AI Opportunity, One Year Later

The Vice President Drew a Line in Paris. The Administration Built Behind It.

It’s been one year since the Vice President took the stage in Paris and broadcast the Trump administration’s priorities on AI: firmly placing opportunity at the forefront.

In the twelve months since, much has been said by naysayers on the risks of AI; there are those that would look to pause AI development entirely or limit the ways we use American AI to the purely aesthetic. But as the Vice President outlined at the AI Action Summit, to focus on the risks of AI would miss it for what it actually is: the greatest opportunity for America to surmount the challenges that have emerged across the past fifty years. A year ago, Vice President Vance laid out four pillars: establishing American AI as the global gold standard, removing obsequious regulatory barrier, eliminating ideological bias, and creating a pro-worker growth path. He declined to sign the summit’s international declaration pontificating on AI. Critics called it reckless - a repudiation of the international order. A sign the Trump administration was “unserious” and unsupportive of pro-human AI. Rather than pen meaningless documents, the administration built a blueprint: the clearest articulation of “Why AI” in the world.

The blueprint has been put into place. But the hardest work - putting the plan into Action - is just beginning.

Winning the Stack, Not Just the Benchmark

From a Silicon Valley perch, it’s easy to forget the lives that Americans actually live: they do not have limitless access to safe autonomous taxis or a paved path to prosperity. Just like any other technology, turning AI innovation into tangible benefits for Americans is accompanied by challenges; but if we let the challenges of electrical fires or internet crashing hold us back from embracing technology, America would lay lightless and disconnected while the rest of the world passed us by.

The Vice President was clear in Paris: “The United States is the leader in AI, and our administration plans to keep it that way.” On the surface, this is true. American frontier models remain the best in the world. Our highest-end models score six percent better on selected benchmarks than the highest-end Chinese alternatives. American data and benchmarks remain the baseline upon which competitors, including Chinese frontier AI, develop their own systems. But six percent is not dominance. It is a lead; a lead that narrows when you look at what’s beneath the software.

The Vice President spent part of his speech discussing the importance of the AI stack: everything we need to unlock tangible security and economic impacts from AI. This includes the models, data, and hardware, as he outlined, but it also includes the physical layers: the energy and infrastructure, the supply chain of raw materials, and the basic and advanced manufacturing required to build AI and deploy it for the gain of Americans. The US leads on the IP-heavy layers at the top of the stack. We are significantly disadvantaged in the physical layers beneath them - and those are the layers that matter most when AI systems start producing real-world effects.

The numbers are stark. America’s energy generation is roughly a third of China’s - and in 2024 alone, their capacity grew by 500 GW compared to our 48.6 GW. Only 12 percent of global rare earth production comes from the US, compared to 70 percent from China, which also controls nearly 90 percent of global refining. In total manufacturing value added, the basic manufacturing layer, China exceeds the US by $2.5 trillion. And while we consider advanced manufacturing to be the final bottleneck for the Chinese, China actually leads America in total output at the 7nm wafer process and above. While the Taiwan Semiconductor Manufacturing Company (TSMC) has began producing the highest-end 4nm wafers in Arizona, which China has not publicly demonstrated in volume, our total expected 2026 production of leading-edge chips will represent only 10 percent of global advanced chip production. This may seem like a barrage of statistics, but it’s the raw truth. We have the world’s best models; but once the AI stack requires you to build things, America is disadvantaged at every layer.

It’s not enough to build the best AI software. Real progress comes in the world of atoms. The question that should keep policymakers up at night is not which country’s model scores highest on a benchmark, but which country has an AI stack that can physicalize AI’s impact: manufacturing autonomous systems, scaling deployment at speed and low cost, and rapidly building and powering the data centers to run it all. On that question, the competition is closer than the headline numbers suggest; in most physical areas, we are behind.

The administration understands this. The next step for American AI dominance is controlling the entire AI stack: not just the models, data, or AI applications, but production of AI-specific hardware, building the best energy production and infrastructure, and developing resilient supply chains for the raw materials that undergird the AI industrial revolution.

Building Behind the Line

To the administration’s credit, the past year has been one of systematic execution. On day one, the President rescinded the Biden-era AI executive order and the onerous Biden Diffusion Rule. In July, the AI Action Plan formalized ninety federal policy positions across three pillars: accelerating innovation, building infrastructure, and leading in international AI diplomacy. Three accompanying executive orders supercharged work on AI exports, data center permitting, and ensuring that federally procured AI was maximally truth-seeking, not bound to ideological bias. In December, an executive order set the groundwork for a national AI policy framework to preempt the patchwork of state-level regulations that threatened to fragment the American market. American AI companies cannot compete globally while buried under fifty different domestic compliance regimes.

