Valar Atomics, NVIDIA and the New AI Infrastructure Trade: Power Is the Prize -( $BA $NVDA )

The next chapter of the artificial-intelligence buildout may be written not only in silicon, but in steel, concrete, uranium fuel and very patient permitting offices. Valar Atomics’ push to develop nuclear “gigasites” alongside the AI infrastructure boom highlights a larger investment theme: America’s race for AI leadership is becoming inseparable from the race to secure abundant, dependable electricity. For many, the opportunity spans more than reactors. It reaches into the AI compute stack led by NVIDIA Corp. (NASDAQ: NVDA), the defense-industrial supply chain represented by Boeing Co. (NYSE: BA), utilities, grid equipment, engineering, fuel-cycle capacity, advanced manufacturing and the increasingly strategic business of building physical things on time.

AI Has Found Its Limiting Reagent: Power

For years, the AI investment thesis had a familiar rhythm: faster chips, larger models, bigger data centers and ever-more impressive capital-expenditure figures. Now comes the awkward, but possibly investable, question: where will all the electricity come from? In a Bloomberg Technology interview, Valar Atomics founder and CEO Isaiah Taylor argued that nuclear energy is particularly well suited to support the AI race because it is clean, scalable and can be concentrated near large data-center campuses. The company says it has demonstrated nuclear generation powering an NVIDIA Blackwell chip, while pursuing a model built around larger, vertically integrated nuclear development sites. That proposition gets directly at the AI sector’s emerging constraint. Computing hardware has become spectacularly capable; the power system has been rather less enthusiastic about matching the pace. Data centers need electricity that is dispatchable, high-density and available around the clock—not merely available when the weather is in a cooperative mood. Valar’s stated strategy is to build hundreds of reactors at dedicated gigasites rather than rely entirely on traditional grid-connected project structures. Its proposed applications include data-center power, industrial energy, hydrogen and cleaner hydrocarbon-fuel production. The company frames the market backdrop as a U.S. power shortfall exceeding $100 billion, driven by demand from data centers, grid modernization and industrial reshoring. Those are company projections rather than independent forecasts, but they capture the growing urgency around reliable power capacity.

Nuclear Is Moving From Policy Topic to Capacity Question

The bullish case for advanced nuclear is not that construction risk has vanished. It plainly has not. The stronger point is that the strategic demand signal has become much harder to ignore. Taylor’s critique is blunt: the United States must move faster if it intends to lead in AI, especially when China continues to build nuclear capacity at a far more aggressive pace. The precise number of Chinese projects changes over time and depends on the reporting date and definition used; a World Nuclear Industry Status Report table counted 32 Chinese reactors under construction as of July 1, 2025. The broader competitive point remains clear: China’s nuclear construction pipeline is substantial. America’s advantage, meanwhile, is increasingly tied to its ability to translate exceptional technology into dependable infrastructure. That means speeding up siting, licensing, fabrication, construction and interconnection, activities that are not as glamorous as a new GPU unveiling, but are considerably more useful when the lights must stay on. Valar’s Ward 250 project illustrates the new urgency. The company says it is working with the Department of Energy, the San Rafael Energy Research Center and the State of Utah, with the goal of bringing its test reactor online before July 4, 2026. Its Ward 250 site was described as having broken ground on September 17, 2025. One sophisticated joke here is that AI may be able to write a sonnet in a fraction of a second, but it still cannot negotiate a transmission queue. For that, America will need engineers, manufacturers, regulators and a great deal of institutional follow-through.

NVIDIA’s Buyback Sends a Powerful Signal

NVIDIA’s role in this story is more than symbolic. The company sits at the center of the compute demand driving the new energy conversation, and it has just underscored its confidence with a record-setting capital-return commitment. On September 28, NVIDIA announced an additional $150 billion authorization under its existing share-repurchase program, raising the remaining authorization to $235 billion. The company expects to execute the remaining program through fiscal 2028. NVIDIA described the $150 billion increase as the largest share-repurchase authorization increase in history. That does not make NVIDIA immune to valuation debates, competition, product-cycle risk or the broader cyclicality that tends to visit even the most admired semiconductor franchises. But it does send an unmistakable message: management sees sufficient financial strength to invest in the AI platform, support its ecosystem and return an extraordinary amount of capital to shareholders. For the nuclear-and-AI thesis, NVIDIA’s announcement reinforces a central market reality. The largest beneficiaries of AI are not behaving as if the buildout is a brief science-fair project. They are allocating capital as though AI infrastructure will remain a strategic priority for years.

