Over the past week, a 40-year-old reactor design was resurrected by ex-SpaceX engineers to power AI data centers. The narrative is seductive: high-energy AI meets zero-carbon baseload. But the real question is not whether the design works. It is whether nuclear can solve the energy crisis for both AI and crypto mining before the hype cycle collapses.
2017 vibes. Proceed with skepticism.
Context: The mPower reactor was originally developed by Babcock & Wilcox in the 2000s, then shelved due to lack of commercial interest. Now, a team of former SpaceX engineers claims to have revived it, targeting the massive power needs of AI data centers. The article describing this revival is a narrative piece, not a technical audit. It provides zero data on reactor type, power output, licensing status, construction timeline, cost, or customer commitments. The only signal is a demand-side story: AI data centers need power, and nuclear is clean and reliable.
But this is exactly the kind of story that crypto miners and infrastructure investors have been chasing for years. The Bitcoin mining industry alone consumes over 100 TWh annually, and layer-2 scaling solutions are shifting computational load to decentralized networks that require always-on servers. The energy demand is real, but the supply side is a mess.
Core: Let’s dissect the mPower revival through the lens of a protocol audit. The four thresholds for any energy infrastructure project are: regulatory approval, engineering replicability, economic viability, and customer commitment. The article fails on all four counts.
Regulatory: The design has not entered NRC review. Without a licensing pathway, the reactor is a paper concept. In crypto terms, this is like a Layer-2 project that has a whitepaper but no testnet. Entropy wins. Always check the fees.
Engineering: The mPower design is from the 2000s. It was shelved for a reason – likely cost overruns or technical immaturity. Reviving it without a demonstration plant or even a pilot project is equivalent to forking a 2017 DeFi protocol and expecting it to handle 2025’s MEV bots. The codebase is old, and the security assumptions have changed.
Economic: No cost data. No LCOE. No PPA. The article does not answer the most important question: can nuclear electricity compete with grid power, gas, or renewables + storage for AI data centers? Based on my audit experience, I have seen protocols that raise billions on a narrative alone, only to crash when the unit economics are exposed. Nuclear is no different.
Customer: The article assumes that AI data centers will pay a premium for nuclear power. But the data center operators I’ve spoken to care about three things: cost, reliability, and speed of deployment. Nuclear fails on speed. The average reactor takes 10-15 years from concept to power. AI data centers are built in 2-3 years. The time mismatch is a systemic risk that the article ignores entirely.
Impermanent loss is real. Do your math. In this case, the impermanent loss is the gap between the hype cycle and the actual construction timeline.
Contrarian Angle: The real blind spot is not the reactor design, but the assumption that AI and crypto energy demand will remain concentrated in locations suitable for nuclear. Both industries are mobile. AI workloads can be moved to regions with cheap hydropower or natural gas. Crypto miners already relocate based on energy prices. Nuclear plants are fixed, require massive water sources, and face local opposition. The revival of mPower is a signal that the market is desperate for a clean baseload narrative, but it is not a solution.
Furthermore, the article uses the "ex-SpaceX engineer" label as a credibility heuristic. In nuclear, engineering credibility is not about rocketry. It is about nuclear safety culture, regulatory experience, and long-term liability. SpaceX engineers have built rockets, not reactors. The cognitive dissonance is similar to a DeFi project hiring a frontend developer to write a smart contract – it may work, but the risk of catastrophic failure is high.
Takeaway: The mPower revival is a narrative play, not a technical breakthrough. For crypto and AI infrastructure investors, the near-term answer is not nuclear. It is grid optimization, gas peakers, and renewables with storage. The nuclear option will remain a paper concept until it passes the four thresholds. Until then, proceed with skepticism. The 2017 vibes are strong, but the math doesn’t lie.