Four of our desks arrived at the same finding this week, independently, about four unrelated industries. In each case the number that decides whether a technology is worth building is the one number nobody publishes.

Our cover story is the clearest case. Three rotating-detonation propulsion programmes cleared real hardware milestones in 2026 — a 300-second continuous burn, more than 470 seconds accumulated, over 4,000 pounds of thrust — and not one of them disclosed a pressure-gain figure, which is the entire thermodynamic reason the cycle exists. The same shape appears elsewhere on this slate. Enhanced geothermal publishes days per well, feet per well and megawatts per power block, but not wells per block — so nobody outside the company can compute megawatts per well, and the sector's cost curve rests on two data points. Fusion plant designs are converging on a headline output figure while the fusion power behind it spreads by 41%. And a six-year depreciation life now underpinning a great deal of compute capital cannot be falsified by the market until 2029.

The pattern is not secrecy. It is that operability is easy to report and efficiency requires a baseline. Where we can compute the missing number ourselves, we do, and we label it as ours.

Twelve features and eight briefings from across the network.

— The Ecliptic, Editorial Desk