or an independent industrial chain. At this point, control over the system will already be in Chinese hands, even if a superficial observer might still be impressed by ‘technological indicators’. In fact, a fully independent production line for mature nodes is more important because it ensures a path for future development. ‘This is because capabilities are developed cumulatively via product platforms.’

Since chips are not a natural resource, they depend on the smooth workings of a system of production. Demand is ultimately more important than supply. As Lu puts it, ‘if you have a market, but no technologies, you can develop technologies. If you have technologies, but no market, then technologies will end up leading you nowhere.’ By developing the largest market for chips, China will be able to resist foreign pressure using its own domestic supply chain. Deprived of the bulk of their customers, foreign companies will shrivel, while Chinese ones will have all the time in the world to develop advanced technologies. ‘All we are missing now is for the Chinese government to make up its mind and take a decision. If we build up our industrial base, who will be most afraid then? It will not be the Chinese who are afraid, but the Americans.’

For the system hacker, the most important resource is not advanced technology but a deep understanding of the dynamics of the industry, of technologies and of the laws of market competition. Above all, a hacker needs to ‘develop the ability to grab hold of the other side by the throat’.⁶⁵ Conversely, the way to prevent a geopolitical rival from gaining access to the source code of the existing system is not through individual technological bans or restrictions but through extensive control over markets and supply chains, the system of production taken as a whole. Depriving China of markets or access to resources and foreign segments of global value chains would almost certainly be more effective than the export restrictions approved in October 2022.

In May 2024, just a few months before the November election, President Joe Biden sharply raised tariffs on Chinese imports, ranging from electric vehicles to solar cells, nominally at least a measure meant to buy time for American companies to catch up with their Chinese rivals in new energy technologies. As well as a tariff increase from 25 to 100 per cent on electric vehicles, duties rose from 7.5 to 25 per cent on lithium batteries, from 0 to 25 per cent on critical minerals, from 25 to 50 per cent on solar cells and from 25 to 50 per cent on semiconductors. Whether the tariffs will help spur an American energy renaissance remains very doubtful. The historical record shows that ‘industry protection without technology transfer is a recipe for bloated and lazy domestic firms making unaffordable, unattractive green goods’.⁶⁶ In the past, Japan and South Korea faced pressure to eventually remove their protectionist walls and let their car-makers be exposed to global competition, but the latest Biden tariffs have no sunsetting timeline, so where will the pressure to innovate come from? ‘The sizeable literature in economics and political science also points out a major medium term cost of imposing tariffs: unproductive businesses become more interested in lobbying for more protection instead of innovation.’⁶⁷ And innovation in this case extends to a whole ecosystem and even a whole energy paradigm.

The tariffs were imposed because the United States had lost the first round in the green energy wars, but China is under pressure as well: if geopolitical tensions continue to increase, Chinese electric vehicle companies may be subject to increasing tariffs or import restrictions in Europe, limiting their plans for global leadership. The American market was always going to be a long shot, but if the door to Europe is shut, ‘it will be a severe blow to the globalisation process of Chinese cars’.⁶⁸ Whether that happens depends on an exceedingly complex game of global political and economic influence of which we have so far seen only the initial moves. The tariffs approved by the European Union against Chinese electric vehicles in June 2024 were different in nature from the Biden tariffs: meant more to slow rather than stop imports from China, they ‘might also incentivise Chinese companies to invest or form joint ventures with European companies’.⁶⁹ The ‘big idea’, as a Politico story put it, might be ‘to use the tariff threat to force Chinese carmakers to come to Europe to form joint ventures and share technology with their European counterparts’.⁷⁰ Turkey announced in July 2024 that it was withdrawing from plans announced in June to impose an additional 40 per cent tariff on all vehicles from China after reaching an agreement with BYD, the Chinese electric vehicle manufacturer, to construct a $1 billion plant in the west of the country.⁷¹

Just as a country will climb the international economic hierarchy by transforming the framework conditions that surround economic life, so can other countries slow down its progress by corroding or depressing those conditions for their rivals, an exercise that is conducted at a deeper level than individual technologies and can be compared to world building. Whoever builds the world of tomorrow will leave other countries with no choice but to find their place within that world. But Washington seems to be making the fundamental error of assuming that a global technological system must take a form similar to the existing one, and thus that the way to stop a rival is to block the known path of technological development. In reality, there are many paths, and thus the attempt to block them, even before all the alternatives become visible, may turn out to be impossible. As Paul Triolo puts it, ‘there is no single path to achieving technological performance levels, only many different, albeit difficult paths. Chinese firms will find the paths that work well enough to continue to drive innovation.’⁷² History is fatalistic only from the point of view of the observer who has already reached the destination, not from the point of view of those who are beginning their journey.

There is a fundamental difference in nature between the normal operations of the world game, where the game programmer may use sanctions, the disconnection of network nodes and other measures to exert power or punish transgressors, and the battle royale for the role of programmer. Here it is not sufficient to rely on sanctions. The goal is to build a complete system under which every state must act and according to which specific outcomes are determined or produced. Thus, sanctions may well be effective at the lower level but ineffective at the level of world building. Ultimately, what a country should do is focus on maintaining its technological edge instead of trying to slow down its rivals. What is scarier for Chinese manufacturers than tariffs? Getting technologically leapfrogged.

Somewhat surprisingly, if decoupling is about building capacity, it would make sense for the United States to decouple from the global system in areas where China has become too dominant – that is, to acquire distinctive power over the way the system works while preventing China from doing the same. Revealingly, Robert Lighthizer, the United States Trade Representative in the Trump administration from 2017 to 2021, talks in his book of decoupling from China rather than, say, decoupling China: ‘The most urgent priority should be strategic decoupling from China.’⁷³ It seems to be just that form of decoupling that China adopted with the Great Firewall, the set of laws and tools used to block access to selected foreign websites and foreign internet platforms from within China. Chinese authorities felt the need to decouple from the global internet in order to preserve their ability to control and shape information as well as current and future internet technology.

What the term ‘decoupling’ expresses is a kind of coalescence of the global superpowers out of a system of shared rules. Different poles of will are formed and demarcated. Superpowers exist outside the global system as its makers or legislators, but the system survives as the product of their efforts, which are distinctively competitive. Decoupling means sovereignty. Before any kind of individuation happened in product or financial markets, there was the individuation of distinctive systems or projects to organise global exchanges, where the task was seen as not being subject to pre-existing rules. If economies and states are to compete to shape the world game, they must first decouple in the sense of acquiring their own autonomous spheres of action. The concept of world building captures this reality by stressing the existence of sovereign actors no longer constrained by a system of shared rules: architects of the world order that are nevertheless profoundly global in their outlook.