According the CMDportal database, Banks worldwide have over USD 460 bn of AT1 outstanding, sold as the layer that should take losses early, while a bank can still be turned around. In practice, these instruments have behaved very differently: they sit idle through the recovery phase and only bite once the bank is already at the point of failure. The capital that supervisors label as going‑concern is treated by markets as gone‑concern. A new FSI Brief, “Strengthening the going-concern role of AT1: options and trade-offs”, co‑authored with Rodrigo Coelho, Yvan Lengwiler, Kumar Rishabh and Rastko Vrbaski, asks whether better design can close this gap. The authors argue that the issue is not only low triggers, set far below the levels where stress actually surfaces, but also conversion terms that fail to threaten meaningful dilution: once the share price falls moderately, AT1 conversion stops penalising shareholders and instead rewards them, much like a write-down. That is consistent with pricing: write-down and conversion AT1 from the same issuer trade at almost identical yields. Policymakers therefore face a stark choice: phase out AT1 altogether, or reform it around three pillars — dilutive, market‑linked conversion, automatic rather than discretionary triggers, and higher trigger levels that activate while recovery is still feasible. Phasing out AT1 would forgo one of the few contractual tools that can overcome shareholders’ reluctance to recapitalise under stress, a function that simply adding more CET1 cannot replicate..