Enterprise AI Transformation Programmes (2025–2026)

Narrative

Financial and business press coverage from mid-2025 through mid-2026 converges on a single organising fact: enterprise AI adoption is now near-universal but value realisation remains concentrated in a small cohort. McKinsey's State of AI survey, fielded across nearly 2,000 respondents in 105 countries, found that 88% of organisations report regular AI use in at least one function, yet only around 6% qualify as high performers attributing more than 5% of EBIT to AI, with nearly two-thirds still not scaling AI across the enterprise. MIT NANDA's GenAI Divide report supplied the more viral, and more contested, figure that 95% of generative AI pilots show no measurable P&L impact, a claim that outlets such as Marketing AI Institute pushed back on for its narrow definition of success even as the number spread widely across coverage.

Token economics has become the dominant 2026 storyline in financial press, marking a shift from indiscriminate rollout to active cost governance. The Financial Times reported that Amazon, Walmart, Cisco, Uber and Meta all introduced usage caps in 2026 after AI spending outran projections, with Uber capping individual tool use at $1,500 per month after exhausting its annual AI budget by April. Wall Street Journal reporting, cited across Forbes and other outlets, found OpenAI missing revenue targets amid a sharp rise in data-centre capital expenditure, feeding a broader "economics of AI breaking down" narrative that Morgan Stanley and Goldman Sachs analysts have offered conflicting reads on, with Bain warning many firms are not earning expected productivity gains while Goldman argues consensus AI capex forecasts remain too low.

The shift from seat-based to token-based, consumption-driven pricing has forced FinOps practices into finance functions that previously had no playbook for volatile, agent-driven cost structures; Deloitte's tokenomics research (originally published in the Wall Street Journal's CIO Journal) frames tokens as "the new currency" and finds unit prices falling even as total spend rises because of surging consumption from agentic workloads. Regulatory coverage centres on the EU AI Act's phased 2 August 2026 deadline for transparency and high-risk obligations, with the European Commission's Digital Omnibus provisionally deferring some high-risk deadlines by roughly sixteen months while leaving general-purpose AI and Article 50 transparency obligations unchanged, creating a compliance trap for firms that assume blanket relief. UK and US financial regulators, meanwhile, are extending existing model-risk frameworks such as the PRA's SS1/23 and the Federal Reserve's new SR 26-2 to cover AI and machine-learning models, reflecting how governance intensity scales with sector risk tolerance rather than arriving as a single new AI-specific regime.

Bloomberg's own coverage, including its Bloomberg Intelligence agentic AI outlook and CTO Shawn Edwards' account of building the AskB terminal feature, illustrates how financial-sector practitioners are treating evaluation-driven development as the binding constraint on production-grade agents, a theme echoed in enterprise surveys from WRITER and Zapier showing human-in-the-loop governance persisting as the default operating model even as agent deployment accelerates. Across this coverage, independent academic work on the AI productivity paradox and NBER's large-scale executive survey finding 90% of firms report no measurable productivity effect after three years point to a genuine, still-unresolved gap between task-level gains and enterprise-level financial impact, a gap financial press increasingly treats as the central unresolved question of the AI transformation cycle.