Big Tech Is Now Grading You on AI Usage — How Meta, Google, and JPMorgan Tied AI to Your Raise

JPMorgan labels you "light," "heavy," or "non-user." Meta tracks your AI-assisted code percentage. Google lets managers mandate AI adoption. Not using AI is now a performance issue.

April 5, 2026 · 8 min read · By Connie

The Shift: AI Adoption Is Now a Job Requirement

For most of 2024 and early 2025, using AI tools at work was encouraged but optional. By early 2026, that has changed at some of the world's largest employers. A detailed Business Insider investigation confirmed that Meta, Google, JPMorgan Chase, Amazon, and Salesforce have all moved to formally include AI tool usage in employee performance evaluations.

The shift is driven by economics. Companies collectively invested hundreds of billions of dollars in AI infrastructure, tools, and licensing in 2024 and 2025. Executives are under pressure to demonstrate returns. One measurable proxy for AI ROI is adoption: are employees actually using the tools we bought? And performance reviews — the mechanism that drives raises, promotions, and career trajectory — are the most powerful lever companies have to force behavior change.

As Mark Zuckerberg told Meta investors in January 2026: "2026 is going to be the year that AI starts to dramatically change the way that we work." Performance reviews are where that change is being operationalized.

Company by Company: What They're Doing

JPMorgan Chase

JPMorgan has implemented the most systematic AI adoption tracking program reported so far. The bank's internal dashboards classify its approximately 65,000 engineers and technologists into three tiers: light users, heavy users, and non-users of AI tools.

New performance goals for the Global Technology team require engineers to "drive excellence" by demonstrating better code quality and higher productivity through AI. The bank has updated its grading system to evaluate employees on both "what you achieve" and "how you achieve it" — with AI usage being a critical component of the latter category.

Meta

Meta has established specific performance goals for engineers that include hitting a target percentage of AI-assisted code. The company formed AI "pods" and launched "Transformation" weeks — dedicated periods of workshops and experiments with AI tools including Claude Code — to drive adoption.

Failing to hit AI adoption targets at Meta can affect your performance rating, which directly determines your annual bonus and eligibility for promotion. Meta CEO Mark Zuckerberg has publicly stated that he personally returned to coding using AI tools, partly to model the behavior he expects from Meta's 70,000+ employees.

Google

Google's approach gives managers direct authority: managers can mandate AI assistant use for their teams. For engineers, this means using AI for coding. For non-technical staff, it means using AI for strategy documents, sales call analysis, and customer insights. Both adoption and output quality are tracked.

Google has invested heavily in internal Gemini deployment across Docs, Sheets, Slides, and Gmail. The expectation is that employees use these tools as part of normal workflows — not as occasional experiments.

Amazon and Salesforce

Amazon has embedded AI tool usage goals into its internal performance framework (Amazon calls these "leadership principles" evaluation criteria). Salesforce has tied Agentforce tool adoption to sales team performance goals, with reps evaluated on whether they use AI agents to support customer interactions.

The Employee Perspective: Fear, Friction, and the "Training Your Replacement" Problem

Not everyone is enthusiastic. Many employees report that mandated AI adoption feels coercive — and the underlying fear is rational: if you use AI tools effectively, you may demonstrate that your role can be done with fewer people. You are, in a sense, contributing to the case for your own eventual redundancy.

Experts note that friction often stems from a lack of trust rather than stubbornness. Many AI tools in corporate environments have well-documented accuracy and bias issues. Employees who have seen AI produce wrong code, hallucinate facts, or give legally problematic advice are reasonably cautious about staking their performance rating on AI-generated output.

The productivity gap is also real. Most companies are not yet seeing significant, measurable productivity gains from AI adoption — suggesting that the current push is as much about signaling competitiveness and justifying AI investment as it is about actual efficiency improvement.

What "Heavy AI User" Actually Means for Your Career

The emerging picture from 2026 data is a "superstar economy" dynamic in tech: high-leverage individual contributors who amplify their output using AI are being compensated at levels that previously required management roles. The ability to write and ship code faster, conduct research faster, and produce analysis faster — using AI as a force multiplier — is increasingly treated as the defining productivity signal.

At JPMorgan, heavy AI users are being offered compute power as part of compensation packages — GPU access, AI tool credits, and preferred access to new AI capabilities. At Meta, engineers hitting AI-assisted code targets are eligible for the highest performance ratings and the associated compensation jumps.

The practical message for knowledge workers: becoming a genuine AI power user is not just about productivity — it is now a direct input into your career trajectory at the largest employers in tech and finance.

How to Become a "Heavy AI User" (Practically)

Frequently Asked Questions