AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: Amazon’s Tokenmaxxing Warning Meets Its Own AI Investment Strategy on ThorstenMeyerAI.com

Before you orderOffer from Amazon

Get the latest gadgets delivered free with Prime

  • Fast, free delivery on millions of items
  • Prime Video, Amazon Music and more included
  • Member-only deals all year
Start your free Prime trial Free trial for eligible customers · Cancel anytime
As an affiliate, we earn on qualifying purchases.

TL;DR

Amazon published guidance on October 27 telling software engineers not to treat token use or other AI activity metrics as performance targets. The memo favors customer and business outcomes, while Amazon’s own Nova experience and its investment in Anthropic provide contrasting examples of AI bets whose reported results differ.

Amazon published new guidance for its software engineers on October 27, warning that targets based on AI token use, tool adoption or code volume can reward activity rather than results. The memo, written by senior engineering leaders, says employees should be judged on impact for customers, the business and their teams—a message that lands as the company faces questions about its own AI portfolio, including the reported scaling back of some Nova models and its large investment in Anthropic.

The memo was published on Amazon’s news site and introduces a revised role guideline for software engineers. It says measures such as tokens used, AI-tool adoption, code changes and reviews can help teams understand work, but become flawed when set as goals. Using them as targets, it warns, can increase costs and system complexity. The memo invokes Amazon’s leadership principles, saying “tokenmaxxing is at odds with frugality” and measuring lines of code is at odds with inventing and simplifying.

The guidance does not say AI tools should be avoided. It says these activity measures currently correlate well with outcomes, and that teams using AI most are delivering more, while teams not adopting it are falling behind. The distinction is between using data to assess performance and turning it into a score employees are expected to maximize. The memo says leaders should examine both low and high adopters to understand where tools help and whether heavier use produces better work.

It also treats a lack of benefit as useful information: “If your work does not benefit from current AI tooling, that is a finding, not a failing.” Engineers are asked to explain why. Leaders, meanwhile, should use the tools in their own production work and identify examples where they helped and where they did not. The memo says the role guideline avoids prescribing specific AI rules because definitions written now may quickly become outdated.

At a glance
reportWhen: Memo published October 27, 2026; Amazon…
The developmentAmazon issued engineering guidance rejecting AI usage metrics as targets and urging staff to judge tools by their outcomes.
Thorsten Meyer AI
Timeline

Six months at Amazon

From token leaderboards to an outcomes memo

Amazon spent 2026 learning the same lesson twice: once with how its engineers use AI, once with its own AI models.

  1. May 2026Token leaderboards backfireFinancial Times (via The Decoder): staff give an internal agent tool pointless tasks to inflate token counts. Reported, anonymous sources.
  2. Late May 2026KiroRank “deprecated”Amazon spokesperson to Business Insider. SVP Dave Treadwell: “Please don’t use AI just for the sake of using AI.”
  3. 24 Jul 2026San Francisco AGI Lab site closesConfirmed by Amazon to GeekWire. Frontier model research continues under Pieter Abbeel.
  4. 28 Jul 2026Most flagship Nova models wound downBusiness Insider, people familiar: Premier, Omni, Reel, Canvas to “keep the lights on”. Amazon: models “remain one of the most important things we’re working on”.
  5. 14 / 30 Sep 2026Nova Sonic v1, Nova Reel, Nova Canvas reach end of lifeAWS Bedrock model lifecycle page. Requests to them now fail.
  6. 26 Oct 2026Anthropic files its S-1Amazon notes: $6.2 billion principal, 12% ownership cap that Amazon can waive. Anthropic commits at least $110 billion to Amazon through April 2036.
  7. 27 Oct 2026The memo: outcomes, not tokenmaxxingUsage and throughput metrics are “deeply flawed” as targets. “Tokenmaxxing is at odds with frugality.”
  8. 29 Oct 2026Amazon’s third-quarter resultsAnnounced by Amazon. Not yet published when this article went live.

Sources: Amazon engineering memo (27 Oct 2026); Business Insider (29 May and 28 Jul 2026); GeekWire (24 Jul 2026); The Decoder (12 May 2026, summarising the Financial Times); AWS Bedrock model lifecycle (read 28 Oct 2026); Anthropic S-1 (26 Oct 2026).
Powered by Thorsten Meyer AI

Amazon Reframes AI Performance Metrics

The memo matters beyond an internal engineering policy because it addresses a growing management problem: how to measure AI adoption without rewarding usage for its own sake. A token leaderboard can be easy to count, but it does not establish that a tool improved software, reduced effort or helped customers. Amazon’s guidance says teams should treat usage figures as diagnostic signals, not as a substitute for outcomes.

That distinction has practical consequences for employees and companies deploying generative AI. Targets can influence behavior, including encouraging work that inflates activity measures. The memo’s approach instead asks managers to examine what changed and why. It also leaves room for engineers to report that a tool is not useful for a particular task, rather than pressuring them to use it regardless.

