Deloitte Report: AI in Finance from Pilot to Scale, "Uncharted Edges" Still Await Breakthroughs

Industry Background: AI in Finance Moves from "Pilot Purgatory" to the Tipping Point of Scale

The narrative around artificial intelligence in the financial sector is shifting rapidly. According to Deloitte's latest report, *The State of AI in the Enterprise*, 74% of financial institutions plan to deploy AI agents with multi-step reasoning and autonomous action capabilities within the next two years, but only 21% of enterprises have mature governance models for autonomous agents. This gap highlights the core contradiction facing the financial industry on the path to scaling AI: technological availability is accelerating rapidly, yet the chasm from experimentation to production-grade deployment remains vast.

The report notes that currently only 25% of global leaders can convert more than 40% of their AI experiments into full production environments. However, 54% of enterprises expect to cross this threshold within the next three to six months, signaling that a wave of scaled deployment is imminent.

Market Impact: Technical Debt and Governance Gaps Constrain Value Realization

The long-standing technical debt in finance—legacy banking infrastructure—has become the primary bottleneck for scaling AI. Sandbox environments run on cleansed data in isolated test zones, but real-time transaction deployment demands rigorous security reviews, compliance checks, and real-time monitoring. As one AI strategy leader at a major European bank candidly stated in the report: "Many organizations built infrastructure and governance systems for traditional AI models, but the emergence of large language models has completely disrupted those efforts. Now 80% to 90% of new use cases are generative AI, requiring an entirely new set of capabilities."

This mismatch has led to a clear performance divergence in the market: 37% of companies use AI on the surface without altering any underlying processes; another 30% are optimizing existing processes but not touching the business model; only 34% of market leaders leverage AI to deeply transform entire business models, core workflows, and product portfolios.

Competitive Landscape: Efficiency Enhancers vs. Model Disruptors

In the fintech space, two types of players are diverging: those using AI for automation and efficiency improvements (e.g., data entry, reconciliation, tier-1 customer support), and those leveraging AI to reshape core financial workflows. The report indicates that 36% of leaders expect at least 10% of operational roles to be fully automated within a year. This shift is fueling a new wave of competition—traditional banks that fail to quickly move beyond the pilot stage will be left behind by agile fintech companies and deep AI transformers.

Geopolitics is also reshaping procurement behavior: 83% of enterprises place data residency and local computing parameters at the core of strategic planning; 77% include the country of origin of AI tools in supplier selection; 58% prioritize building technology stacks with local vendors. "Sovereign AI"—designing, training, and deploying models on locally controlled infrastructure using locally governed data—has become a boardroom reality.

Enterprise Implications: Building AI-Native Capabilities, Balancing Automation and Talent InvestmentDeloitte's global AI leader Nitin Mittal emphasized: "We see enterprises with great AI ambitions, shifting from experimentation to embedding AI into the core of business, focusing on scale and impact. Leaders should release enterprise value by consciously weaving AI into business workflows and better coupling human-machine intelligence."

• For financial institutions, the most urgent tasks at present are:
• Establish mature AI governance models, especially risk management frameworks for autonomous agents, to address regulatory compliance and data privacy challenges (73% of enterprises rank data privacy and security as the top AI risk, and 50% worry about legal, intellectual property, and compliance issues).
• Invest in technology infrastructure, especially the new-generation computing power and data management systems that support generative AI.
• Reshape the talent model: Jim Rowan, Deloitte's US AI leader, pointed out, "Successful enterprises not only invest in automation and algorithms but also in talent. AI is inspiring new ways of working. This dual-focus strategy—simultaneously enhancing both talent and AI tool capabilities—enables teams to embrace reshaped business models and lay the foundation for competitive advantage."

Future Outlook: Key transformation window in the next 12-24 months

• Next 12 months: More AI agent pilots are expected, but only enterprises with mature governance can safely deploy them in core processes. Fintech companies with lighter technical debt may be the first to scale.
• Next 24 months: Autonomous AI agents will accelerate deployment in areas such as customer service, compliance monitoring, and trade execution. The trend of sovereign AI will further intensify, and regional competition for AI infrastructure will increase.
• Next 3 years: AI will be deeply embedded into core workflows of the financial industry. Traditional banks' "IT core systems" may be replaced by AI-native platforms. Enterprises that fail to establish governance and talent systems will face significant competitive disadvantages.

Deloitte's report reveals a clear signal: the AI race in the financial industry has shifted from "whether to adopt" to "how to scale." Enterprises that can cross the pilot trap and balance innovation with risk will seize the advantage in the next cycle.