J.P. Morgan projects that generative AI will need to enhance IT services workloads by more than 55% to offset the deflationary effects on project costs, according to a recent analysis from the brokerage.
Analysts estimate that automating coding tasks with large language models could lower project costs by approximately 35% on a comparable basis. This forecast assumes a 90% reduction in routine coding, which currently makes up about 40% of a typical project’s full-time equivalent (FTE) hours.
Further cost reductions are anticipated in documentation and testing, while areas requiring advanced reasoning—like requirement analysis and architecture—are expected to experience less price pressure.
The report highlights that prompt engineering, a new role in the landscape, will likely counterbalance some of the deflation, making up 15% of pre-AI project costs or 23% of post-AI project expenses. Additionally, code review is expected to double in importance due to the need for oversight of AI-generated code.
Debugging, which generative AI can both streamline and complicate, is projected to remain cost-neutral. J.P. Morgan’s calculations indicate that if prices decrease by 35%, a minimum 55% increase in workload is necessary to maintain net revenue levels.
The report outlines several strategies to meet this requirement, such as a 25% rise in both project volume and complexity. Analysts believe the adoption of generative AI is on course to surpass this 55% workload increase, pointing to historical trends where companies reinvested savings from earlier technology shifts—like cloud computing and offshore delivery—into new initiatives.
Early indicators, such as a 3% revenue increase reported by Genpact (NYSE:G) among clients using agentic solutions, suggest a similar pattern of reinvestment is emerging.
J.P. Morgan also noted a shift in the makeup of IT work. Before AI, coding represented 40% of project costs; after AI adoption, this figure is expected to fall to 6%. In contrast, prompt engineering, debugging, and code review are set to claim a larger share of the new cost structure, indicating a reallocation of labor rather than a straightforward reduction.
