The tariff regime instituted by the Trump administration will prompt businesses, governments, and investors across frontier AI hardware and software supply chains to carefully reconsider their strategies and which entities they choose to partner with. The new environment is creating openings for a more globally dispersed set of actors to be involved in AI supply chains and creating new incentives to develop critical operations, talent pools, and intellectual property assets outside of the U.S. and China.
In the wake of the tariff announcement, investing in and trading with both the U.S. and China has become more risky. Additionally, the value proposition for investing and operating in the U.S. is quickly changing to one premised more on short-term factors rather than long-term stability. To make the most of these new conditions and to secure its lead in AI capabilities, the U.S. government would need to double-down on incentivizing and investing in domestic AI hardware and software research and development - but it is not yet clear to what extent the U.S. government will do this.
As such, the current state of flux presents remarkable opportunities for governments and businesses worldwide to become more involved in frontier AI hardware and software supply chains. Successfully taking advantage of these opportunities will not be easy, however. It will require long-term planning and committed investment over multiple years, all within a macro environment that includes heightened risks of global recession and international conflict.
It is now more risky to invest and trade with the U.S. or China: Both the U.S. and China are becoming more risky for outsiders to invest in and trade with. This risk exists in an immediate sense as a result of the incredibly high two-way tariffs between the U.S. and China, as well as in a larger and more ongoing sense due to how economic and military tensions between the U.S. and China have been climbing over the past eight years and now seem to be approaching some sort of local maximum. Neither de-escalation nor guarantees about more stable tariff levels seem imminent. Engagement with the U.S. promises considerable uncertainty, while engagement with China promises potential exclusion or reprisal by the U.S. and its partners.
The value proposition for investing and operating in the U.S. is changing to one premised more on short-term factors than on guaranteed stability: Predictable and stable national governance was an enormous advantage for the United States. It is a key reason why the U.S. became a global magnet for immigration, investment, and trade over the last few decades. This historical continuity of federal policy is being replaced with fast-moving, disruptive, and highly-centralized decision-making. It is tough for those outside of the Trump administration’s inner circle to anticipate policy directions. The consequences of this are not necessarily all negative. This type of system could hold advantages for the U.S. government’s ability to respond swiftly in times of global instability and crisis.
However, the case for investing in the U.S. now is a fundamentally different value proposition that hinges on relationships and faith in the current administration rather than on the broader institutional environment and reputation of the U.S. Over time, continued changes that are as drastic as these tariff rules could also alter global perceptions of transfers of power between Presidential administrations. These could become events that increase anxiety about policy instability and unpredictability due to abrupt and large-scale policy changes. In many ways, the value proposition for investing and operating in the U.S. is taking on some similar characteristics to the value proposition of investing and operating in an emerging market.
To make the most of the new trade regime, the U.S. government needs to double down on incentivizing and investing in domestic AI hardware and software research and development: The U.S. does not currently have manufacturing capacity, talent, and dedicated capital to be able to domestically replace China’s role in the hardware supply chain. To be able to eventually take on and lead in this capacity, the Trump administration and Congress would need to pass new government-backed incentive and investment programs that significantly augment or replace the CHIPS program. There have been some moves in this direction - such as the establishment of a U.S. Investment Accelerator and the announcement of TSMC's $100 billion investment in new chip fabrication and R&D facilities in the U.S. - but it is still highly uncertain if funding and commitments will happen at the scale necessary to revitalize the U.S. chip industry.
Realistically - because developing domestic R&D and manufacturing capacity of this scale takes a significant amount of time - such initiatives would work best if they are kicked-off in tandem with carefully structured long-term partnerships on hardware trade, intellectual property, and talent with countries such as the Netherlands, South Korea, and Japan and the leading hardware companies in those countries. Without new funding and incentives to build local hardware R&D and manufacturing talent and capacity - and create special partnerships with countries with core competencies in those areas - America’s approach to AI supply chain strategy will weaken.
Given these conditions, new opportunities exist for governments and businesses seeking to increase their foothold in state-of-the-art AI research and development. A successful AI strategy will involve securing stable access to key resources for developing frontier AI systems - and then leveraging those resources to nimbly navigate an unpredictable future.
Factors that governments and businesses should consider to find success with their AI strategies include securing access to critical intellectual property and R&D talent, securing access to large amounts of capital that can back long-term plans, and prioritizing building and maintaining “sovereign” AI capabilities. Additionally, governments and companies would benefit from considering how to position themselves to stay at the frontier through the duration of a large-scale conflict and a global recession.
Securing access to scarce pools of critical intellectual property and R&D talent: Certain pools of R&D talent and intellectual property assets across the hardware and software stacks for AI development are highly localized and scarce. Now - while trade relationships are still in flux - is a critical time for governments and companies to ensure they maintain long-term access to these pools of talent and IP. Examples of such talent & IP pools include semiconductor manufacturing & manufacturing equipment research and development, high-performance computing, uniquely high-quality training datasets, or large language model architecture development.
