Skip to content
The ColumnOpen letter· No. 2496

To those who still doubt the nuclear race for AI

Introduction: a demonstration that deserves your attention

Premium reading
MadMax
Key takeaways
  1. Introduction: a demonstration that deserves your attention
  2. What happened on July 1 , 2026
  3. Dear reader, I'm writing this open letter because an event that took place on July 1, 2026 , in the Utah desert deserves attention, even though it didn't make the evening headlines.
Transparency

Facts, quotes, and cited links remain in the body. Interpretations are framed as analysis or opinion according to the format.

Introduction: a demonstration that deserves your attention

What happened on July 1, 2026

Dear reader, I'm writing this open letter because an event that took place on July 1, 2026, in the Utah desert deserves attention, even though it didn't make the evening headlines. California startup Valar Atomics ran its advanced nuclear reactor, the Ward 250, to directly power an Nvidia RTX Spark PC built on the Blackwell architecture, during a public demonstration in Emery County.

This was not a mere public-relations exercise. The reactor had reached criticality a few weeks earlier, on June 18, 2026, clearing a major regulatory and technical milestone for any new nuclear technology.

Why I wanted to tell you directly

I'm writing to you because the link between nuclear energy and artificial intelligence is becoming one of the defining stories of the decade, and yet it remains largely undercovered by mainstream media.

The data centers powering the world's most powerful AI models consume a colossal amount of electricity, and the question of who will produce that electricity, and how, will shape the technological geopolitical balance of the coming decades.

I won't hide my enthusiasm for this kind of initiative. Watching a young American company concretely solve, on the ground, one of the biggest obstacles to Western AI growth is the kind of news that restores confidence in our capacity for innovation.

Who is Valar Atomics, and why its bet matters

A young company with outsized ambition

Valar Atomics is an American startup specializing in advanced nuclear microreactors, which raised roughly $450 million in its latest funding round, valuing the company at around $2 billion.

Its Ward 250 reactor was designed specifically to meet an urgent need: providing a stable, dense power source with no direct carbon emissions, capable of feeding demanding industrial facilities like artificial intelligence compute centers.

Still modest output, but a powerful symbol

During the demonstration, the reactor was running at roughly 37% of capacity, producing about 100 thermal kilowatts, an output still far from what's needed to power a full data center, but enough to prove the concept's technical viability under real-world conditions.

This modest figure shouldn't overshadow the essential point: it's the first time an American advanced nuclear reactor of this type has directly powered a chip designed for artificial intelligence, a technological and symbolic milestone.

I'd rather be honest with you: 100 kilowatts isn't much at industrial scale. But I remember that every great technological leap started with a modest proof of concept before becoming massive infrastructure.

The Nvidia partnership and the 30-megawatt data center project

An ambitious joint feasibility study

Valar Atomics and Nvidia announced, alongside this demonstration, the launch of a joint feasibility study for building a 30-megawatt artificial intelligence compute center, powered in a closed loop by nuclear energy and designed to use no local water at all.

This water-conservation feature is no side detail: conventional data centers consume massive amounts of water for cooling, an increasingly sensitive issue in arid regions like Utah, where this project is being considered.

Why Nvidia is engaging in this kind of partnership

Nvidia, whose chips dominate the global market for artificial intelligence semiconductors, has a direct interest in securing reliable power sources for its future data centers, at a time when AI's energy demand is growing faster than traditional power grids' capacity.

This kind of collaboration between chipmakers and nuclear innovators could well become the norm rather than the exception in the years ahead, as energy pressure on the sector keeps intensifying.

A data center with no local water consumption, powered by next-generation nuclear energy: if someone had described this project to me five years ago, I probably would have smiled skeptically. Today, it's a serious feasibility study between two very real companies.

The Department of Energy's reactor pilot program

A race against the clock set by Washington

Valar Atomics' demonstration is part of the Reactor Pilot Program run by the United States Department of Energy, which had set an ambitious target: get at least three advanced reactors to criticality before the symbolic date of July 4, 2026.

According to available information, three companies actually hit that target on time: Valar Atomics, Antares Nuclear, and Deployable Energy, a result the Department of Energy hailed as proof of the acceleration underway in the American nuclear sector.

