The Hard Part Won’t Be Exporting US AI—It’ll Be Making It Stick

Until recently, the United States was not seriously courting AI partners in the Global South. It had been warning them against partnering with China, but had not been offering a compelling alternative to be their economic partner of choice to deploy AI. With President Trump’s AI Action Plan, that posture has changed. The United States plans to win the AI race by “exporting its full AI technology stack—hardware, models, software, applications, and standards—to all countries willing to join America’s AI alliance.” To succeed, it will need to pursue the right partners, make offers that meet their ambitions, and resist the urge to lead with virtue over value.

First, the United States will have to choose its partners carefully because it does not have the resources to create favorable, low-risk conditions for its AI firms everywhere, the way China’s state-backed model does. Every international AI engagement requires financial backing because, while private companies may provide the data centers, fiber networks, and advanced AI tools, they rely on conditions that make massive, long-term, and often risky infrastructure projects commercially viable.

In many countries in the Global South where the United States wants to engage, those conditions don’t exist. China solves for that: Its policy banks offer large, concessional loans to foreign governments to fund entire infrastructure projects, which are then contracted to Chinese firms. That setup not only helps the recipient country, it de-risks the engagement for Chinese firms and makes otherwise risky overseas ventures commercially attractive and scalable.

The U.S. approach is different. Tools such as the Export-Import Bank (EXIM) and the International Development Finance Corporation (DFC) don’t primarily fund comprehensive projects directly. Instead, they offer loan guarantees, insurance, or co-financing to encourage private capital to step in, but the commercial risk still largely sits with the firm. That makes engagement in markets with weak institutions, uncertain returns, or high political risk fundamentally more commercially risky for U.S. companies than their Chinese counterparts.

If the United States tries to match China in creating favorable, de-risked conditions for its AI companies across too many markets, it will spread itself too thin. It should therefore focus its financial support on markets where targeted de-risking can make a real difference—where U.S. companies have something distinctive to offer, and where the strategic returns for the United States are most meaningful.

It can hone in on such countries by identifying those that are already leaning into AI’s promise, and that have strong digital foundations but lack the capital, compute power, or specialized infrastructure to build full, independent AI ecosystems. For example, both Kenya and Nigeria boast burgeoning tech sectors, a growing appetite for AI solutions, and the potential to anchor broader regional ecosystems. What they truly need is a partner capable of helping them scale their vision, and that’s precisely where the United States can take the lead.

Second, the United States will need to offer tailored partnerships that help countries use American AI to advance their specific national priorities and development goals. The AI Action Plan talks about offering “full-stack AI export packages,” a prepackaged bundle of U.S. hardware, software, and models that it can deliver to willing partners. That may sound like a comprehensive offer, but on its own, it risks treating countries as clients rather than economic partners—recipients of American technology rather than co-builders of something lasting.

It also risks sending the signal that the United States values them mainly as pieces in a strategic rivalry with China, rather than as valuable partners with priorities of their own. That model simply won’t work in today’s global landscape, where countries in the Global South are keenly aware of their leverage and are looking for partners who will support development aligned with their national interests, not simply advance what they see as foreign strategic or corporate agendas.

What many countries truly want isn’t just access to advanced AI tools, but a transformative partnership that empowers them to build indigenous capacity, apply AI to their unique challenges, and drive their own long-term economic growth.

China exemplifies this approach: In Kenya’s Konza Technopolis, a flagship government-led smart city project, China financed the initiative through loans, then Huawei stepped in to not only build a national data center and smart ICT network, but also to deliver public safety and traffic management systems. It has invested in training local workers to run and maintain the infrastructure, and has launched extensive programs, including AI literacy hackathons, to build a local workforce capable of developing new AI-powered services and solutions on top of its systems in sectors like agriculture and health care.

If the United States hopes to compete with China in such regions, it will need to offer more than standardized, one-time AI packages. It needs to compete on outcomes and show what American AI can help countries achieve. That means aligning its offering to work with partners to apply AI in ways that unlock local economic value and deliver tangible progress in areas like agriculture, education, and health care.

Third, and equally critical for long-term commitment, the United States will need to address the fear that potential partners have of being cut off from essential technology. Many governments worry that if they rely too heavily on U.S. AI systems, particularly for critical public infrastructure and essential services, access could disappear overnight due to political disputes or shifts in U.S. foreign policy.

To overcome this apprehension and build trust, the United States should establish clear, transparent, and binding agreements that guarantee consistent access to U.S. AI technology and services, even amidst geopolitical shifts. Providing ironclad assurances is vital for securing long-term commitment from partners and demonstrating the reliability of the U.S. technology ecosystem.

Finally, the United States needs to resist any urge to turn its AI efforts into a values contest, focusing instead on pragmatic economic partnership. In recent years, the United States has relied on getting countries to align with U.S. AI priorities by drawing on the residual goodwill of traditional aid partnerships and casting China as the authoritarian boogeyman.

The pitch was clear: “Partner with us because we have democratic values.” However, the sudden changes to aid structures and perceived inconsistencies in U.S. foreign policy have effectively nullified that “ethical AI partner” pitch, and the goodwill has evaporated. China, meanwhile, is still around, actively financing AI projects and, on its face, helping countries move forward on their own terms.

If the United States wants to be taken seriously as a contender, it cannot afford to go back to trying to sell itself as the principled alternative to China. Instead, it should start acting like a credible economic partner and center its counteroffer on economic competition and shared prosperity, not a values crusade.

The Trump administration is right that for the United States to win the AI race, it will have to get other countries to build with it, bet on it, and lock in long-term alignment around its technology stack. But that’ll take more than sermonizing about values or selling its stack. It will require the United States to prove it’s a partner worth building a long-term AI future with.

Social Media Preview Credits: Generated by DALL-E