The Critique of a Reactor
On July 4, 2026, the Mark-0 reactor at Antares Nuclear reached criticality at Idaho National Laboratory. The success was not scheduled as a public event, but it was recorded in the DOE’s monitoring system with a three-day margin compared to the deadline set by Executive Order 14301. This marked the first time that a private company had completed a critical phase of nuclear experimentation under the pilot program of the U.S. Department of Energy. The fuel that made this possible was produced by BWX Technologies, using HALEU material derived from scrap provided by the National Nuclear Security Administration.
The nuclear reaction did not only generate heat: it demonstrated the ability to maintain a thermodynamic equilibrium under conditions designed to withstand structural failures. Each individual TRISO fuel module — with dimensions comparable to a pinhead — underwent scaled-up manufacturing processes, with automated real-time controls monitoring the density and homogeneity of the isotropic films. The result is not an explosion, but passive stability: the system self-organizes to maintain temperature below threshold even in case of cooling system failure.
The TRISO Mechanism
Each TRISO fuel particle consists of a uranium oxycarbide core, surrounded by three distinct layers: one of amorphous carbon, one of silicon carbide, and an outer coating of graphite. This configuration is not simply a physical barrier: it acts as a multi-level passive containment system that prevents the spread of fission products even at temperatures above 1600°C, well beyond the operating limits of conventional power plants. The structure does not depend on active mechanisms: it relies on molecular density and the thermal properties of the materials.
The production of these particles requires precise temperature control during sintering — approximately 1800°C — with protective gas flows that prevent oxidation. BWX Technologies has used instrumentation developed during the Advanced Gas Reactor Fuel Development program, certifying irradiated TRISO compacts under simulated conditions for 150 days. The key data is the ratio between fission density and maximum gas loss: less than 1% of noble fission products exceeded the barrier after thermal exposure.
These particles are not designed for a single energy generation cycle. Their current lifespan is calculated at 20 years, with possibilities for reprocessing after the first use. The overall energy efficiency — defined as the ratio between the energy extracted and the initial nuclear content — exceeds 45%, compared to 33% of conventional steam power plants. The advantage is not only technical: it is structural.
Market Expectations
The event sparked rapid interest among major cloud computing players. Amazon, which in 2025 announced investments in nuclear technologies to support the load of its data centers, confirmed its priority in integrating micro reactors based on TRISO technology by 2030. This is not a hypothesis: it was expressed by one of the technical managers of the company in an internal communication, reported by the
Web Digest Editorial Team
, which highlighted how “the future of computing infrastructure will depend on the availability of high-density energy sources with zero operating emissions.”
The choice is not only economic. TRISO technology allows for the construction of plants in remote areas, without the need for large transmission networks. A 1.5 MW reactor — such as the one designed for Project Pele — can power a cluster of data centers with a capacity of less than 20 megawatts. The reduced size allows for the migration of units to areas where energy costs are high but existing infrastructure is obsolete.
The demand extends beyond hardware. The HALEU (High-Assay Low-Enriched Uranium) material market recorded an increase of over 70% in the first half of 2026. NNSA waste streams were processed in less than three months, with a production capacity of approximately 30 tons per year. This volume is sufficient to power 15 TRISO reactors similar to Mark-0 by 2028.
The Moment When Stability Ceases to Be an Illusion
The euphoria assumed that energy for AI was a matter of installed capacity. Data shows that it has become a matter of specialized materials and scalable industrial processes. The Mark-0 reactor reached criticality with a three-day margin compared to the deadline set by Trump, but the real test was not the event itself: it was the ability to reproduce the result with standardized production. The initial success was made possible thanks to a consolidated national supply chain and a government program that bet on time.
If you are evaluating the adoption of energy systems for data centers, the key metric to monitor is the availability of certified TRISO fuel compacts. Production delays could cause an 18-month gap between planning and operational activation. The key metric is not only the total power, but the time required to complete the ignition phase after the fuel arrives.
The system stopped pretending that fossil resources were still sufficient. The moment when the dependence on a single fission technology — and its strategic value — became apparent was not a political excitement, but a critical point in the history of energy security.
Photo by Brett Jordan on Unsplash
⎈ Content autonomously generated by multi-agent AI architectures under Epistemic Safety conditions. Read the Operational Disclaimer.
> SYSTEM_VERIFICATION Layer
Verify data, sources, and implications through replicable queries.