The electric grid isn’t just infrastructure; it is civilization itself—a vast, century-old machine now cracking under new pressures from AI, electrified transport, and reindustrialization. The core problem isn’t a lack of new power generation but a crippling bottleneck in delivery. This conversation with Doug Bernauer of Radiant and Drew Beglino of Heron explores two radical technological responses: portable, factory-built micro-nuclear reactors and solid-state power electronics designed to rebuild the grid from the edges outward. They argue that we can no longer rely on the centralized, top-down model of the past. Instead, the future lies in a decentralized, software-defined, and modular architecture where power can be generated and managed locally, turning the grid into a flexible, resilient network that can grow organically to meet explosive new demand.

Bernauer’s vision stems from designing a Mars colony at SpaceX, where he realized solar power was insufficient and was directed by Elon Musk to look at nuclear. His company, Radiant, is creating a trailer-sized, one-megawatt micro-reactor that can be factory-built, shipped anywhere, and powered on within 48 hours, offering a clean, resilient alternative to diesel generators for remote military bases, disaster relief, or islands. Crucially, it uses meltdown-proof fuel and is designed as a product with a service model—deployed, operated for years, and then removed, aiming to make nuclear a viable, on-demand option rather than a multi-decade construction project.

Beglino’s work at Heron addresses the grid’s physical constraints with solid-state transformers. Replacing the giant, oil-cooled transformers that have defined the grid for a century, his company’s devices use power semiconductors and software to convert voltage at high frequencies. This makes them smaller, more efficient, and software-defined, enabling dynamic control and bidirectional power flow. His first product converts DC power from solar, batteries, or data centers to the grid’s AC voltage, but the broader vision is to enable a grid that can intelligently integrate diverse, distributed energy sources, manage spiking loads from massive data centers, and even facilitate a future DC-based microgrid architecture.

Both founders emphasize a philosophy of manufacturing honed at Tesla and SpaceX: solving the delivery problem through modularity, factory production, and extreme supply chain focus. They see a symbiotic relationship between their technologies—where portable nuclear provides steady, off-grid baseload power, and advanced power electronics enable that power to be integrated and managed intelligently within modern, DC-native systems that also include solar panels and batteries. This represents a fundamental shift from viewing the grid as a static, centralized utility to seeing it as a dynamic, upgradable platform.

Surprising Insights

• Nuclear waste might be safer above ground: Doug Bernauer posits that leaving uranium in the ground is arguably more dangerous from a health perspective, as it naturally decays into radon, whereas using it in a modern reactor captures its energy and allows for controlled containment.
• Data centers could lower electricity rates for everyone: Drew Beglino challenges the common narrative, arguing that because data centers consume power at a near-constant, high rate, they increase the overall utilization of the grid infrastructure, which can spread fixed costs over more kilowatt-hours and potentially reduce average rates.
• The power electronics market could rival the scale of the entire grid: While the U.S. grid’s peak power is under a terawatt, Beglino notes that the supporting power electronics conversion needed at every stage (generation, transmission, distribution) could represent a market several times larger—around 800 gigawatts for Texas alone.
• The “nuclear industry” barely exists yet: Bernauer provocatively states that a true, innovative nuclear industry is only just being born, with multiple startups aiming to achieve “criticality” with new designs for the first time in the coming year, moving from concept to tangible product.
• The biggest grid constraint isn’t technical but logistical: Decades of flat demand led to a hollowing out of engineering talent and supply chains for grid hardware. The current bottleneck is less about a lack of brilliant ideas and more about rebuilding the domestic manufacturing and expertise to execute at scale.

Practical Takeaways

• Rethink grid architecture from the edge: Instead of waiting for centralized grid upgrades, consider how decentralized generation (like solar plus storage) and new interconnection technologies can create resilient microgrids for critical facilities, from data centers to factories.
• Embrace DC where possible: Modern loads like data centers, batteries, solar panels, and even new micro-reactors are natively DC. Designing local power systems to use DC internally can reduce conversion losses and complexity before interfacing with the main AC grid.
• Prioritize manufacturability and modularity for infrastructure: Whether for power plants or grid hardware, the factory-based, productized approach—focused on modular design, automation, and low on-site construction—leads to faster deployment, higher quality, lower cost, and easier permitting.
• Use software to turn grid loads into assets: Data centers and other large loads should invest in grid-interactive technologies that allow them to stay online during disturbances and even provide grid-stabilizing services, transforming them from a stability risk into a resilience asset.
• Advocate for regulatory modernization: Support policies that accelerate the deployment of new energy technologies, from streamlined permitting for micro-reactors to updated standards that allow software-defined power electronics to replace electromechanical grid equipment.