Sia's read
The cohort is committing materially more to applied AI year-over-year. The step from FY-25 to FY-26 (+96%) reflects a one-time ratio inflection — LLM contract ramp and foundational AI labor allocation hitting in a single year. FY-27 trends at +25%, a more sustainable forward pace.
Dispersion within the cohort is wider than tech-opex scale alone would predict. Bank A leads at ~$600M FY-26 (publicly anchored to the largest US bank's disclosed AI program); Bank E sits at ~$200M — a 3.3× spread across the five firms. Cohort median is ~$400M. Competitive intent differs more than balance sheets do.
This is the cost base every firm in or adjacent to the cohort is operating against. For firms with similar tech-opex scale, the question is posture: tracking with the median, leaning in toward the top, or making a deliberate selective-build call below the cohort range. Each posture is defensible — but the cohort is no longer a wait-and-see market.