Defense Adds a Second Tailwind

The energy story is not occurring in isolation. It is unfolding alongside renewed U.S. emphasis on defense technology, domestic manufacturing and resilient supply chains. Boeing and private defense-technology company Anduril recently advanced in the U.S. Army competition for the Integrated Fires Protection Capability Increment 2 Second Interceptor. The system is intended to defend fixed and forward operating bases against cruise missiles, militarized unmanned systems and related airborne threats. Boeing’s proposed interceptor incorporates a solid-rocket motor from Anduril Rocket Motor Systems and a modular architecture intended to support quicker, lower-cost upgrades. For Boeing (NYSE: BA), the development does not guarantee a final award, and the companies have not disclosed the value of a potential contract. Still, it supports a broader theme: Washington’s strategic priorities increasingly reward production capacity, modularity, faster iteration and domestic industrial depth. That framework has a clear overlap with the nuclear conversation. Whether the challenge is microreactors, missile interceptors, advanced chips or high-voltage equipment, the prize increasingly goes to companies that can bridge the gap between a credible prototype and repeatable production. In other words, the market is rediscovering that hardware execution is not a nuisance; it is the moat.

A Takeaway

The emerging bull case is not simply “buy nuclear.” It is that AI is forcing a repricing of physical infrastructure. Many may want to watch five developments:

• Nuclear project execution: Demonstrations are important, but repeatable construction, licensing progress, fuel access and operating performance will separate aspirational ventures from durable enterprises.
• AI power contracting: Long-term power-purchase agreements, co-located generation and data-center siting decisions could provide the clearest evidence that AI demand is reshaping energy markets.
• Grid bottlenecks: Transmission and interconnection delays remain a practical constraint. Companies that relieve those bottlenecks may become as strategically relevant as companies that generate the power.
• Capital discipline: Nuclear and grid projects require serious capital, so financing structures, off-take agreements and cost control deserve at least as much attention as promotional timelines.
• Public-market pathways: Valar Atomics and Anduril are private, which means public investors may initially access the theme through NVIDIA (NASDAQ: NVDA), Boeing (NYSE: BA), utilities, industrial suppliers, electrical-equipment providers and nuclear-fuel-chain participants rather than through the startups themselves.

The market’s newest infrastructure trade may therefore be bigger than a single reactor developer or chip company. It is the industrial system required to turn AI ambition into reliable computing capacity: semiconductors, power generation, transmission, defense manufacturing and the old-fashioned ability to build at scale. That is a compelling narrative for Wall Street because it joins two forces many already understand: accelerating AI demand and renewed national focus on industrial capability. The risk is that timelines slip and costs rise, two habits infrastructure projects have perfected over the decades. The opportunity is that the companies which solve those problems could become essential suppliers to the most capital-intensive technology cycle in modern history.

The Sources

1. Bloomberg Technology: “Valar Atomics Raises $1 Billion to Power the AI Era” (YouTube video)
2. Valar Atomics: The New Atomic Age
3. Valar Atomics: Ward 250
4. Anduril: Boeing and Anduril’s Midrange Interceptor Advances in U.S. Army Competition
5. Boeing: Boeing and Anduril Midrange Interceptor Advances in U.S. Army Competition
6. NVIDIA: NVIDIA Announces a $150 Billion Share Repurchase Authorization Increase
7. Reuters: Nvidia Boosts Share Buyback by a Record $150 Billion as AI Boom Fuels Growth
8. Yahoo Finance: Nvidia Announces $150 Billion Stock Buyback
9. World Nuclear Industry Status Report: Nuclear Reactors Under Construction