--- title: Stop counting tokens, start counting what ships | daily.dev description: Rationing AI coding usage with caps and seat limits optimizes the wrong metric, this sponsored HPE piece argues. Cost per token measures consumption, not... canonical: https://daily.dev/posts/stop-counting-tokens-start-counting-what-ships-jh7lr6jmj twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Stop counting tokens, start counting what ships | daily.dev og:description: Rationing AI coding usage with caps and seat limits optimizes the wrong metric, this sponsored HPE piece argues. Cost per token measures consumption, not... og:url: https://daily.dev/posts/stop-counting-tokens-start-counting-what-ships-jh7lr6jmj og:image: https://api.daily.dev/og/posts/JH7lR6jmJ.png og:image:alt: Stop counting tokens, start counting what ships og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Stop counting tokens, start counting what ships **[MIT Technology Review](https://daily.dev/sources/technologyreview)** · 6 min read · 0 upvotes · 0 comments ## Summary Rationing AI coding usage with caps and seat limits optimizes the wrong metric, this sponsored HPE piece argues. Cost per token measures consumption, not results, so a better metric is cost per merged pull request. Uber reportedly exhausted its 2026 AI budget by April after expanding Claude Code use, and Gartner predicts AI coding token costs will surpass average developer salaries by 2028. HPE's hybrid platform routes routine workloads from 150+ internal apps to open-weight models on HPE Private Cloud AI, cutting token costs 30 to 50%. It also found the top 5% of users drive most consumption, using roughly 450 times more than the lightest users on the same license. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What metric should I use instead of token count to measure AI coding agent ROI Cost per merged pull request is recommended: total AI-assisted development cost divided by the number of changes that pass review and are merged, tracked alongside rework and defects so cheaper changes don't hide lower quality. If that cost falls as AI usage rises without hurting quality, more consumption is converting into useful software. _Teams weighing AI coding spend against output can track frameworks like this through daily.dev._ ### Why did Uber run out of its AI coding budget so fast Uber reportedly exhausted its entire 2026 AI budget by April after expanding its use of Claude Code, illustrating how agentic coding tasks can use 10 to 50 times more tokens than a basic chat prompt because an agent plans, generates code, runs tests, and repeats the cycle to fix errors. _Developers budgeting for agentic coding tools can follow cost trends like this on daily.dev._ ### How much can routing to open-weight models reduce AI coding token costs HPE reports token costs down 30 to 50% after routing routine workloads from more than 150 internal applications to open-weight models running on premises, including on HPE Private Cloud AI, while reserving frontier or on-premises flagship models for difficult planning and novel reasoning tasks. _Teams evaluating hybrid inference setups can compare approaches like this on daily.dev._ ## Similar posts on daily.dev - [The token bill comes due: Inside the industry scramble to manage AI’s runaway costs](https://daily.dev/posts/the-token-bill-comes-due-inside-the-industry-scramble-to-manage-ai-s-runaway-costs-glxojg09s) · TechCrunch · 0 upvotes · 0 comments - [Microsoft’s quiet Claude Code retreat and the real cost of enterprise AI](https://daily.dev/posts/microsoft-s-quiet-claude-code-retreat-and-the-real-cost-of-enterprise-ai-8ycsrgrmm) · The Next Web · 44 upvotes · 6 comments - [Whether tokenmaxxing or tokenminimizing, you’re measuring the wrong thing](https://daily.dev/posts/whether-tokenmaxxing-or-tokenminimizing-you-re-measuring-the-wrong-thing-w9guajhob) · Dev Interrupted · 0 upvotes · 0 comments - [The new economics of AI cost management](https://daily.dev/posts/the-new-economics-of-ai-cost-management-tguezsxg2) · databricks · 1 upvotes · 0 comments --- Tags: [#open-source](https://daily.dev/tags/open-source), [#ai-agents](https://daily.dev/tags/ai-agents), [#llm](https://daily.dev/tags/llm), [#dora-metrics](https://daily.dev/tags/dora-metrics) [View this post on daily.dev](https://daily.dev/posts/stop-counting-tokens-start-counting-what-ships-jh7lr6jmj) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Stop counting tokens, start counting what ships","url":"https://daily.dev/posts/stop-counting-tokens-start-counting-what-ships-jh7lr6jmj","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/stop-counting-tokens-start-counting-what-ships-jh7lr6jmj"},"datePublished":"2026-11-02T13:43:21.254Z","dateModified":"2026-11-02T13:43:52.566Z","description":"Rationing AI coding usage with caps and seat limits optimizes the wrong metric, this sponsored HPE piece argues. Cost per token measures consumption, not...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/ab408d476268ad1497bb7f944abe454c?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/ab408d476268ad1497bb7f944abe454c?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"MIT Technology Review","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"MIT Technology Review","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/a8d03889722f4f22b3986e0b395ff076","url":"https://daily.dev/sources/technologyreview"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/stop-counting-tokens-start-counting-what-ships-jh7lr6jmj","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"open-source,ai-agents,llm,dora-metrics","timeRequired":"PT6M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"MIT Technology Review","item":"https://daily.dev/sources/technologyreview"},{"@type":"ListItem","position":3,"name":"Stop counting tokens, start counting what ships"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/stop-counting-tokens-start-counting-what-ships-jh7lr6jmj#faq","mainEntity":[{"@type":"Question","name":"What metric should I use instead of token count to measure AI coding agent ROI","acceptedAnswer":{"@type":"Answer","text":"Cost per merged pull request is recommended: total AI-assisted development cost divided by the number of changes that pass review and are merged, tracked alongside rework and defects so cheaper changes don't hide lower quality. If that cost falls as AI usage rises without hurting quality, more consumption is converting into useful software. Teams weighing AI coding spend against output can track frameworks like this through daily.dev."}},{"@type":"Question","name":"Why did Uber run out of its AI coding budget so fast","acceptedAnswer":{"@type":"Answer","text":"Uber reportedly exhausted its entire 2026 AI budget by April after expanding its use of Claude Code, illustrating how agentic coding tasks can use 10 to 50 times more tokens than a basic chat prompt because an agent plans, generates code, runs tests, and repeats the cycle to fix errors. Developers budgeting for agentic coding tools can follow cost trends like this on daily.dev."}},{"@type":"Question","name":"How much can routing to open-weight models reduce AI coding token costs","acceptedAnswer":{"@type":"Answer","text":"HPE reports token costs down 30 to 50% after routing routine workloads from more than 150 internal applications to open-weight models running on premises, including on HPE Private Cloud AI, while reserving frontier or on-premises flagship models for difficult planning and novel reasoning tasks. Teams evaluating hybrid inference setups can compare approaches like this on daily.dev."}}]} ```