Reliability Growth: FOAK to NOAK

Availability is expected to improve across the first-of-a-kind, next-of-a-kind, and best-of-a-kind builds, but that improvement is a projection until hardware demonstrates it.

Why first units are least reliable

First-of-a-kind (FOAK) machines are where unknown failure modes surface. As designs mature through next-of-a-kind (NOAK) and best-of-a-kind (BOAK) builds, components are refined, maintenance is understood, and availability typically climbs. This reliability growth is a well-observed pattern in complex systems.

Applying it honestly to the burner

The burner has not been built, so its FOAK availability is not measured. The 0.86–0.995 range spans this expected growth. It would be dishonest to quote the top of the range as a delivered figure, and dishonest to ignore that growth is expected. Both statements are true: early units will be less available, and the design intent is to improve.

What must precede any claim

• The breeder (Hyperion) reaches FOAK first tritium ~2030, demonstrating the harder fuel-cycle physics.
• The burner's plug, regime, and fuel gates must be resolved before a burner FOAK exists.
• Availability growth is validated by operating hardware, not by extrapolation.

Reliability growth is a reason for optimism about the mature machine and a reason for caution about the first one. Resilient sovereign power is designed for the availability we can defend today, with growth treated as upside to be proven.

Optimism as upside, not baseline

Reliability growth is a well-attested pattern, but treating projected NOAK or BOAK availability as if it were a delivered figure would be dishonest. The design baseline is what the current machine can defend; growth is upside to be proven in hardware. This keeps the resilient-power case anchored to demonstrated capability while acknowledging the expected trajectory.