A deliberate political push to revive American nuclear power

This government initiative reflects a clear intent by the American administration to reposition the United States as the global leader in advanced civilian nuclear power, a sector where China and Russia have built a considerable lead over recent decades in terms of the number of reactors built.

The Department of Energy explicitly presented this pilot program as a step toward launching a "new atomic age" for America, with direct implications for national energy security.

Let's say it plainly: while some Western countries hesitated on nuclear power out of political fear, China and Russia kept building. Seeing Washington take back the initiative with this kind of pilot program is a signal I welcome without reservation.

The geopolitical stakes behind this energy race

AI as the new technological battlefield

The ability to produce enough reliable, abundant electricity is becoming a decisive factor in the global race for artificial intelligence, on par with the computing power of the chips themselves or the quality of the algorithmic models developed.

A country or economic bloc unable to supply the energy its data centers need will find itself mechanically disadvantaged in the AI race, no matter how strong its software engineering talent may be.

Why the West cannot afford to lose this ground

Faced with a China that keeps building nuclear plants and investing massively in energy infrastructure dedicated to artificial intelligence, the United States and its Western allies must urgently accelerate their own production capacity, or risk losing a major strategic advantage.

Advanced nuclear power, with its energy density and reliability, appears to be one of the few solutions capable of meeting this colossal demand without relying exclusively on fossil fuels or sensitive foreign imports.

I've said it for a long time to anyone who will listen: the West's technological sovereignty also depends on its energy sovereignty. You cannot win the artificial intelligence race with a power grid from the last century.

The remaining regulatory and technical challenges

A still-young technology facing a heavy regulatory framework

Despite this promising demonstration, advanced microreactors like Valar Atomics' still need to clear numerous regulatory hurdles before they can be deployed at commercial scale, notably before the American Nuclear Regulatory Commission.

The approval process for this kind of technology remains long and complex, which could slow the pace of deployment relative to the ambitions expressed by the company and its industrial partners.

Scaling up, a major industrial challenge

Going from a demonstration producing about a hundred kilowatts to a full 30-megawatt data center represents a considerable leap in scale, one that will require substantial additional investment and rigorous technical validation at every step.

It would therefore be premature to view this demonstration as definitive proof that advanced nuclear power alone will solve artificial intelligence's energy crisis in the short term.

I'd rather speak to you frankly than sell you a dream: between a successful demonstration and large-scale commercial deployment, there are often years of thankless regulatory work. Let's stay enthusiastic, but clear-eyed.

The water question, an underrated issue in the AI sector

Water consumption is becoming an increasing concern

Conventional data centers are heavy water consumers for their cooling systems, an increasingly critical issue in regions facing growing water shortages, such as certain areas of the American Southwest.

The joint project between Valar Atomics and Nvidia, explicitly designed to operate without local water consumption, directly addresses this increasingly pressing environmental concern for the tech industry.

An additional argument in favor of advanced nuclear power

This approach could strengthen the social and environmental acceptability of advanced nuclear power among sometimes wary local communities, by demonstrating that this technology can coexist with limited natural resources rather than further depleting them.

This is an argument that advocates for advanced civilian nuclear power should push harder in the public debate, rather than focusing solely on arguments about power output and energy reliability.

Here's an argument we don't hear enough in the public nuclear debate: its ability to conserve water rather than waste it. I think this aspect deserves a more central place in the sector's messaging.

What this means for the future of the semiconductor industry

A growing dependency between chips and energy

Valar Atomics' demonstration illustrates an increasingly unavoidable reality for the semiconductor industry: the performance of the most advanced chips, like those built on Nvidia's Blackwell architecture, can no longer be separated from the question of their power supply.

This growing interdependence is pushing chipmakers to get directly involved in energy projects, a strategic shift that would have seemed unlikely just a few years ago.

A model that could inspire other tech giants

If the partnership between Valar Atomics and Nvidia proves successful, it could inspire other tech giants to forge similar alliances with nuclear-sector innovators, accelerating the energy transition of the global data center industry.