On permitting, the Council on Environmental Quality has worked to streamline National Environmental Policy Act (NEPA) reviews for priority projects. The bipartisan Simplify Permitting and Enable Efficient Development Act (SPEED Act) passed the House in December 2025 and awaits Senate action. The administration must continue with these projects: Make America Build Again.

In the world of infrastructure, the administration’s Stargate initiative has mobilized up to $500 billion in private investment for AI data centers and the energy to power them. The US Investment Accelerator can identify partners and deploy government capital through direct investment, subsidization, and tax incentives in the areas of the stack we’ve identified as critical. Private capital is often better suited for effective development; the government’s role is to attract it through deregulatory action and streamlined application processes, not run American AI through top-down bureaucracy.

Supply chain is one of the least discussed elements of the administration’s AI strategy, and Pax Silica has rung true as a winning solution to our issues. Launched in December 2025, it brought together America and its allies around a shared commitment to secure the full technology supply chain. If the twentieth century ran on oil and steel, the twenty-first runs on compute and the minerals that feed it. Pax Silica is the administration’s answer to a world in which China controls 70 percent of rare earth mining and 90 percent of refining: an alliance structure designed not for diplomatic signaling or the proliferation of meaningless Memorandums of Understanding, but for the hard work of ensuring the physical foundations of the AI revolution are built by America and our trusted partners.

Pax Silica was a strident first step. But supply chain resilience requires further effort: more bilateral critical minerals deals through the State Department, structured US off-take and financing through the Export-Import Bank (EXIM) and Development Finance Corporation (DFC), and a hard look at whether the archaic Minerals Security Partnership needs restructuring or replacement. Organizations like the DFC and EXIM have begun breaking ground on closing America’s critical mineral deficiencies, announcing Project Vault’s strategic minerals reserve as a unique policy lever to support the American supply chain as a buyer of last resort. Whether it’s the rare earth magnets that power autonomous vehicles or the actuators that allow robots to empower Americans in the jobs of the future, America has begun building the world’s most important pieces of technology at home. Less MOUs, more mending.

Silicon Valley is not always best at determining what products will actually change Americans’ lives. America has routinely underinvested in the foundational areas of technology that are not aesthetically impressive, but are crucially important nonetheless. The Trump administration is breaking this pattern.

Truth-Seeking AI

The Vice President’s third pillar, eliminating ideological bias, was quickly turned into a point of mockery for his European audience. The Paris speech referenced AI image generators depicting America’s Founding Fathers as racially diverse or declining to produce content based on ideological filters. The examples were easy to dismiss - an amusing oversight from tech companies kowtowing to cultural pressure. The underlying point, however, was not.

AI will always carry biases of various kinds. The goal is not a theoretical consensus or false neutrality, but truth-seeking AI: systems that give Americans straight answers, allow users to understand the biases embedded in the tool, and teach people how to engage with AI critically. If Americans cannot trust these systems, adoption stalls and AI’s potential goes unrealized.

Indeed, truth-seeking AI goes beyond “wokeness” or political mercuriality; it is the foundation for pro-human AI. How can we trust AI that lies to us, or misrepresents its intentions? Effective use and deployment of AI requires robustness. Truth, more than any other principle, is essential to realizing the benefits of AI.

The AI Action Plan and accompanying executive order addresses this through a cross-cutting priority requiring trustworthy systems free from ideological bias. But the deeper investment is in AI education: teaching Americans not just to use these tools, but to think clearly about them. Truthful AI is the first step – a thoughtful population is the second. The Presidential AI Challenge and the AI Education Task Force are early steps, and a clear marker of how seriously the administration is taking training the AI-empowered America of tomorrow. A country that leads in AI development will have financial success; a country that leads in AI literacy will have prosperity.

Cybernetics, Not Replacement

The Vice President’s mention of a pro-worker growth path was the most forward-looking element of the Paris speech and remains the area where the stakes are highest. Vice President Vance was right that AI will make Americans more productive, more prosperous, and more free. But for most Americans, that’s still a promise, not a paycheck.

