The internet is splintering along ideological lines, with many governments – in particular the US, China, EU, India and Russia – competing to reassert the authority of nation-states.
Fragmentation is now moving down to the internet’s fundamental, hidden infrastructures: technical standards, supply chains and submarine data cables.
In contrast to regulatory divergence, which can be addressed in the future, these structural battles risk locking in friction and fragility at great cost. The greater the intervention lower down the internet stack, the higher the risk that fragmentation may be locked into the internet’s architecture for good.
One benefit of the global, interoperable internet is that it avoids much of the traditional friction that comes from physical borders. Communication and transactions are near-instant, breaking many geographical constraints. But as states try to reimpose their authority by regulating both the good and bad of the internet’s disruptive potential, this global infrastructure may splinter into many sovereign internets. This phenomenon risks undermining the principles on which decades of economic and social opportunity have been based.
Shedding Light on the Splinternet
While the regulatory component of this so-called splinternet has been covered at length elsewhere, states are increasingly intervening deeper down the internet stack to assert control on the fabric of the internet itself. The battles at this hidden frontier are part of a new and more worrying trend. While regulatory divergence creates challenges of its own, regulations are malleable and can be revisited and aligned in future. In contrast, structural changes to the standards, supply chains and submarine cables that underpin the internet cannot be reversed without significant upheaval and economic cost. The greater the intervention lower down the internet stack, the higher the risk that fragmentation may be locked into the internet’s architecture for good.
Internet fragmentation has been the subject of many academic analyses. In 2018, Hall and O’Hara set out the battle between four competing ideological models:
The technology-centred Silicon Valley model that focuses on maintaining an open internet to reflect its original, idealistic vision.
The rights-based European model that seeks to minimise uncivil behaviour such as harmful content, anti-competitive practices and privacy intrusions, even if it could come at the cost of innovation.
The surveillance-driven authoritarian model, led by China but becoming increasingly attractive to other nations seeking greater controls through enhanced surveillance and identification powers in the name of “social cohesion”.
The DC Commercial Internet model favoured by former President Donald Trump that advocates for the interests of private actors over the public good of an interoperable system.
However, as rapidly as the internet ecosystem proliferates, these four visions have given way to multiple models of a sovereign internet – each driven by differing priorities but united in their aspiration for control:
The internet risks fragmenting due to an ideological battle of sovereign internets.
Laissez-faire/lack of friction
Low barriers to entry help innovation
Superstar companies dominate
Strong public-values regulation
De facto global regulator
Bureaucracy stifles innovation, creating few major players and limiting geopolitical leverage
Huge market mitigates regulatory constraints
State champions dominate; civic restrictions
Tight control of infrastructure; limited but growing regulations for services
Huge market creates geopolitical counterweight
State champions dominate; civic restrictions
Successful tests of a “national internet” enable censorship at scale
Weak internet economies; civic restrictions
These national internet models provide a snapshot of the overt regulatory approaches to influencing the development of the internet, but underneath them lies a web of technical standards and hardware that is becoming a new, hidden frontier of geopolitical competition. When examined individually, the motives of states could be viewed as benign or even beneficial. However, when aggregated, the potential effect of these attempts to assert control over fundamental structures of the internet may lock in global fragmentation for good.
Fragmentation Through Competing Standards Proposals
As set out earlier, the layers of the internet stack are largely independent and unaware of each other. A new proposal from China called New IP, in practice submitted by Huawei and other industry collaborators, would involve a shift away from this model. It seeks to reconceptualise the network layer of the internet (the dumb pipes), which ordinarily transports data packets without being aware of content. There are two main consequences of this:
Political control of the network: Data packets contain both content and descriptions of where they are being transmitted. By enabling surveillance techniques such as deep-packet inspection, New IP could break the independence of this network layer – so the dumb pipes are no longer dumb – and enable political interference.
Traceability: Descriptions of New IP contain references to indelible identifiers and blockchain architectures that cannot be edited and would enable complete traceability of content and networks. This contrasts with today’s system of temporary IP addresses, which provide devices with some level of identity protection. In essence, as internet governance expert Carolina Caeiro argued at the 2021 RightsCon Summit, this proposal would turn the network into a surveillance tool.
The proposals are sold on the idea that the internet’s current architecture cannot support today’s accelerating growth in traffic, so a new model is needed. The moves towards greater digital sovereignty are also a response to the dominance of foreign infrastructure providers such as NEC, Fujitsu, Ericsson, Siemens, Alcatel and others in China’s telecoms markets in the 1980s and 1990s. There are many more secure, privacy-protecting and technically neutral solutions to the New IP model and its lack of any clear provisions to protect privacy and other rights. There is also growing concern that New IP could lead to the type of capture, transfer and use of personal data that violates states’ obligations under international human-rights law.
