A golden age should be recognizable in what people can do. Can a small business serve customers with tools that once required an entire department? Can a student enter a serious technical education without shutting the door on the rest of life? Can a startup challenge an established company? Can a community welcome new industry and become better off because it is there?

Those are the ambitions I want American leadership to organize around: artificial intelligence, quantum computing, semiconductor manufacturing, space startups, higher education, and the energy to support them.

Give credit where it is due.

The companies building this future deserve to be named. Their engineers, researchers, technicians, and investors are turning ambitious ideas into useful products and productive capacity. I believe they are helping pave the way for an American golden age.

Micron is investing in American memory-chip manufacturing and research. NVIDIA develops the accelerated-computing hardware and software at the center of much of the AI economy, while AMD provides another major source of AI computing hardware. These are the kinds of advanced industries America should want to strengthen.

IREN and CoreWeave supply data-center and cloud infrastructure for demanding AI workloads. OpenAI and Anthropic are developing models and tools that people can use to write software, analyze information, and improve everyday work. The connection matters: chips and electricity become computing capacity, and that capacity becomes a tool in someone’s hands.

The opportunity reaches further. IBM is advancing quantum computing, and Rocket Lab builds launch vehicles, spacecraft, and satellite systems. International partners matter too: TSMC’s Arizona manufacturing work with NVIDIA shows how allied expertise can strengthen production on American soil. American renewal can grow through cooperation with companies and talent from other countries.

I am pro-capitalism. Entrepreneurship, investment, and the prospect of earning a profit are powerful reasons to take risks and build something useful. We should want these companies to succeed—and want the next generation of founders to have a fair chance to compete with them. Fair taxation, open competition, and responsible treatment of communities are how we make that success more durable. Let’s be smart about the capitalism we practice.

Cooperation as a competitive advantage.

Competition does not require isolation. A company can compete fiercely with another company while purchasing its technology, sharing infrastructure or developing something neither could build as effectively alone. Countries can do something similar: preserve their independence while strengthening their capabilities through trade, common standards and strategic alliances.

This is the connection to the Wilsonianism I am arguing for. Cooperation can be a source of competitive strength. Agreements allow participants to combine different advantages without surrendering their distinct interests. The objective is to become more capable together while retaining the freedom to compete.

The companies building AI infrastructure illustrate this principle. Their relationships form an overlapping network of suppliers, investors, infrastructure providers and customers. The same company can occupy several positions at once. Partnerships can change as technologies improve and new competitors emerge.

Open the interactive network in its own window →

Calling these arrangements “circular financing” does not settle whether they create value. Neither does calling them “cooperative competition.” The meaningful questions are what gets built, whether customers need it, and whether the resulting market offers more capability, greater reliability and meaningful choice.

That last point matters most. Cooperation between large companies should expand opportunity for the businesses and people who depend on them. If it merely locks customers into a few powerful providers, it falls short of the golden age we should be building.

The principle is not that every partnership is good, or that more cooperation is always better. It is that well-designed cooperation can make competition more productive. Stronger American companies and stronger international alliances can advance the same ambition: giving people better tools, more opportunity and greater power to build something of their own.

Public leadership should support that ambition.

International supply chains in practice

South Korea’s SK hynix showcased 48 GB, 16-layer HBM4 at CES 2026, describing that version as under development. HBM, or high-bandwidth memory, stacks memory chips vertically to provide the capacity and bandwidth used by AI accelerators. The company also displayed 36 GB, 12-layer HBM3E. The distinction between a development demonstration and commercial availability matters when assessing supply.

Cross-border investment also includes production in the United States. SK hynix reported breaking ground on its Indiana advanced-packaging facility in August 2026. Alongside TSMC’s Arizona operations, it illustrates how a foreign supplier can participate in both international supply networks and American manufacturing.

Sandisk and Japan’s Kioxia extended their Yokkaichi manufacturing joint venture through 2034, with their Kitakami agreement aligned to the same date. Their work concerns NAND flash storage, a different function from HBM’s role alongside AI processors. Separately, Sandisk is scheduled to join the S&P 100 effective September 21, 2026. Index membership is a financial-market designation; the manufacturing partnership documents the underlying international production relationship.

Donald Trump is right to recognize the importance of AI infrastructure. His July 2025 executive order made expanding data centers and supporting infrastructure a federal priority. I agree with the ambition to strengthen America’s position in this technology. Each policy used to pursue it still deserves scrutiny.

Democrats should be capable of that distinction too. Opposition to a particular project may be justified by its costs or location. Turning opposition to data centers into a general position against the infrastructure of computing would surrender an enormous opportunity.

The better political argument is to represent the communities being asked to host that infrastructure. Demand a worthwhile deal. Help shape the future being built there.

Small businesses can gain real capacity.

Imagine an independent business owner who needs a website. AI can help produce an initial version in minutes, then assist with revisions, product descriptions, and customer questions. The owner still has to check that the site works, that the promises are accurate, and that customers’ information is protected. But the distance between having an idea and trying it has become much shorter.

The same possibility extends to drafting emails, preparing estimates, organizing customer inquiries, and assisting with sales follow-up. Used well, these tools can give a small firm some capabilities associated with a much larger company, at a fraction of the organizational overhead.

