Amazon Addresses AI Usage Among Engineers

Amazon has recently advised its software engineers to refrain from artificially inflating their AI usage to appear more productive, as revealed in an internal memo from September that was made public this week. This practice, known as “tokenmaxxing,” has been encouraged by various tech companies but seems to be facing scrutiny now, especially with rising usage costs.

According to reports, the memo explicitly discourages “tokenmaxxing,” where engineers excessively consume AI tokens to show heightened engagement with the technology. Additionally, it redefines the roles of software development engineers and managers now that AI tools are integral to the coding process. The memo suggests, “We’re just seeing the beginning of AI’s effect on software engineering.”

The tech industry is grappling with escalating AI tool costs and struggling to ascertain their actual impact on productivity. Some companies have imposed limits on token usage among employees or have opted for more affordable open-source alternatives, including Chinese technologies. However, Amazon has not taken this route, as a company spokesperson indicated there are no restrictions on employee token use.

Amazon has encountered similar issues before; in May 2026, it dismantled an internal leaderboard after finding that employees manipulated AI usage metrics to enhance their rankings, an instance of the very behavior the recent memo cautions against.

The memo highlights three key metrics—token usage, AI tool adoption rates, and the volume of code changes and reviews—as “deeply flawed” indicators of AI utilization. It warns that relying heavily on these metrics could inflate costs and complicate engineering tasks. While not advocating for the complete disregard of these figures, the memo stresses that they should be evaluated “with care and judgment” in conjunction with other data. It further states, “We believe that it is best to measure, and be measured by, long-term outcomes.”

The document also provides space for engineers to assert that AI may not be helpful for specific tasks. It states, “If your work does not benefit from current AI tooling, that is a finding, not a failing,” and emphasizes that an engineer can showcase the required judgment by explaining where AI falls short in aiding their work.

Despite these caveats, the memo recognizes AI as a pivotal element of the future of software engineering, even as many details remain to be finalized. The memo notes, “While we have a vision for what the future looks like, there’s a lot that’s still emerging,” asserting that AI will undoubtedly play a central role in software engineering practices moving forward.

Within Amazon Web Services, a different metric is being employed. Marc Brooker, a vice president and distinguished engineer at AWS, mentioned that the company tracks “feature velocity,” assessing “how many launches are we doing, how many successful changes are we making in production” to evaluate AI’s impact on engineering output. He described Amazon’s overall AI spending approach as being centered around return on investment, explaining, “If there is an important problem to solve for our customers, important problem to solve for our business, we’re going to invest tools, time, tokens, human effort, etc. into those.”

He also pointed out the delicate balance between discipline and experimentation, stating, “What we don’t want to see is wasted effort where we are going and spending those tokens solving problems that don’t matter and sometimes it’s hard to know, so we do need to have some appetite for experimentation.”

The memo comes amid broader discussions from Amazon’s leadership regarding AI’s impact on the workforce. CEO Andy Jassy has indicated that AI will “change the scope and speed” of everyday work, previously mentioning that AI-driven automation could reduce Amazon’s corporate workforce in the years to come, while also creating new job opportunities.

As companies of all sizes, from Amazon to smaller startups, evaluate how best to leverage AI’s potential, discussions continue on finding a balance that maximizes benefits while avoiding pitfalls.