Summary: Purpose: analyze the US push to revive nuclear and extend coal under the banner of AI demand, examine the politics and economics, and map the likely winners and losers. This post untangles the claims, repeats the hard facts, and asks the sharp questions leaders and investors should be asking now.
Interrupt — Engage: The White House has put a spotlight on nuclear and handed coal a short reprieve. The message is simple and loud: AI needs massive, steady power, and government must pick the sources. You hear the promise of “second chances” and “generational restitution.” You also hear lines about restarts, big loans, and tech giants signing long-term purchase deals. What do those moves buy you? What do they cost? Let’s trace the threads and test the claims.
Where we came from: past bailouts and failed bets
The story begins with the 2017 subsidy push for coal and nuclear. That attempt was blunt: protect plants losing money to cheap gas and renewables. Taxpayers would have covered a lot. The plan failed. Since 2020, three nuclear plants have shut and several new-build projects stalled after a decade and billions spent. Coal’s share of the US power mix dropped from around 45 percent in 2010 to 17 percent today. That decline didn’t appear because regulators were mean—it happened because capital, technology, and markets moved elsewhere.
What changed in 2025: AI becomes the energy argument
The current moment is different because AI gives political leaders and industry a headline that sells. In May 2025, the administration used executive orders to declare nuclear a priority for AI workloads, even ordering 10 large reactors by 2030 and reshaping the nuclear regulator. The Department of Energy created a pilot program that has helped small startups progress faster. Energy Secretary Chris Wright said AI’s progress “will be accelerated by rapidly unlocking and deploying commercial nuclear power.” That’s a clear framing move: nuclear is pitched as strategic infrastructure for technology leadership. It’s an emotional and political argument as much as a technical one.
Tech money, public support, and the optics
Large tech firms have jumped in on the narrative. Google, Amazon, and Microsoft signed deals to buy nuclear power for data centers. Microsoft joined the World Nuclear Association and backed the high-profile restart of Three Mile Island, supported by a $1 billion federal loan. Retired reactors are being eyed for restarts; some companies even support reopening plants that closed recently. Public approval of nuclear is the highest since 2010. That’s social proof: companies in the spotlight, federal backing, and warmer public sentiment together create a story people can rally behind.
Mirror: second chances. Generational restitution. Big promises.
You want “second chances.” You want “generational restitution.” You want power fast and reliable. That’s understandable. It also raises immediate questions: how do you square promises of reactors ready in a few years with the reality of long, costly construction? How do you square large federal deals and sudden valuations with concerns about political influence and overpricing? Those are practical, not partisan, questions.
Construction costs, timelines, and the physics of building plants
Most nuclear costs come from building the plants, not from regulation per se. Construction budgets balloon, timelines stretch, and contractors encounter supply chains and labor limits. Critics warn of “juiced-up valuations” for small modular reactor firms, especially those close to political power. The government’s $80 billion deal with Westinghouse in October lacks public detail on deliverables and timelines. Big contracts without clear milestones create uncertainty, not certainty. No amount of executive will can change construction physics overnight: labor, materials, testing, certification, and grid integration take time.
Coal gets a stay of execution, but not a clean bill of health
In April, the administration also used executive powers to extend the life of some coal units and ease pollution rules. Two plants slated to retire were ordered to stay online. More than two dozen generating units that were scheduled to retire are now staying in service, getting multi-year reprieves. Yet here’s the mirror again: coal’s public image is poor. Tech companies want to avoid direct ties to coal. No major cloud provider has declared plans to buy coal power. Instead, many utilities that are cutting coal are looking at nuclear as replacement capacity. Coal’s lifeline is real but fragile—political, not commercial—and may be temporary.
The carbon question and investor reality
Some firms still explore carbon capture for coal, but high-profile failures make that a risky bet. Investors care about emissions. Brett Rampal, a nuclear adviser, said nuclear has been “the neglected and unjustly villainized hero of the energy world” and that 2025 felt like “generational restitution.” That captures the mood among proponents. But Rampal and others also acknowledge that emissions matter for investor decisions. The market penalizes dirty power where customers and lenders care.