As new trade barriers and relationships solidify, accessing these scarce pools will likely become even more difficult than is already the case. Securing access to these talent and IP pools could be done via negotiated long-term commitments with the governments and companies that currently house these people and assets. If governments or companies do not possess comparable assets to offer access to in return, they could proffer benefits in entirely different sectors where they have a global comparative advantage (perhaps defense domains or unique natural resources). Governments could also attempt to move these scarce pools by creating uniquely attractive immigration, tax, and regulatory policies that attract this talent and the owners of these assets to relocate their activities.
Securing access to capital: Building and using large amounts of computational infrastructure to train and deploy the next generations of AI models will be very expensive. As such, governments and businesses with access to large amounts of capital will have unique leverage over being able to steer new directions for AI research and development. Some existing business and government entities are or could plausibly provide access to these large amounts of capital. Private sector examples include Softbank and Berkshire Hathaway. Government-backed entities include Norway’s Government Pension Fund, Singapore’s Government of Singapore Investment Corporation (GIC) and Temasek Holdings, the United Arab Emirates’ Mudabala Investment Company and MGX, and Saudi Arabia’s Public Investment Fund.
As access to capital becomes more sought after, wealthy individuals and large institutional endowments could also play a critical role in enabling the buildout of new AI infrastructure and R&D programs. Given the amount of leverage that capital providers can hold over the direction of these initiatives, governments or businesses should carefully evaluate which private sector or governmental funding partners are most aligned on basic governance values and long-term strategic objectives for how new hardware, infrastructure, and AI systems are built, tested, and deployed.
Building and maintaining sovereign AI capabilities: Governments and businesses that secure access to critical IP, R&D talent, and capital will be able to build and maintain “sovereign” AI capabilities that provide lasting advantages in a more fragmented and less stable geopolitical environment. Here, “sovereign” means locally owned and operated infrastructure and capabilities that are reasonably near the generative AI model development frontier.
Some countries have already made sizable steps in this direction. France is an interesting example. Notably, France signed an agreement with the United Arab Emirates, for the Emirati investment entity MGX to provide €30 to €50 billion towards funding the development of domestic data center infrastructure - which neatly complements France’s existing comparative advantage in homegrown nuclear energy. French regulatory and tax rules are still a deterrent to attracting IP owners and R&D talent to the jurisdiction. However, Macron is explicit about his attempt to move the country in a direction that increases its independence from both the U.S. and China in AI.
Positioning for large-scale conflict: Several factors could contribute to increased chances of large-scale international conflict over the next decade. These factors include reduced international trade linkages (which reduce the additional economic cost of countries going to war with each other), reduced trust in the deterrent effect of international military alliances, a prioritization of building up military capabilities and expanding defence budgets across Europe and Asia, and multiple points of simmering tension throughout Europe and Asia (including in Eastern Europe, the Taiwan Strait, the South China Sea, and the Indo-Chinese border).
Maintaining access to IP, R&D talent, and capital during a large-scale international conflict could make a substantial difference for governments and businesses to either stay at or lose touch with the frontier of AI research and development. The trade war is likely just one instance of many to come over the next few years in which governments and businesses must choose partners at the risk of losing out on others. Staying “non-aligned” is becoming increasingly difficult.
Additionally, the already constant state of global cyber conflict will only increase in intensity alongside the tensions towards physical conflict. Those preemptively investing in research security across their AI hardware and software supply chains will have better chances of protecting their comparative advantages on an ongoing basis from theft or disruption by cyber-offensive actors.
Positioning for global recession: The current state of heightened uncertainty caused by the new trade regime is considerably increasing the risks of a global recession in the near future. Maintaining access to IP, R&D talent, and capital during a global recession will be very difficult and take unique risk-tolerance. There is a large chance that many leading model development companies re-allocate resources from research and development towards product commercialization and growth. A lack of visible markers of progress from these leading model developers could increase pressure on all actors throughout the ecosystem to perform the same re-allocation.
However, once the global economy calms, those who have stuck with long-term commitments to building up their hardware, software, infrastructure, talent, and governance capacity at the edge of AI development will emerge ahead of the pack. If such a global economic downturn is not navigated well, there is a risk that more authoritarian and centralized actors emerge ahead and that more democratic and market-driven ecosystems fall behind irrevocably.
We have entered a unique period of opportunity and risk in trade and international relations. Governments and businesses should take advantage of this moment to recalibrate their AI R&D and supply chain strategies. Those who can secure stable and long-term access to intellectual property, talent, and capital - and double-down on resolutely building out their sovereign AI capabilities amidst chances of economic downturns and large-scale conflict - will position themselves well for the future.
AI supply chain research is a priority at Augur. Subscribe to keep up with our future updates on these topics and drop us a note if you are interested in partnering on projects related to any of the above.
...to keep up with research updates and announcements