The administration has not been idle. Executive Order 14278, “Preparing Americans for High-Paying Skilled Trade Jobs of the Future,” set a target of one million new active apprentices per year. In August, the administration delivered America’s Talent Strategy: a framework built on industry-driven training, upward mobility, streamlined systems, outcome-based accountability, and flexibility. The Work Pell Grant, created through the One Big Beautiful Bill, extended Pell eligibility to short-term credential programs for the first time, unlocking a new funding stream for the kind of training that actually leads to jobs.

These are serious policy tools. The question is whether they will be deployed at the scale and speed the moment demands.

Here is the problem in plain terms: Americans are sold a dream of white-collar jobs that are decreasingly meaningful, going into debt for careers that do not fulfill them, while critical industries are creating high-quality, economically stable jobs without the workforce to fill them. Nearly 40 percent of the nuclear workforce is expected to retire within the decade. The Navy needs 100,000 shipbuilding workers; only 5,000 are certified for nuclear-grade welding. TSMC Arizona created 4,500 new positions and struggled to hire for them. Eighty-three percent of aircraft maintenance technicians are expected to leave the workforce within ten years. These are not hypothetical gaps. They are the bottleneck standing between the administration’s reindustrialization vision and reality.

The Talent Strategy’s emphasis on industry-driven approaches and apprenticeship expansion is exactly right. But the execution vehicle matters. What Americans need is not another federal program they have to navigate; they need a package: training with guaranteed employment, transparent timelines, and a clear path from enrollment to a career that pays better than the median four-year degree.

The right frame for this is cybernetics: the art of empowerment, augmenting man with machine to accomplish more than either could alone. This is the Trump administration’s answer to automation anxiety. Not replacement, but enhancement. Not coding bootcamps, but industrial workforce development in the specializations that actually build the AI stack: semiconductor manufacturing, nuclear power operations, industrial cybersecurity, grid maintenance, robotics and automation, shipbuilding, and avionics. These are jobs that provide meaningful salaries, do not require a four-year degree, and offer the kind of career stability and dignity that the white-collar economy has increasingly failed to provide.

The Talent Strategy provides the policy architecture. The Work Pell Grant, Perkins V, and the Workforce Innovation and Opportunity Act (WIOA) provide the funding authorities. What is still needed is the operational vehicle: public-private partnerships that connect industry workforce needs directly to training institutions, guarantee job placement for graduates who meet performance standards, and market these careers with the kind of branding that makes young Americans want to sign up. Early cohorts should be visible - not another quiet grant program, but a flagship that shows Americans how technology will improve their lives.

A country that builds the best AI but cannot staff its semiconductor fabs, wire its data centers, or maintain its nuclear fleet has not won the AI race. It has built an engine without a workforce to run it. The administration has the framework. Now it needs to operationalize it at a pace that matches the ambition.

What the Next Year Demands

Looking at 2026, we need to take stock of what the Trump administration has accomplished and be resolute about what challenges remain.

The infrastructure investments are flowing. The regulatory environment has begun to be cleared. Pax Silica, the AI Exports Program, and the Technology Prosperity Deals are building the international architecture. The full AI stack, from minerals to models, is now the organizing framework for American technology policy. These are real achievements. The world scoffed at America’s “naive vision” a year ago.

Who’s actually delivered?

But America’s position in the AI race is more precarious than the headline numbers suggest. Our models are whiskers away from China’s, despite our advantages in chips and talent. We are likely to create AI systems that unlock real security and economic impact around the same time. When that moment arrives, the primary concern will not be benchmark scores. It will not be a tenuous vision of superintelligence. It will be which country has an AI stack that can physicalize that impact at scale. The US must create what we might call stack sovereignty: an AI stack that is resilient to exogenous factors and highly controlled and influenced by the United States.

Achieving stack sovereignty means continuing to remove regulatory barriers, deepening allied supply chain partnerships, and deploying capital - public and private - to build across every layer of the stack. It means making bets that take years to pay off: securing critical mineral supply chains, closing the manufacturing gap, building energy capacity, and developing the workforce to run it all. These are not the kinds of investments that generate headlines. They are the kinds that determine whether a country leads a technological revolution or merely participates in one.

The Vice President was right in Paris: this is about opportunity. But seizing that opportunity requires building in the world of atoms, not just bits. The administration set the terms a year ago. Now it must build the future those terms demand.

Ryan Hassan is a writer and researcher focused on the philosophy of science and technology, American reindustrialization, and national security. He writes for Pirate Wires, is a research fellow with Build American AI, and a New Science Fellow.