This dynamic could profoundly transform how the Western tech industry plans its energy infrastructure for decades to come.

I see in this partnership a potential model far broader than just the case of Nvidia and Valar Atomics. If other tech giants follow suit, we could witness a structural transformation of the energy that powers Western artificial intelligence.

Critical voices and precautions to keep in mind

Legitimate questions about safety and waste

Some nuclear safety experts and environmental advocates continue to raise legitimate questions about the management of radioactive waste generated by these advanced microreactors, as well as about the long-term safety protocols applicable to decentralized facilities of this kind.

These concerns deserve to be taken seriously and should not be dismissed out of hand in the name of technological enthusiasm, even though they don't call into question the overall relevance of the project.

Essential regulatory transparency

For this kind of technology to durably earn public trust, it will be essential for Valar Atomics and other players in the sector to maintain rigorous transparency about their safety protocols and performance results, under the supervision of relevant regulatory authorities.

This transparency will be all the more crucial as the sector seeks to win over local communities sometimes reluctant to see new nuclear infrastructure built in their area.

I'm not the type to dismiss legitimate criticism just because it dampens my enthusiasm. Transparency on safety must remain an absolute priority, or public trust will collapse at the first incident.

The role of microreactors in the broader energy transition

One piece of the puzzle among others

This demonstration should be placed in a broader context: advanced nuclear power is only one piece among others in the energy puzzle needed to meet artificial intelligence's growing demand, alongside renewable energy and energy storage technologies.

No single technology will likely be enough on its own to solve AI's energy crisis; it will instead take a combination of complementary solutions to meet this exponential demand.

Why diversity of energy solutions is a strength

This diversity of energy solutions is actually a strength rather than a weakness for Western economies, by reducing their dependence on a single technology or a single energy source.

Microreactors like Valar Atomics' thus complement an already diversified energy portfolio, rather than replacing it entirely.

I'm always wary of narratives that present a single technology as the miracle solution to all our energy problems. The West's real strength lies in its ability to combine several complementary solutions rather than betting on a single card.

Possible effects on employment and the local economy

Expected economic benefits in Utah

The deployment of projects like Valar Atomics' in Emery County, Utah, could generate significant economic benefits for a region historically dependent on the coal industry, as it transitions toward new employment sources tied to advanced energy.

This local economic reconversion, if fully realized, could serve as a model for other American regions facing similar industrial and energy transition challenges.

Potential for creating skilled jobs

The jobs generated by this kind of project generally require a highly skilled workforce in nuclear engineering, industrial maintenance, and the management of complex systems, which could help revitalize American nuclear expertise over the long term.

This dynamic fits into a broader context of reviving nuclear expertise in the United States, a sector that had seen a marked slowdown over several decades.

Watching a region historically tied to coal reconvert toward advanced nuclear power to fuel artificial intelligence is a powerful image of successful industrial transition. I sincerely hope other regions will follow this example.

What this demonstration reveals about American entrepreneurial spirit

A culture of calculated risk-taking

The trajectory of Valar Atomics, a young company that managed to bring an advanced reactor to criticality in record time compared to the nuclear sector's historical standards, illustrates an American corporate culture defined by calculated risk-taking and rapid execution.

This ability to move fast while still meeting the nuclear sector's strict regulatory requirements sets this kind of startup apart from traditional large industrial groups, which are often slower to innovate.

An example for the Western tech ecosystem

This success, still modest in terms of power output, could nonetheless inspire other Western entrepreneurs to launch ambitious technology projects in the energy sector, a field long seen as the exclusive domain of large established corporations.

It's precisely this kind of entrepreneurial boldness that has always allowed the West to maintain a technological edge over its international competitors.

This story reminds me why I remain fundamentally optimistic about Western innovation capacity, despite all the challenges. A small team getting an advanced nuclear reactor running in record time is exactly the kind of boldness worth celebrating.

What Western allies should take away from this

A lesson for Europe and Canada

Beyond the American case alone, this demonstration should serve as a lesson for all Western allies, particularly in Europe and Canada, where debates on civilian nuclear power often remain hostage to short-term political considerations rather than a clear-eyed assessment of future energy needs tied to artificial intelligence.

