Last year, it was the year of generative artificial intelligence (AI) with the mass adoption of large language models (LLMs) such as OpenAI’s ChatGPT and image generators such Midjourney. Hiding beneath interfaces, AI models, and piles of data, this boom of capabilities and applications has been driven by hardware: AI chips. Albeit not very visible for users and consumers, the semiconductor value chain for dedicated AI chips plays a crucial role in the development and deployment of generative AI. It will influence issues such as whether China will invade Taiwan, if the US can maintain its technological edge and remain the global superpower, and whether Europe’s geopolitical vision of open strategic autonomy is viable. As such, the AI semiconductor sector becomes increasingly in the crosshairs of politicians and governments. In this piece, we will look at the geopolitical and economic dimensions of this ‘AI chips war’ that is at the heart of the AI race.
Developing and manufacturing dedicated chips to train AI models have been dubbed the ‘gold’ of AI. This is as far as the frontier of technological innovation goes, pushing both at the edge of current physics and computer science. Given that it is such a high-tech frontier of innovation, the actual ‘fabrication’ – that we mostly associate with technology supply and value chains – of AI chips is just a small part of the total value chain. The AI semiconductors value chain involves multiple stages:
Each of these stages is vital for the production of high-performance chips that meet the specific requirements of AI applications, such as the ability to process large datasets and perform complex computations efficiently.
However, this process is fraught with bottlenecks, significantly impacting the pace at which AI technologies can advance. One of the most pressing challenges is the industry's heavy dependency on specific countries for raw materials, particularly rare earth elements where China dominates supply and other specialized materials necessary for the latest generation of semiconductor manufacturing. These materials are crucial for the production of chips that are capable of handling the high-speed, high-efficiency demands of AI algorithms. Similarly, the Dutch company ASML is the leading company for lithography systems for semiconductor production, and thus a major market player in the value chain. The concentration of these resources in a few geographical locations introduces vulnerability into the global supply chain, subjecting it to geopolitical tensions, trade disputes, and regulatory changes. For instance, any disruption in the supply of these critical materials, whether due to political conflict or export restrictions, can have cascading effects on the availability and cost of semiconductors worldwide.
AI is arguably the next general purpose technology (GPT) and therefore of huge strategic importance for countries and geopolitics. Russian President Vladimir Putin has declared that the control of artificial intelligence will be crucial to global power, stating that “Artificial intelligence is the future, not only for Russia, but for all humankind. It comes with colossal opportunities, but also threats that are difficult to predict. Whoever becomes the leader in this sphere will become the ruler of the world.
In this spirit, access to advanced chips to train AI models is a – or the most – critical factor in the advancement of AI, with far-reaching implications for national security and technological sovereignty. By powering the hardware that drives data processing, analytics, and autonomous decision-making, a nation's ability to produce and procure these chips can significantly influence its position in the global hierarchy of technological capability and security. National security has become closely intertwined with AI semiconductor supply, given their application in defense systems, communication networks, and critical infrastructure. Furthermore, a disruption in the semiconductor supply can hamper a country's defense mechanisms, making it vulnerable to security threats. Technological sovereignty, the capacity of a country to independently fulfill its technological needs, also heavily depends on semiconductor access. It is a measure of a country's autonomy in tech development, allowing it to set its agenda for AI advancement without undue foreign reliance or influence.
Given the strategic importance of owning and having access to AI chips, the US and China are now rushing to maintain ownership and control over the technology. In August 2022, just before the mainstream attention went to (generative) AI, the US launched a huge investment package called the ‘CHIPS and Science Act’ to boost American domestic chip production and development, and last month more billions were announced for ‘advanced chips’ that are suited for training and developing AI models. These are good examples of how strategic the AI semiconductor industry has become. First of all, it represents a significant investment by the US government to revitalize and support domestic semiconductor manufacturing, research, and development – despite the higher costs associated with the later stages of the value chain being done in the US and thus a higher willingness for inflation. Furthermore, the measures are designed to reduce dependence on foreign chip production, which has become a concern for economic and national security amid supply chain vulnerabilities exposed by recent global events. By investing in its semiconductor infrastructure, the US aims to recapture a leadership position in chip manufacturing, which has been ceded to Asia over recent decades. This shift is not just about economic gains but also about having a fail-safe against the geopolitical uncertainties that threaten the uninterrupted supply of these vital components. Ensuring a robust and resilient semiconductor supply chain is now a strategic imperative, one that countries are addressing with a mixture of policy initiatives, subsidies, and partnerships with private industry. These efforts underscore the critical intersection of technology, trade, and economic policy in the global competition for semiconductor dominance.
