--- title: The same AI request emits 8.5 times more CO2 in Texas... description: Swedish AI provider Berget AI has released an open-source method for calculating the CO2 emissions of individual AI requests. A typical request to the open... canonical: https://daily.dev/posts/the-same-ai-request-emits-8-5-times-more-co2-in-texas-than-in-sweden-ccuzpfima twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: The same AI request emits 8.5 times more CO2 in Texas than in Sweden | daily.dev og:description: Swedish AI provider Berget AI has released an open-source method for calculating the CO2 emissions of individual AI requests. A typical request to the open... og:url: https://daily.dev/posts/the-same-ai-request-emits-8-5-times-more-co2-in-texas-than-in-sweden-ccuzpfima og:image: https://api.daily.dev/og/posts/cCuZPfImA.png og:image:alt: The same AI request emits 8.5 times more CO2 in Texas than in Sweden 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. # The same AI request emits 8.5 times more CO2 in Texas than in Sweden **[Hacker News](https://daily.dev/sources/hn)** · 6 min read · 2 upvotes · 0 comments ## Summary Swedish AI provider Berget AI has released an open-source method for calculating the CO2 emissions of individual AI requests. A typical request to the open model Kimi K3 emits about 0.4 grams of CO2 in Sweden, 1.6 grams in Ireland, 2.5 grams in Germany and 3.2 grams in Texas on identical hardware, 8.5 times more in Texas than Sweden. The method was built with researchers from SEI, WattTime and Klimatkollen, follows the Green Software Foundation's SCI for AI specification, and is MIT licensed on GitHub. It includes hardware manufacturing emissions and is available as the npm package @bergetai/co2-calculator. Berget AI customers can track emissions per request in the company's console. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### How much CO2 does a typical AI inference request emit depending on where the data center is located? Emissions vary heavily by grid: a typical request to the open model Kimi K3 emits about 0.4 grams of CO2 in Sweden, 1.6 grams in Ireland, 2.5 grams in Germany, and 3.2 grams in Texas, all on identical hardware. That makes the Texas request 8.5 times more carbon-intensive than the same request run in Sweden, driven by grid intensity differences (Sweden ~8 g CO2e/kWh vs Texas ~333 g CO2e/kWh). _Developers picking inference regions can weigh carbon impact alongside latency and cost via daily.dev._ ### Is there an open-source tool to calculate the carbon footprint of AI API requests? Yes, Berget AI released an MIT-licensed, open-source calculator and npm package called @bergetai/co2-calculator (version 1.1.1) that computes per-request emissions based on model, region, hardware, and time of day, following the Green Software Foundation's SCI for AI specification. The code and methodology are published on GitHub, and a hosted calculator is available at co2.berget.ai. _Teams adding carbon metrics to their AI stack can track tools like this one on daily.dev._ ### How much CO2 does manufacturing a data center GPU add to AI emissions calculations? Embodied hardware emissions are estimated at about 1,000 kg of CO2 per data center GPU, with refurbished GPUs counted as zero additional emissions. For a Kimi K3 request in Sweden, 89% of the total emissions come from hardware manufacturing rather than electricity, since Sweden's grid is extremely clean at roughly 8 g CO2e per kWh. _Anyone budgeting the real cost of AI hardware can follow sustainability methodology updates on daily.dev._ ## Community take How the wider developer community reacted, aggregated from 1 discussion and 146 comments across hackernews (as of 2026-10-23). **TL;DR:** Discussion barely engages with the CO2-per-request methodology itself and instead spirals into a long, heated debate about whether meat consumption is a bigger environmental problem than AI, with side arguments about energy markets, data center siting, and whataboutism. **Sentiment:** 5% positive · 25% mixed · 70% skeptical **The case for** - Some argue the tool usefully highlights that fossil-fuel-heavy grids like Texas's are the real driver of emissions differences, not the AI models themselves. - A few suggest per-request CO2 tracking could support tariffs or CO2 taxes that push the industry toward efficiency. **The pushback** - Many argue meat/beef consumption dwarfs AI's CO2 impact per unit, calling the focus on AI emissions a form of selective outrage. - Some push back that building data centers in dirty-grid regions is an added, avoidable choice unlike other established industries. - Several note that building efficient datacenters elsewhere doesn't actually reduce Texas-style buildouts, it just adds more total capacity and emissions. - One detailed rebuttal points out the gap is really about grid fossil-fuel mix (gas/coal vs hydro/nuclear/wind), not AI hardware or model efficiency at all. - Debate over whether clean energy capacity is a fixed, zero-sum resource that AI inference competes