Bank of America AI spending is entering a larger operating phase after Chief Executive Brian Moynihan said the bank expects to double its AI expense budget in 2027. The increase follows a year in which the bank expanded employee access, deployed more AI-supported workflows and tied more of the program to measurable productivity and service results.
Key Takeaways
- Bank of America expects to double its AI expense budget in 2027
- Moynihan said roughly 130 to 140 implemented AI and machine-learning uses cost about $400 million and generated about $800 million in company-measured benefit
- Bank of America reported about 200,000 active AI users and more than 400,000 prompts per day in its second-quarter 2026 materials
- More than 18,000 customer service representatives use EricaAssist, according to the bank
The shift matters because Bank of America is treating AI as an enterprise operating issue rather than a narrow technology project. Its disclosures now connect AI with software development, customer service, employee training, workflow design and governance. Recent executive remarks also show that management is focusing on how quickly AI spending produces measurable benefits.
Bank of America’s second-quarter 2026 materials showed about 200,000 active users of general-purpose AI productivity tools generating more than 400,000 prompts a day. The bank also reported more than 2,000 departmental AI templates, more than 300 approved AI and machine-learning use cases, 114 live generative AI use cases and 34 fully deployed generative AI uses.
Bank of America AI Budget Puts Execution at the Center
Bank of America grouped its AI uses into customer-facing experiences, front-line productivity, workflow transformation and workforce efficiency. The presentation also distinguished general-purpose productivity tools from role-specific applications and multi-process systems embedded in workflows.
That structure matters because the bank is measuring adoption at several levels rather than treating every AI tool as the same type of deployment or stage of operational maturity.
At the September 14 Barclays Global Financial Services Conference, Moynihan said the bank had roughly 130 to 140 implemented AI and machine-learning uses. He said those uses had cost about $400 million and generated about $800 million in benefit, figures that reflect the bank’s own measurement of the program.
Moynihan also said Bank of America planned to double its AI expense budget in 2027. The disclosure puts a clearer financial marker on a program that had previously been described largely through adoption counts, productivity measures and client-service applications.
Management has not disclosed a single dollar figure for the total 2027 AI budget.
One of the bank’s most detailed productivity disclosures involves software development. Its second-quarter materials said about 19,000 developers were using real-time coding assistance and that productivity had increased by more than 20%.
More recent executive comments have referred to about 20,000 developers using coding agents, with productivity gains running in the 15% to 20% range.
The figures were disclosed at different points in the rollout and use slightly different measures, so they should not be treated as identical benchmarks. Together, they show that software development is an area where the bank is tracking AI-supported productivity with specific operating metrics.
Leadership Roles Expand as AI Reaches More Teams
Co-President Jim DeMare said on September 23 that AI tools had reached roughly 90% to 95% of the company, giving employees greater familiarity with the technology.
“Then it is about general productivity, and how is it improving their life at work and how is it improving workflows,” he said.
That emphasis on adoption has also reshaped leadership responsibilities. Jeffrey Busconi, head of Corporate Strategy and Operational Excellence, serves as the bank’s AI catalyst, with responsibility for accelerating AI usage and value across the company.
Bank of America’s 2025 leadership changes also placed continued expansion of AI-based tools and client innovation within the responsibilities of Co-Presidents DeMare and Dean Athanasia.
The structure places AI oversight closer to executive decision-making. Similar AI leadership roles have appeared elsewhere in banking as institutions connect technical deployment with operating efficiency, risk management and customer experience.
Training is part of that framework. Bank of America reported more than 600,000 completed AI training courses by the end of the second quarter.
It also said more than 20,000 wealth professionals were using AI to support meeting-related activities, while thousands of other employees were using role-specific tools for research, presentations and client-service tasks.
Client Service Expansion Brings Governance Into Focus
Bank of America has also expanded AI directly inside customer-service workflows. In July, the bank added generative AI capabilities to EricaAssist, a human-assisted system used by more than 18,000 customer service representatives.
EricaAssist summarizes why a client is calling, surfaces relevant information and recommends next steps while the employee remains responsible for the conversation.
Bank of America says the system can provide contextual guidance in under three seconds and has reduced average call times by nearly one minute per interaction. Those are company-reported operating measures and may vary by workflow.
The deployment builds on Erica, the bank’s consumer-facing virtual financial assistant. Bank of America said in March 2026 that 20.6 million users had interacted with Erica nearly 700 million times during 2025, taking cumulative interactions above 3.2 billion.
The growing use of AI in online banking has also increased attention on controls. Bank of America’s 2025 annual report says expanded AI use, including autonomous agents and third-party services, can create added risks involving data, system behavior, inaccurate outputs, cybersecurity, legal obligations and regulatory requirements.
Moynihan addressed that tension in September when discussing autonomous agents and AI-generated output.
“For our sense, we’re trying to control the risk,” he said, describing human checks and accountability as part of the bank’s approach.
The bank’s Code of Conduct was updated in January 2026 to address responsible AI use and employee responsibilities around approved and non-approved applications. That places AI behavior within formal company standards rather than treating it solely as technical guidance.
On September 22, Bank of America joined five other banks in publishing principles for agentic commerce, where AI agents may take a larger role in selecting and paying for goods and services. The framework centers on transparency, safety, privacy and data, choice and interoperability.
Bank of America said the principles are intended to support trusted use of AI agents in commerce. For Bank of America AI governance, the framework adds an external-facing set of principles alongside internal controls described in the bank’s Code of Conduct and annual filing.
Frequently Asked Questions
How much will Bank of America spend on AI in 2027?
Bank of America has not disclosed a single total dollar figure for its 2027 AI expense budget. Moynihan said in September 2026 that the bank expects the budget to double next year.
How widely are AI tools used at Bank of America?
Bank of America reported about 200,000 active users of general-purpose AI productivity tools in its second-quarter materials. DeMare later said AI tools had reached roughly 90% to 95% of the company.
What is EricaAssist?
EricaAssist is a human-assisted AI system used by more than 18,000 Bank of America customer service representatives. The bank says the system can provide contextual guidance in under three seconds and has reduced average call times by nearly one minute per interaction.
How does Bank of America manage AI governance?
Bank of America AI governance includes formal employee standards, human oversight and controls addressing risks outlined in the company’s annual filing. The bank also participated in the September 2026 principles for agentic commerce covering transparency, safety, privacy, choice and interoperability.
