America’s AI Export Strategy: Welcome Progress, but More Is Needed

In July, the U.S. government released its long-awaited national AI strategy. Alongside the main document, three executive orders were announced, including one focused squarely on promoting the export of the American AI technology stack. Former U.S. Chief Technology Officer Michael Kratsios underscored the ambition during his recent remarks at the APEC Digital and AI Ministers Meeting, noting that a Department of Commerce-led program would “carry out the development and deployment of the U.S. AI stack, packaged for each customer nation’s convenience.”

This is a significant development. For years, China has pursued a state-backed model of technology export, heavily subsidizing its national champions to capture global market share. Huawei’s expansion in telecommunications infrastructure across Africa, Europe, and Asia illustrates how subsidies and strategic financing entrench Chinese technology globally. In contrast, trusted telecommunications infrastructure vendors have struggled, not because their products lack quality or innovation, but because the U.S. has lacked sufficient financing tools and export promotion mechanisms. Our financing system for U.S. products and services has lagged behind the urgency of strategic competition.

Successive administrations have failed to reform U.S. financing instruments to meet the scale of the challenge. Reauthorizing and expanding the budget of the Development Finance Corporation (DFC) is an important first step, but it is only a first step. If Washington wants its AI export strategy to succeed, it must do more than package and market American technology abroad. It must expand the scope of the AI stack, adopt a more pragmatic risk tolerance, and address the underlying reasons why some foreign governments hesitate to deepen their reliance on U.S. technology providers.

Expanding the AI Stack to Reflect Strategic Realities

Kratsios correctly identified many of the core components of the U.S. AI stack: AI-optimized hardware such as chips and servers, data center storage, cloud services, data pipelines and labeling systems, AI models, and cybersecurity. But if the goal is to offer a competitive and truly comprehensive alternative to Chinese technology providers, the U.S. must broaden its vision.

A crucial gap is illustrated by telecommunications and networking infrastructure. The global AI economy rests on a foundation of data centers—and they rely not just on servers and chips, but also on ultra-high-capacity networking, including data center interconnects and subsea cable infrastructure that allow facilities to share workloads, replicate data, and scale efficiently. This is an area where U.S. firms, from optical networking providers to switch manufacturers, remain global leaders. Excluding this capability from the AI export package would be a mistake.

Without U.S.-backed offerings in interconnects and telecom systems, foreign governments may find themselves turning to Chinese suppliers for the “pipes” of their AI infrastructure. Once embedded, those vendors secure not just contracts but also strategic influence. Washington must ensure that the AI export program reflects a full digital stack, from hardware and cloud services down to the fiber-optic links that make large-scale AI possible.

A Smarter Risk Tolerance

The second area where the strategy must evolve is financing. The current U.S. approach remains overly risk averse. The Commerce Department and financing arms such as the DFC need to adopt a sharper lens. What is the strategic cost of inaction?

If a Chinese company wins a contract to build a data center or AI pipeline abroad, the consequences go far beyond one lost sale. The contract revenue in part flows back into Chinese R&D budgets, strengthening their global competitiveness. The data processed by Chinese-built systems can help train better algorithms. And critically, the presence of Chinese firms in a country’s digital backbone provides Beijing with ongoing leverage.

In this context, U.S. financing tools must accept a higher level of calculated risk. For a time-limited period, Washington should be willing to underwrite loans or guarantees that might not meet traditional commercial thresholds if the alternative is ceding a strategic foothold to China. Even a modest short-term loss to the U.S. taxpayer is far cheaper than the long-term costs of losing this race. Unlike grant programs, export loans are repaid over time, and unlike doing nothing, they ensure that the next generation of digital infrastructure is built on trusted foundations.

Addressing Partners’ Concerns Head-On

Finally, the U.S. must grapple with the reasons why some governments are hesitant to buy American. Too often, Washington assumes that if its technology is superior, partners will line up to procure it. In reality, purchasing decisions are deeply political.

First, allies are watching America’s broader commitments. When U.S. leaders question NATO, equivocate on Indo-Pacific security guarantees, or signal retrenchment, foreign governments naturally hedge. Building their digital systems on American technology feels riskier if they are uncertain about America’s long-term strategic reliability. Washington must make clear that those who adopt secure, U.S.-aligned systems are reinforcing—not undermining—their security umbrella. By contrast, governments that turn to untrusted vendors risk exposing not only their infrastructure, but also the effectiveness of U.S. and allied force projection in times of crisis.

Second, U.S. allies are increasingly assertive in regulating digital markets. For example, the European Union has imposed new taxes, regulations, and competition rules affecting major American tech firms. Washington does not need to agree with these policies, but it must recognize that domestic concerns in those markets are real. The U.S. should signal that it respects the sovereign right of other nations to regulate and tax—so long as these measures are not discriminatory. Dismissing these concerns risks alienating potential partners and pushing them toward Chinese alternatives.

If Washington does not treat the concerns of allies as legitimate, it will lose deals by default. If it instead engages respectfully, acknowledges differences, and makes the case for why secure American systems ultimately benefit partners’ own sovereignty, it can turn hesitation into cooperation. 

The Trump Administration deserves credit for recognizing that the U.S. must export not only agricultural goods and aircraft, but also the digital infrastructure of the future. By creating an AI Exports Program, Washington has taken an important step toward ensuring that American innovation—not Chinese subsidies—sets the standards for the AI era.

But the job is not finished. To succeed, the program must expand the scope of the AI stack to include telecom and interconnect capabilities, adopt a smarter risk tolerance in financing, and address the underlying concerns of foreign governments with honesty and seriousness.

The stakes could not be higher. AI is not merely another technology—it is a general-purpose capability that will shape military power, economic growth, and global influence for decades to come. Allowing China to dominate the global AI infrastructure market would be a strategic error of historic proportions. With foresight, flexibility, and resolve, the U.S. can still ensure that the AI future is built on trusted foundations.

The views expressed in this article are those of the authors and do not reflect either way the views of Venable LLP.