CIB PERFORMANCE AND OUTLOOK
The first half of 2025 confirmed a fragile economic environment. The OCDE projects global growth at 2.9% this year (after 3.1% in 2024), but trajectories diverge: a US rebound (+3.8% in Q2), modest recovery in Europe (+1.6%), limited growth in the UK (+0.3% in Q2), and emerging markets under pressure from new US tariffs (effective average rate at 18.6%, the highest since 1933). Against this backdrop, capital markets revenues rose 13% YoY, but performance dispersion across banks remains pronounced. Structurally, revenue and cost dynamics across geographies have followed similar cycles since FY2023, downturn, recovery, and acceleration, but the amplitudes differ: US banks achieved a steeper rebound in revenues, while European peers exhibited greater volatility in cost efficiency. UK banks remain closer to the European profile.
On the non-financial side, ESG integration accelerates: According to our reserach, European banks score 36% higher than US peers, with reporting converging toward IFRS S1/S2 despite delays, while investors demand integrated ESG-financial data and deploy GenAI tools to assess double materiality.
From AI to Agentic AI in Banking and CIBs
AI investments topped $250bn as of 2024YE and could reach $1.2tn by 2030, with banking alone attracting $26bn in 2024, set to quintuple by 2030. In CIBs, most use cases focus on cost savings, cutting up to 25% in compliance, operations, and customer service costs, though ROI on revenues remains unclear.
Generative AI is scaling fast: BNP Paribas targets 1,000 use cases in production by 2025YE (c.€500m in value), Standard Chartered has deployed SC GPT to 70,000 staff in 41 markets, and JPMorgan runs 300+ AI use cases within an $18bn tech budget. Goldman Sachs's Marquee shows how advanced analytics and execution can be monetized, paving the way for agentic AI.
This new frontier promises autonomous decision-making and deeper transformation but also raises risks around trust and compliance, with Gartner warning that 40% of projects may be cancelled by 2027. Success will hinge on robust data governance, cross-functional collaboration, and strong talent strategies, as people remain central to scaling AI effectively.