The policy arrives alongside a mixed picture of Amazon’s own AI investments. Business Insider reported in July that most flagship Nova models had been moved to maintenance, while Amazon’s filings show a substantial increase in the reported value of its Anthropic holdings. Those developments do not prove that one strategy succeeded and the other failed, but they sharpen the question the memo raises: which AI investments deliver measurable value?

Products Worth Considering

Bose TV Speaker (2nd Gen) Compact Soundbar with Bluetooth and HDMI Connectivity

Bose TV Speaker (2nd Gen) Compact Soundbar with Bluetooth and HDMI Connectivity

As an affiliate, we earn on qualifying purchases.

From Usage Leaderboards to Outcomes

The memo follows earlier reports of internal tracking and employee incentives around AI use. The Financial Times reported in May that some Amazon staff assigned unnecessary tasks to an internal agent called MeshClaw to raise token counts. Those accounts were attributed to employees and are not independently established by the memo. The same reporting described a target for more than 80% of developers to use AI weekly.

Separately, Amazon confirmed to Business Insider at the end of May that its KiroRank leaderboard had been deprecated. The company described it as an informal tracker created by employees and said it was never intended to promote AI use for its own sake. Senior Vice President Dave Treadwell also told staff, “Please don’t use AI just for the sake of using AI.” The October guidance makes that position part of an official engineering role guideline.

On the investment side, Business Insider reported in July that most flagship Nova models had been moved to “keep the lights on.” Amazon’s lifecycle information lists Nova Canvas, Nova Reel and the first Nova Sonic as retired. At the same time, Amazon’s June 30 filing put its Anthropic holdings at about $190 billion, compared with roughly $18 billion invested, according to the source report. Those figures describe the stated value of the holdings, not cash proceeds from a sale.

““Tokenmaxxing is at odds with frugality.””

— Amazon engineering memo

What the Investment Figures Show

The memo does not disclose how Amazon will evaluate individual engineers or whether teams will stop collecting token and adoption data. It says the measures can still correlate with outcomes, leaving open how managers should weigh them alongside customer and business results. Amazon has not provided a public accounting in the memo of productivity gains or costs attributable to AI tools.

The reported Nova and Anthropic figures also have limits. Business Insider’s account of Nova’s status is not a full public assessment of every model or use case. The roughly $190 billion Anthropic figure is a reported carrying value in a filing, not a realized return, and it can change with valuation and accounting. The source does not establish that the two investments can be compared directly or that the memo was prompted by either portfolio outcome.

Amazon Earnings Due October 29

Amazon is scheduled to report third-quarter results on October 29, one day after the memo’s publication. Those results may provide further financial context for the company’s AI spending and investments, but the figures and any related commentary were not available in the source material. The memo itself sets out no deadline for a new measurement system.

For now, the stated next step is operational: engineering leaders are expected to use AI tools on their own work, compare tasks where the tools help with those where they do not, and discuss the reasons with their teams. Whether Amazon later publishes clearer outcome measures—or how it will track adoption without turning it into a quota—remains open.

Key Questions

What did Amazon tell its engineers about AI use?

Amazon’s October 27 memo says token use, AI adoption, code changes and reviews should not become performance targets. It asks engineers and managers to focus on customer, business and team outcomes.

Is Amazon telling employees to stop using AI?

No. The memo says AI adoption measures currently correlate with important outcomes and that teams not adopting AI are falling behind. Its warning is against maximizing usage as a goal, not against using tools where they help.

What is tokenmaxxing?

In the reported Amazon example, tokenmaxxing means using an AI tool for unnecessary tasks to increase the number of tokens processed and improve a usage ranking. Token counts measure activity, but do not by themselves show that work improved.

What does Amazon’s Anthropic investment figure mean?

The source report says Amazon’s June 30 filing valued its Anthropic holdings at about $190 billion, against roughly $18 billion invested. That is a reported paper value, not a confirmed cash gain from selling the stake.

Source: ThorstenMeyerAI.com

You May Also Like

What Makes Desktop CNC Routers So Appealing

Pioneering versatility and precision, desktop CNC routers open new creative horizons, but what exactly makes them so appealing?

Why ByteDance’s AI Innovations Could Revolutionize Autonomous Driving

ByteDance is exploring autonomous-driving technology through its Seed world-model team, but has not confirmed plans for a commercial vehicle business.

Understanding ‘Run’ In Frontier AI For Mac Studio Users

Exploring what ‘run’ means for frontier-scale AI models on the new Mac Studio with 512GB memory, including capabilities, limitations, and implications.

Can Shippy Teach Us How To Build More Efficient AI Agents?

Ai2 details how Shippy, its maritime AI, emphasizes reliability through auditable instructions, deterministic tools, and human verification, beyond model capability.