Telecoms & technology stack
TCP/IP model (Current)
China’s New IP proposal
3GPP, IEEE, ETSI, ITU
3GPP, ITU, ETSI, GSMA
The New IP proposal also threatens the future of multi-stakeholder governance institutions that traditionally oversee the design, development and maintenance of internet standards and protocols. Instead, China proposes that the International Telecommunication Union (ITU), a UN body in which only nation-states can vote, should be empowered at the expense of bodies like the Internet Engineering Task Force (IETF). This means bypassing the multi-stakeholder standards organisations that incorporate voices from industry, civil society and governments, and which largely support a free and open internet. In turn, China’s proposal provides the opportunity for national governments, which support more tightly censored and regulated models of the internet, to have greater power in shaping its future.
Currently, China’s New IP proposals have been rejected by the IETF and other multi-stakeholder bodies but have gained some traction at the ITU. The proposals may return in different forms if they are not immediately successful. However, while international standards can facilitate adoption of technologies, they are not completely necessary. China is already piloting New IP domestically and may look to export it to emerging digital economies that are dependent upon Chinese investment and infrastructure in the absence of effective alternatives.
Fragmentation Through Semiconductor Supply-Chain Competition
Semiconductors have become a critical, foundational technology and strategic industry in the 21st century. Supply chains are highly globalised: chip architectures are routinely designed in the US or UK but fabricated in Taiwan or South Korea. However, a recent US White House supply-chain review expressed the need for more self-reliance in chip production. This is catalysed by national security concerns, given China’s ongoing territorial claims over Taiwan, which could threaten the world’s leading chip manufacturer, TSMC.
The frontier for commercial production, at present, is developing process nodes at 3 nanometres (nm). While US firm Intel has plans to offer this capability, only TSMC and Samsung (South Korea) are in advanced stages of doing so. Beyond the US, South Korea and Taiwan, no other country, including China, boasts a semiconductor manufacturer that is near this milestone.
Sources: Eurasia Group and TBI
Example products using chips built at each process node
90nm - Playstation 2 65nm - Microsoft Xbox 360 45/40nm - Playstation 3 32/28nm - Intel Core i3 and i5 processors 22/20nm - iPhone 6 (A8 chip, 20 nm) 16/14nm - iPhone 6S, iPad Pro (A9 chip, 14nm; A9X & A10 chips, 16 nm) 10/7nm - 2nd-gen iPad Pro (A10X chip 10nm), iPhone XS (A12 chip, 7nm), iPhone 11 (A13 Bionic chip) 5nm - iPhone 12 (A14 chip); M1 Macbooks
China’s SMIC is lagging – it is currently only planning 7nm chips, which TSMC and Samsung are already producing at scale – while the US has also pressured TSMC not to sell to Huawei, one of China’s largest technology vendors. As a result, as global demand for semiconductors has increased, the fixed supply of expensive chip foundries combined with US sanctions has created a shortage. One by-product of this is a new industrial hierarchy, with US tech giants the winners and carmakers, many of which are based in the EU, the losers. Similarly, ASML, which supplies chip-manufacturing machines to the semiconductor industry and is the EU’s biggest tech company by market capitalisation, has also been affected by US export controls on China.
While building new chip foundries in the EU and US may be rational given today’s fractious geopolitical environment, the trade-off is likely added cost and complexity for global supply chains. As states increase their self-reliance when it comes to chip fabrication, there is a risk that device manufacturers may be forced to use specific manufacturers or foundries. While this is already happening to an extent given TSMC’s dominance of the sector, partnerships based on political allegiances are worse for consumers because they reduce competition and create a greater risk of lock-in than those based purely on technical capability.
Fragmentation Through Proprietary Submarine Cables
Underpinning the global internet is a network of surprisingly vulnerable undersea cables. This backbone accommodates approximately 97 per cent of internet and voice data and facilitates around $10 trillion of financial transactions daily. There are four main suppliers: Alcatel Submarine Networks (France), SubCom (US), NEC (Japan) and Huawei Marine Networks (China). Increasingly, new players such as Google, Facebook and Microsoft are also investing directly in cable infrastructure to accommodate the massive web traffic they generate.
While submarine cables are not free of regulation, since the internet’s inception there has been a consensus that the global free flow of data is a common good and, if regional variation is necessary or desirable, this should happen much further up the stack (for example at the content-regulation level). However, US pressure on Europe over China’s “Peace Cable”, a new cable travelling over land from China to Pakistan and then under the sea to reach France, cuts against this consensus. Huawei, which now controls almost 10 per cent of the subterranean market, is the third largest shareholder in the company building the new cable.
There are a number of physical and digital risks to submarine cables, and while most incidents are related to accidental activity there is legitimate concern of potential deliberate sabotage. Fears of surveillance and data theft through the creation of backdoors during cable construction, tapping landing stations or tapping cables at sea, as well of cyberattacks, are also beginning to dominate the policies and politics of subterranean cables. As geopolitical debates increasingly co-opt this foundational infrastructure, a variety of responses are possible: states may keep some traffic away from vulnerable cables, or trends towards digital sovereignty may expand to this layer, creating a race for proprietary submarine cables.
A network of sovereign cables could lead to huge inefficiencies in the global internet network, which relies on transporting information as fast as possible. It could also ultimately leave states more vulnerable if forced to rely on a smaller number of cables, instead of an interconnected global network with built-in redundancies. This is not an immediate risk, but leaders must be aware that subterranean geopolitical competition could splinter the internet at the very deepest level.