Furthermore, the Biden Administration in October 2022 implemented significant restrictions on China, including cutting off China from certain semiconductor chips made with US tools to slow Beijing's technological advancements. This move aimed to limit China's access to advanced technology and prevent the flow of semiconductors to China for military purposes, such as developing nuclear weapons. These restrictions marked a substantial shift in US policy towards exporting technology to China and could significantly impact China's chip manufacturing industry by disrupting support from American and foreign companies using US technology. Similarly, ASML, the Dutch company that manufactures lithography systems for semiconductor production, has had its export license to China revoked by the Dutch government for certain lithography systems to develop the latest, most advanced AI chips. This decision was made in response to pressure from the United States, which has imposed new restrictions on exports of semiconductor chips and lithography machines to China. In response, China launched its own semiconductor investment fund in September 2023 of about $40 billion to boost its domestic semiconductor industry. All of this cannot be unseen from the rising interest and attention to (generative) AI that emerged last year.
Being a ‘laggard’ in the digital economy, the European Union's (EU) ‘European Chips Act’ from September 2023 to increase its semiconductor production capacity exemplifies a strategic response to these geopolitical challenges. Recognizing the risks of dependency on external sources, the EU has been investing in efforts to bolster its semiconductor industry. This includes funding for research and development, incentives for building fabrication plants, and policies aimed at attracting talent in the semiconductor field. The aim is to ensure a stable and secure supply of semiconductors to support its burgeoning AI sector and reduce reliance on non-EU countries. By doing so, the EU seeks to strengthen its technological sovereignty, secure its AI ecosystem from potential supply chain disruptions, and assert its autonomy in the global power dynamics of technology.
As such, the geopolitical landscape surrounding AI advancement and semiconductor technology is increasingly influenced by tensions and strategic moves by major powers, particularly between the US and China, with significant implications for global security, economic stability, and technological sovereignty. One clear example is that the struggle for dominance in AI technology, as highlighted by Putin's statement on AI's crucial role in global power, underlines the importance of advanced chips in shaping the future of international relations. The ability to develop, produce, and control these chips determines a country's military, economic, and technological strength. As AI chips become central to defense systems and critical infrastructure, nations prioritize securing their semiconductor supply chains to safeguard national security and maintain technological independence.
As such, all major powers are very hawkish on owning their own AI chips. For example, China's determination to achieve technological self-reliance and counter US restrictions, means that it is being put at odds with other major countries, such as Taiwan. Taiwan's critical role in the global semiconductor supply chain is one of the reasons for the rising China-Taiwan tensions. China's ambitions to assert control over Taiwan are intensified by the strategic value of Taiwan's semiconductor industry, as a conflict could enable China to dominate this critical industry, but also risks disrupting global supply chains, making Taiwan's status a flashpoint in US-China geopolitical rivalries. Similarly, companies could be used as geopolitical players, for example by using OpenAI’s – a major generative AI player –tools to catch spies and hackers, but also use the technology to spy on users and their data. Similar to the energy and food industry, domestic AI and chip technology will increasingly be considered as crucial geopolitical assets for countries.
Given these geopolitical considerations, the next piece will investigate the AI Stack that is crucial for understanding the complexities and geopolitical dynamics of AI technology. As such, we will inquire about the economic challenges and bottlenecks of the supply chain in meeting generative AI’s chip demand. From this, we will look at Big tech's control over AI Stack and their strategic positioning within that Stack by major companies like Nvidia, Amazon, Meta, Apple, and Google, to highlight their efforts and strategies to become the leading generative AI firm.
