While the world is momentarily concerned with how to best regulate or govern Artificial Intelligence (AI), an important geopolitical contest is emerging over access to and use of AI and AI-related technologies. The implications of this conflict are vast, driving a global redistribution of power to those who can successfully develop and use AI systems from those who cannot.

Democratic actors have mobilized to develop a shared understanding for the future of AI based on western values through the creation of global AI initiatives that promote the responsible use of AI. Simultaneously, new regulations are being developed to prevent the use of AI-systems in ways that could undermine free, democratic societies. But these regulations are undermined by states such as Russia and China, which are making significant strides in building an alternative, authoritarian future for AI that is already being exported and adopted abroad.

Faced with rising digital authoritarianism, additional action is needed to limit and control access to the world’s most advanced AI technologies. While initial steps toward developing new global regulatory regimes are important, regulation on its own is not enough. Democracies must now focus on winning the race for AI supremacy, which will be determined by technological dominance and digital innovation, ownership of AI infrastructure, and strategic integration of private sector industry.

Technological Dominance and Digital Innovation

Technological domination has always been a key component of geopolitical power – AI is no exception to this rule. Those who build and use the best AI systems will maintain a geopolitical advantage. Most of the innovation and development in the AI ecosystem will be driven by private sector companies, competing against their economic rivals for dominant positions in global markets.

While significant research is being conducted into how to build better AI systems, AI is already impacting our world in potentially undesirable ways. These risks are well-known – from bias and discrimination in our everyday lives to assisting in the creation of bioweapons and cyber threats.

In democratic countries, these risks may be even more profound. At the same time, concerns around privacy and harm may put western companies at a disadvantage in terms of the research and activities they can carry out.

Authoritarian regimes are not stymied by such concerns. Instead, these actors enjoy a competitive advantage. This advantage is driven by their ability to gather data and deploy AI systems in a way that democratic states will never be able to do, especially when it comes to obtaining and utilizing sensitive personal or biometric data. When combined with strategic initiatives and funding to drive the creation of more authoritarian AI systems, this advantage could challenge the West’s current dominance in the AI industry.

Ownership of AI’s infrastructure

AI systems are often discussed in the context of models, algorithms, or data, rather than cables, concrete, and steel. Yet, AI is, at its core, dependent on physical infrastructure and hardware.

AI systems are trained on data sent across the globe, traveling the world’s interconnected, complex tangle of under-sea cables. The systems that make use of this data are powered by servers and graphics processing units (GPUs), based primarily in large data centers. In the case of training and developing new foundation models (FMs), the cost – both financially and computationally – is only accessible to well-funded and technologically competent organizations.

This infrastructure is a physical, tangible, asset – control over which is essential for achieving AI dominance.

The United States currently has the market advantage. American private sector companies drive most of the world’s leading AI innovations and own the most advanced infrastructure. It is American companies that own most of the world’s deep sea cables, as well as data and cloud computing centers. While countries like Taiwan are currently responsible for manufacturing a majority of the world’s state-of-the-art computer chips, these manufacturing processes are dependent on specialized equipment owned by and developed within the United States.

China is pushing back against U.S. dominance in AI infrastructure by growing its control over the internet’s subsea cable infrastructure. As part of their “digital Silk Road” project, Chinese telecommunication providers are investing rapidly into the development of new data centers, especially in areas where western providers have a limited presence. China is also developing into a world leader in the production and manufacturing of chips, prompting the Biden administration in 2022 to initiate new rules and regulations that will limit China’s ability to access cutting-edge hardware from the United States.

This policy shift led to Chinese companies, such as Bytedance, to purchase $1 billion worth of hardware necessary for their AI ambitions. An investment of this size is beyond the reach of most companies and states, further demonstrating the importance of the private sector in the brewing geopolitical conflict over AI.

Strategic Integration of Private Sector Industry

Large private sector corporations are, and will continue to be, an essential part of AI geopolitics. Private sector companies own the majority of AI infrastructure and are responsible for driving AI innovation.

Due to the centrality of these companies in the world’s emerging AI supply networks, they are well positioned to set the rules of the game: controlling which states or companies can use their infrastructure to develop, train, or deploy AI.

This new type of power is described by Farrell’s and Newman’s 2019 article “Weaponized Interdependence: How Global Economic Networks Shape State Coercion,” where they argue that states are able to “weaponize networks to gather information or choke off economic and information flows, discover and exploit vulnerabilities, compel policy change, and deter unwanted actions.” The private sector is a critical component of these networks.

To win the AI race, governments will need to support the private sector and include them as a critical component of a broader geopolitical strategy.

* * *

AI developments have led to a transformation in our understanding of power. To remain competitive, investments must be made to support the creation of a strong domestic AI industry that is able to develop and maintain state-of-the-art AI-systems and models. This is an expensive endeavor that is beyond the reach of many countries, especially those in the developing world. In addition to creating new power inequalities, the rise of the domestic AI industry will lead many states to become increasingly dependent on the private sector for their AI and data needs.

Acknowledging this risk, international actors like the European Union (EU) are pursuing a policy of digital sovereignty to give the EU and its Member States more control over their digital assets and digital supply chains. These calls for digital sovereignty have led to the proposal and development of the GAIA-x initiative, which aims to provide a secure data infrastructure for Europe free from the influence of large private sector actors. Whether or not the project is successful remains to be seen, but it has experienced significant setbacks and runs the risk of being overtaken by private sector interests in any case.

In the geopolitics of AI, the private-public sector dynamic will be essential to monitor. Important questions on the power relations between technology companies and states are emerging, with some experts already predicting the eventual demise of the state at the hands of the tech industry. Countries that are unable to develop their own AI industry, due to prohibitively high costs or lack of access to resources, will become the new battlegrounds for the future of AI as external companies fight to grow their market share. The results of these clashes will strongly influence the future development of AI power in ways that are still impossible to predict.

IMAGE: Visual representation of artificial intelligence. (via Getty Images)