America Just Entered the AI Systems Race

The Department of Commerce’s new call for proposals under the American AI Exports Program marks a quiet but profound shift in U.S. strategy. For years, U.S. policy has focused on protecting advantage—controlling chips, restricting exports, and safeguarding frontier models. Those efforts remain necessary. But they are not sufficient.

This program signals something different: The U.S. is beginning to compete not just on AI technology—but on AI systems.

From Technology Competition to Systems Competition

The premise behind the program is straightforward but consequential. The U.S. government is asking industry to form consortia capable of exporting full-stack AI systems—integrated packages that combine hardware, data pipelines, models, cybersecurity, and applications.

This reflects a deeper reality – AI competition is no longer about who builds the best model. It is about who deploys the most complete, scalable, and trusted systems globally. As we have argued previously, “AI competition is no longer about who invents the best model… it is about who can scale trusted compute across borders and boundaries.”  Commerce has now operationalized that insight.

A Strategic Correction—But Only a First Step

The program represents a long-overdue correction in U.S. strategy.

It acknowledges three critical truths:

1. The Battlefield Is Global Deployment

The decisive terrain is not the lab—it is the market. The U.S. can build the world’s most advanced AI systems and still lose if those systems are not widely adopted.  Adoption—not just capability—determines long-term advantage.

2. The Unit of Competition Is the Stack

AI is not a standalone product. It is an ecosystem: compute infrastructure, energy systems, connectivity, models, applications. Countries do not buy algorithms. They adopt systems that must operate reliably within their infrastructure, regulatory environment, and budget constraints. That is why Commerce is emphasizing “full-stack” solutions—and why consortia, not individual firms, are now the organizing unit.

3. Scale Requires Coordination

Winning in global markets requires more than innovation. It requires, financing, diplomacy, commercial presence, standards alignment. As discussions at our recent NYU–WISC roundtable made clear, “the central challenge is no longer conceptual design. It is translating design into deployable, financeable, and politically durable infrastructure.”  This program is an attempt to build that machinery.

The Hard Part: Execution

If the program succeeds, it will reshape global AI markets. But success is far from assured. The challenge is not defining the architecture. It is executing it at scale. Three gaps stand out.

1. Financing Remains the Bottleneck

The program offers access to financing tools. But access is not the same as deployment.

In many target markets, AI adoption competes with grid expansion, telecom upgrades, workforce development. Capital is the binding constraint. Competitors are not just selling technology—they are bundling financing, construction and long-term service agreements. Unless U.S. offerings can match that model, superior technology will not translate into installations. As I noted in our submission to Commerce, “the Program’s success will be determined by whether it wins installations.”

2. The Volume Market Is Not the Frontier

U.S. firms dominate the high end of the market. But the global contest will be decided elsewhere. Most countries do not need the most advanced model. They need systems that are affordable, adaptable, deployable under real-world constraints Mid-tier and open-weight models—paired with implementation support—will determine adoption across emerging markets. If the U.S. focuses only on the frontier, it risks ceding the volume market—and with it, long-term ecosystem influence.

3. Systems Must Be Built With, Not Just For, Partners

Countries adopting AI systems increasingly demand sovereignty over data and operations, participation by local firms, and integration with domestic infrastructure. This is not a barrier. It is a requirement for durability. As we have emphasized, integration of national champions is not concession—it is incentive alignment.  Systems that exclude local actors may deploy. Systems that include them endure.

Competing in a Mixed-Trust World

A final challenge is structural. Many target markets already operate on infrastructure that includes components the U.S. does not fully trust. Telecommunications networks, cloud environments, and digital systems are often hybrid. The strategic question is not whether to engage in these environments—but how. If participation requires perfect conditions, large portions of the world will default to alternative providers.

If engagement is possible in mixed-trust environments, then security must be embedded in architecture, governance must be enforceable, risk must be managed, not avoided. This is a design challenge—not a binary choice.

What Success Will Require

The U.S. begins from a position of strength with leading AI firms, advanced semiconductor design, deep capital markets and trusted governance frameworks. But advantage at the frontier does not guarantee global scale.

Success will require:

• Financing that moves at market speed
• Persistent commercial and diplomatic engagement
• Competitive, scalable offerings beyond the frontier
• Clear and predictable policy signals
• Integration with trusted partners and local firms
• Architectures that can operate in diverse environments

In short, it will require aligning policy with how systems are actually deployed.

The Stakes

This program is not just about exports. It is about which systems the world runs on. The country whose AI stack becomes embedded in the daily operations of governments, businesses, and institutions will shape standards, data flows, governance models and long-term economic relationships. Frontier leadership is the starting point.

System-level deployment is what determines the outcome.

A Defining Moment

The Commerce Department’s call for proposals is an important step. It reflects a growing recognition that the U.S. must compete in the market—not just at the frontier. But recognition is not execution.

The architecture exists. The strategy is emerging. The task now is to build the institutional, financial, and commercial capacity to deliver at scale. Because in this race, the winner will not be the country with the best technology. It will be the country whose systems the world adopts.

Development Impact: As AI competition shifts to deployable systems, countries will be shaped by whether those systems empower local capability or create dependency. For emerging markets, access to scalable, financeable, and adaptable AI infrastructure will determine long-term growth and economic sovereignty.