Geopolitics in the age of AI
The enormous potential in terms of applications in almost all the possible areas of the human activity is making AI one of the new "general-purpose technologies", and one of the industries which both private actors and, even more, governments, consider as a strategic priority.
AI is indeed a perfect example of the so called "dual-use technology", that is of a multi-purpose technology which, once developed, can be used both for civilian and military (included security) purposes. As the current conflict opposing Russia to Ukraine is showing, AI is deeply changing both the strategies and the tactics of modern warfare. AI provides a ready-to-use massive amount of information which allow the troops on the ground to target more easily the enemy – a sort of "Uber for the artillery", as one official effectively put it.
Not surprisingly, AI has become a key component of great power competition, and a major strategy for enhancing national competitiveness and security. According to experts in the field, at the moment China is one year behind the US in terms of AI capabilities, but the US may soon lose its leading position. In order to speed up its AI processing capabilities, for instance, China launched in 2017 the New Generation AI Development Plan for 2030. The document is clear about China's intentions to become an AI innovation center by the beginning of the next decade, being the new technology essential for strengthening the country's military and economic capabilities. These goals will be achieved by means of a close coordination between the private sector and the government, directly or by means of state-owned "champions" in the field. In the meantime, both the EU and the US have, almost simultaneously, issued "chips acts", which aim at both re-shoring chips production and improving their domestic technological capabilities in the field.
In the jargon of contemporary geopolitics, AI is considered one of the most relevant "realms of power", a modern version of the "pivot", the control over which assigns a de facto position of enduring dominance. Not surprisingly, in a recently published National Security Strategy document, the US administration openly mentions AI as one of the fields in which it is mandatory to ensure that "strategic competitors cannot exploit foundational American and allied technologies".
In order to function, AI applications ultimately need machines, computers able to store and process an enormous (and growing) amount of data. Much (if not all) depends on the basic unit which fuels the machines' computational power, i.e. chips which have to be specifically designed and manufactured to be used in AI data processing. AI demand for customized semiconductors offers immense opportunities to private companies capable of producing chips at cutting-edge technological level in terms of transistor density, but also customized in order to perform the specific tasks required by AI systems.
So far, the global value chain governing chips production has been an example of borderless efficiency. Integrated circuits designed in the US and Europe are produced in South-East Asia employing US-made machine tools and sophisticated equipment. Chips for AI machines are not an exception; their technological sophistication meant that a significant portion of their production is located where there are the most sophisticated "foundries", capable of manufacturing chips with a density measured in nanometers, typically in Taiwan and South Korea.
The present concentration of technological capabilities in leading-edge AI chips in US and "US-friendly" countries gives the opportunity for the incumbent power (America) to weaponize semiconductor technology in order to keep the emergent challenger (China) at bay, through a mix of strategies ranging from export restrictions, to the so-called "foreign direct product rule" (which de facto prohibits the export of products incorporating US design or technologies), including moral suasion in order to avoid the acquisition of semiconductor companies by Chinese investors.