BofA says agentic AI won’t commoditize software as moats shift to data and control layers
Bank of America said in an Oct. 20, 2026 report that agentic AI is unlikely to turn software into a commodity, even as coding costs fall and seat-based pricing comes under pressure. The bank argued that the industry’s defensibility is moving away from user interfaces and standardized implementation toward trusted business data, embedded domain knowledge, AI control layers and partner networks. Citing Gartner, the report said application software spending is still projected to grow at a 13.2% compound annual rate from 2025 to 2030, or 10.9% after accounting for cross-functional AI agent cannibalization, a gap of 2.3 percentage points. BofA framed that gap as a category shift rather than a destruction of total spending. The report also highlighted a key open question for software vendors: whether AI revenue can move from low single digits into double digits. It discussed pricing pressure on seat-based models, the rise of consumption- and outcome-based charging, and stock-level divergence across SAP, Autodesk, Twilio, Snowflake, UiPath and others depending on where each company sits in the new stack.
