JPMorgan Chase is raising the bar for its software engineers—and the message couldn’t be clearer. Internal documents reviewed by Business Insider show the bank has rolled out new performance objectives for its global developer workforce, the bulk of its 65,000-strong Global Technology division. The updated goals, published on the company’s intranet earlier this month, require all software and security engineers to “drive excellence” by adopting AI—demonstrating measurable gains in code quality, speed, and productivity. Changes are set to reflect in employees’ official goals by the end of March.
The bank isn’t just nudging people. Five engineers told Business Insider that AI adoption has become a recurring theme in manager conversations, intranet posts, and internal dashboards granular enough to show whether someone has installed GitHub Copilot—and how often they’re actually using it.
JPMorgan’s $20 billion tech bet comes with strings attached for its engineers
JPMorgan is projected to spend roughly $20 billion on technology in 2026, dwarfing peers like Goldman Sachs. That investment now comes with direct expectations for its workforce. One engineer told Business Insider their manager said the rollout of new AI tools carries an “expectation” that output should show “a noticeable increase” quarter over quarter.
A pilot of Anthropic’s Claude Code is also set to go live as early as April—joining four other large language models developers already use, two from OpenAI and two from Anthropic’s Claude.
Tracking who’s keeping up—and who isn’t—is now part of the job
The pressure isn’t unique to Wall Street. Google, Meta, Amazon, and Salesforce are all factoring AI use into performance reviews and promotion decisions, according to a Wall Street Journal report.
At JPMorgan, engineers who lag on adoption risk being seen as underperformers. The bank is also overhauling how it grades employees more broadly—moving to two core dimensions: “what you achieve” and “how you achieve it.” Workers will be sorted into three buckets: “stand out,” “achiever,” and “needs improvement.” For developers sitting on the AI sidelines, that last category is looking a lot closer.