This is more than a prediction about convenience. Research on an AI assistant used by customer-support workers found productivity improvements, particularly among less experienced workers. One workplace study does not prove every business will benefit. It does show a concrete mechanism for helping people do more with the resources they already have.

I previously wrote about AI expanding an independent citizen’s capacity. The business version of that argument is equally compelling. Skill, judgment, and imagination can travel further when useful tools are affordable.

I do not see the companies creating these capabilities as enemies of the American worker. Their technology can uplift people who learn to use it: a technician diagnosing a problem, an employee taking on more sophisticated work, or a business owner serving customers who were previously out of reach. Greater capability can create paths to better work and new enterprises.

Some tasks will be automated, and some workers will face displacement. Taking that seriously means giving people a practical chance to adapt through training, affordable tools, and time to learn. Employers share that responsibility. A worker’s opportunity to benefit should depend on more than whether they could afford to prepare on their own.

That opportunity will not distribute itself automatically. Large firms and authoritarian governments can use AI too. Affordable access, competing providers, portable business data, and education will influence who benefits. My optimism is about what people can accomplish with these tools—and the choices we can make to keep that opportunity open.

We should have been building the power supply.

Some of us saw AI coming years ago. We did not have to predict every model or the exact electricity demand to see that a more electrified, technologically advanced society would need serious investment in energy.

The 2019 Green New Deal resolution called for clean power, modern electricity grids, and affordable access. It was not a specific forecast of today’s AI buildout. But its underlying insistence that America needed to modernize its energy system was relevant to the future now arriving.

By December 2024, the Department of Energy was explicitly addressing data-center demand. Its report estimated a 4.4 percent share of U.S. electricity consumption in 2023 and projected 6.7 to 12 percent by 2028. That range was a forecast, but the planning challenge was already clear.

My preferred answer combines renewables, storage, transmission, efficiency, and nuclear power. Different places will need different combinations. Small modular reactors could become part of that supply; the DOE-backed TVA and Holtec projects target deployment in the early 2030s. Those are targets, not electricity available today. We should develop promising technology while building the capacity available sooner.

Existing infrastructure changes the timetable.

Speed matters in this buildout. A site with usable power connections, installed equipment, and an experienced operating team has a head start over a project still waiting for those pieces. Existing infrastructure represents years of planning and investment; it deserves recognition when discussing how quickly new computing capacity can reach customers.

Applied Digital reported fully energizing the first 100 MW building at Polaris Forge 1 in November 2025, part of its campus deployment for CoreWeave. Nebius’s second-quarter 2026 shareholder letter describes an operating AI cloud alongside a substantial construction and contracted-power pipeline. These companies, along with IREN and CoreWeave, have infrastructure and operating experience that a new entrant cannot acquire overnight.

The details still differ by site. Energized capacity, contracted future power, and proposed expansion are different stages. Nebius’s May 2026 agreement with Bloom Energy, for example, planned on-site fuel cells to shorten the time to power and reduce dependence on new transmission construction. That is a delivery strategy, rather than evidence that every planned facility already has all the grid capacity it needs.

Clear conditions can support timely delivery.

Community agreements should recognize infrastructure already in place and focus on the actual additional demands of a project. Clear, predictable conditions can help operators plan and residents understand the benefits and costs. Where new grid connections or expansion are needed, operators should pay the costs attributable to their projects, with protections against passing abandoned-project costs to households. Additional power supply should have credible financing and delivery schedules.

Water use, backup-generator emissions, noise, and construction impacts should be measured, disclosed, and subject to enforceable limits suited to the location. Communities need access to the evidence and a way to obtain remedies when commitments are broken.

The financial deal matters as much as the engineering. Tax revenue, permanent jobs, training opportunities, and local procurement should be stated realistically. A town should be able to compare those benefits with infrastructure and service costs before approving tax concessions. A project that cannot meet reasonable conditions may need a different design or location.

Building well is how this industry earns durable public support. Washington can help with research, energy planning, and clear rules. Local residents should be participants in the bargain.

Prepare people to use what we build.

Education is central to this future. Students need to learn the underlying subjects and how to question machine-generated answers. Schools should distinguish between assignments that assess independent knowledge and work that permits AI assistance. Professional competence still requires someone who understands what the tool produced.

Universities, community colleges, technical training, and apprenticeships all have a place. An advanced economy needs researchers and entrepreneurs alongside electricians, technicians, construction workers, and skilled manufacturing teams. The opportunity should reach beyond those who already have an expensive credential.

In What Went Wrong, I argued for fair taxation, room for challengers, and greater discipline about war. Those principles support this agenda. A country that protects competition and preserves the conditions for peace gives its people more room to invest in the future.

A golden age worth building would make ordinary people more capable: able to learn, start businesses, solve problems, and improve their communities. AI and advanced industry can help deliver it. Leadership should make that possibility practical.

Read the companion essay: What Went Wrong →

Disclosure

I may own shares in some of the publicly traded companies discussed in this essay and may therefore have a financial interest in their performance. This essay expresses my personal views and is not personalized investment advice or a recommendation to buy or sell any security. Do your own research, including reviewing company filings and investment risks. Investing involves the risk of loss.