Countries that delay massive investment in advanced nuclear power risk finding themselves structurally disadvantaged in the AI race, unable to supply the energy their own compute centers need without relying on costly foreign imports.

Desirable Western coordination

Closer coordination between the United States and its allies on advanced nuclear regulatory standards could accelerate the deployment of these technologies across the entire Western bloc, rather than leaving each country to reinvent its own regulatory approach in isolation.

This reinforced cooperation could also accelerate the sharing of technical expertise between Western companies and research labs, a considerable advantage against rivals that have no hesitation pooling their national efforts.

I say this with conviction: the West would gain enormously by harmonizing its regulatory approaches to advanced nuclear power rather than letting each country move forward alone. This fragmentation costs us precious time against more coordinated rivals.

The broader context of the global nuclear revival

A movement that goes beyond American borders

Valar Atomics' demonstration fits into a broader revival of civilian nuclear power observed in several Western countries, from the United Kingdom to France, where governments are gradually reconsidering their energy policies in light of growing needs tied to artificial intelligence and decarbonization.

This international convergence suggests the political moment is particularly favorable for accelerating investment in advanced nuclear power, provided Western governments know how to seize this window of opportunity quickly.

A window of opportunity that won't stay open forever

The West's geopolitical rivals won't wait for internal political debates to resolve themselves before continuing to invest massively in their own civilian nuclear capabilities, which makes the urgency to act all the more pressing for Western decision-makers.

Every additional year of delay in deploying Western advanced nuclear power potentially represents an additional year of lead ceded to competitors less scrupulous about transparency and safety.

I'll keep repeating it: the window of opportunity for the West to regain the advantage in advanced nuclear power won't stay open forever. Every year of political hesitation is paid for in real technological delay.

Conclusion: why I'm writing you this letter today

A call to follow this story closely

I'm writing this open letter because I believe this story deserves your ongoing attention, well beyond fleeting technological curiosity. The convergence of advanced nuclear power and artificial intelligence will shape a major part of our collective energy and technological future.

The decisions made today by companies like Valar Atomics and partners like Nvidia will have repercussions reaching well beyond the tech sector alone, all the way to our electricity bills and our national energy security.

A call for informed vigilance rather than blind enthusiasm

I invite you, then, to follow this story with the same mix of enthusiasm and caution that I try to cultivate myself: recognizing the real importance of these advances, without giving in to naive technological optimism disconnected from the regulatory and social challenges that remain.

It's this informed vigilance, I believe, that will allow our Western societies to make the most of these innovations while managing their risks responsibly.

I close this letter with the same conviction I opened it with: this small reactor in the Utah desert tells a story much bigger than itself, that of a West refusing to give up its technological lead without a fight.

By Maxime Marquette, columnist

Columnist's transparency note

Who I am and my acknowledged biases

I hold a firm interest in Western technological innovations in the energy and artificial intelligence sectors, with the conviction that the West must maintain its strategic lead in these fields against its international rivals.

This open letter is built exclusively from public corporate statements and verifiable journalistic reporting; I had no access to any confidential technical information from Valar Atomics or Nvidia.

What I don't know and my method

I cannot guarantee that the 30-megawatt data center project will materialize within the envisioned timeline, nor that the regulatory challenges will be resolved quickly. My method consists of cross-checking corporate press releases against several independent journalistic and institutional sources.

Sources

Primary sources

Secondary sources

Get the tech columns

AI, platforms, digital power: the next analyses straight to your inbox.

Cite this article

Maxime Marquette (2026). To those who still doubt the nuclear race for AI. MadMax. https://mad-max.co/en/article/a-ceux-qui-doutent-encore-de-la-course-nucleaire-pour-lia

How does this piece make you feel?
MM
Maxime Marquette
Independent columnist

Maxime Marquette writes most of the analyses and columns published on MadMax — geopolitics, technology, and current events, no filler.

The Newsletter

Enjoyed this piece? Get the next one.

One chronicle a week, straight to your inbox. No noise.

Comments

0 / 2000

Be the first to weigh in.

This article was generated with AI assistance, under human supervision.

Open letter3152 words4 min read