Market forces pushing back: cheap renewables and storage
Let’s not pretend policy trumps market fundamentals. Utility-scale solar and onshore wind remain among the lowest-cost sources of power per megawatt-hour. Battery storage is improving and pairing renewables with storage changes the calculus for grid reliability. Meanwhile, other countries keep building renewables at scale. China has driven down emissions recently by accelerating renewables. Even if Washington prioritizes nuclear and extends coal, the market will still favor low-cost renewables where permitting, land, and supply chains permit rapid deployment.
Tech’s dilemma: clean image versus reliable loads
Tech firms face a conflict: they need low-carbon, reliable power for AI datacenters, but their public brand depends on climate commitments. Nuclear fits the low-carbon bill; coal does not. That’s why we see Microsoft and others leaning into nuclear deals and avoiding public coal partnerships. The optics matter: being powered by coal would clash with marketing and investor relations. So tech money naturally flows to nuclear or gas-plus-options—less to coal.
Politics, valuations, and the risk of capture
There are real governance risks. Rapid federal spending, big loans, and regulatory reshuffles can create favoritism and inflated valuations. When private firms benefit from political alignment, scrutiny follows. Investors and citizens should ask for transparency: what are the contracts, performance milestones, cost caps, and accountability measures? How will taxpayers be protected if projects fail or get delayed? Those are not rhetorical questions; they demand clear answers.
Global competition: energy policy as a tool for AI leadership
If the goal is to beat China on AI, energy policy matters. AI scale depends on compute, and compute depends on power. The US playbook here has to balance speed, cost, and climate credibility. China’s recent emissions drop came from rapid renewable builds. If Washington bets on older technologies that are expensive or slow to deploy, it risks lagging where speed of deployment counts. Policymakers should ask: do our choices increase national competitiveness or slow it down?
What success looks like — and what failure looks like
Success will be modular, measurable, and mixed. Nuclear can contribute if projects run to clear, enforceable milestones and if the industry proves it can build on predictable schedules and costs. Coal can provide temporary capacity in emergencies, but long-term reliance would harm climate goals and public trust. Failure looks like large federal outlays with no concrete delivery, inflated valuations, and delayed reactors that miss 2030 targets while renewables and storage undercut the economics.
Questions to push policymakers and CEOs — calibrated, not combative
How will timelines for new reactors be independently verified? How will loans be structured to protect taxpayers if projects slip? What clear metrics will decide whether retired plants should be restarted? What contingency plans exist if AI compute demand grows faster than new capacity? Who will enforce cost controls and anti-corruption safeguards? These are open-ended questions that force clarity rather than cheerleading.
Labeling the emotion: there’s optimism, relief, and a dash of triumphalism. There’s also skepticism, and rightly so. Ask yourself which feeling you trust. No single answer fits every stakeholder. Communities near plants want jobs and safe operations. Investors want predictable returns. Tech CEOs want reliable, low-carbon power without reputational damage. Policymakers want a competitive edge. Balancing those needs is the hard part.
Practical moves for leaders who want results
– Make contracts public and milestone-driven. Tie payments to construction, testing, and grid delivery milestones.
– Require independent audits of cost estimates and supply chains.
– Use pilot programs to validate small modular reactors at scale before ordering dozens of large units.
– Protect ratepayers and taxpayers with risk-sharing mechanisms that limit one-sided upside for contractors.
– Pair nuclear investment with accelerated permitting for renewables and storage to keep options open.
Closing: the trade-offs and the choices
You can cheer the surge in nuclear investment and welcome the tech sector’s involvement. You can also say No to opaque deals and No to expecting instant fixes. The policy moves of 2025 open a second act for nuclear and give coal a breath of life. That’s real. But markets, physics, and public opinion remain. If you want power for AI and a shot at climate goals, chase a mixed strategy that enforces accountability, rewards on-time delivery, and keeps low-cost renewables and storage in the plan. That’s how you turn promises into delivered capacity that lasts.
What should government prioritize next: rapid scale of proven technologies, or concentrated bets on large, politically backed projects? Which risks are worth taking to secure compute for AI? Which promises need clearer evidence before taxpayers sign on? Mirror back your concerns: “I want reliable power, soon.” Then ask: how fast can policy and markets deliver it without costly surprises? Let’s keep the conversation practical and hard-edged.
#EnergyPolicy #Nuclear #AI #Coal #DataCenters #PowerPlay #GridStrategy #CleanBaseload
Featured Image courtesy of Unsplash and Nicolas HIPPERT (C82jAEQkfE0)