with, with disagreement on whether this framing is valid. **By community** - hackernews (heated): The thread turns into a prolonged, contentious argument about meat vs. AI emissions and whataboutism, with only a technical subthread addressing why Texas's grid is dirtier. **Hottest debate:** Whether criticizing AI's CO2 footprint is hypocritical/selective outrage given that meat consumption and other personal choices have far larger carbon impacts. **Open questions** - What actually counts as a single 'request' for emissions accounting purposes (e.g., agentic workflows with many sub-steps)? - Would building AI datacenters in low-carbon regions actually offset dirtier ones elsewhere, or simply add more total capacity and emissions? **Highlights** > For 2025 the grid carbon intensity in Texas was 363gCO2eq/kWh, while it was 22gCO2eq/kWh in Sweden, so 16,5x more in Texas. Right now it's 427 in Texas vs. 50 in Finland ie. 8,54x more, like in the article (yes I looked the wrong country at first). The reason is half of Texas electricity production is still fossil fuels (gas & coal) while mostly hydro, nuclear and wind in Sweden. So it's not really about ai, the models or the hardware. Fossil fuels make CO2, that's it. > — [blaze33 on hackernews · 2 comments](https://news.ycombinator.com/item?id=50205966) > why not both? some perspective: 0.5 Pwh/yr - worldwide data center energy 4-5 PWh/yr - meat total farm to table 166 PWh/yr - worldwide total energy production the true bottleneck and cost we have is energy, carbon is a polite proxy if you can meaningfully cut down the energetic cost, that's very valuable either way maybe tomorrow you'll want to spin up a GPU on the cloud to briefly run a model, pick a data center: DC #1: EU Efficient, but has ~100ms additional latency DC #2: TX Runs on jet engine turbines, neighbors complain of noise and air pollution env impact is something worth considering regardless of industry > — [achille on hackernews](https://news.ycombinator.com/item?id=50205888) > Chat GPT output; do take with a grain of salt. TX generated 4x the absolute solar electricity as Sweden and 5x per person in the last 12 months. > — [prasadjoglekar on hackernews · 3 comments](https://news.ycombinator.com/item?id=50206131) > Thanks. My next question would be what's a typical request. The article defines it as "production-typical request of 5,000 input and 387 output tokens, weighted by a typical day's traffic" for the Kimi K3 model. Not sure how that translates to my scale of the blue button to RIIR. > — [thesdev on hackernews](https://news.ycombinator.com/item?id=50205990) > Except they wouldn't move to Texas but would also build in Texas. So you don't get a less polluting datacenter but you get one less polluting and one more polluting datacenter. A Swedish datacenter doesn't simply replace one in Texas, it's an additional one. > — [croes on hackernews](https://news.ycombinator.com/item?id=50206279) **Source threads** - [hackernews](https://news.ycombinator.com/item?id=50205229) · 106 points · 146 comments ## Similar posts on daily.dev - [How Many Spoons Does Your AI Coding Tool Cost?](https://daily.dev/posts/how-many-spoons-does-your-ai-coding-tool-cost--9tixdpv0g) · Kilo Blog · 0 upvotes · 0 comments --- Tags: [#open-source](https://daily.dev/tags/open-source) [View this post on daily.dev](https://daily.dev/posts/the-same-ai-request-emits-8-5-times-more-co2-in-texas-than-in-sweden-ccuzpfima) ```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":"The same AI request emits 8.5 times more CO2 in Texas than in Sweden","url":"https://daily.dev/posts/the-same-ai-request-emits-8-5-times-more-co2-in-texas-than-in-sweden-ccuzpfima","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/the-same-ai-request-emits-8-5-times-more-co2-in-texas-than-in-sweden-ccuzpfima"},"datePublished":"2026-10-23T08:10:50.450Z","dateModified":"2026-10-23T16:57:11.665Z","description":"Swedish AI provider Berget AI has released an open-source method for calculating the CO2 emissions of individual AI requests. A typical request to the open...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/331c8420fd42908f66920a9b0d818010?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/331c8420fd42908f66920a9b0d818010?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Hacker News","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":"Hacker News","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/hn","url":"https://daily.dev/sources/hn"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/the-same-ai-request-emits-8-5-times-more-co2-in-texas-than-in-sweden-ccuzpfima","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":2},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"open-source","timeRequired":"PT6M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Hacker News","item":"https://daily.dev/sources/hn"},{"@type":"ListItem","position":3,"name":"The same AI request emits 8.5 times more CO2 in Texas than in Sweden"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/the-same-ai-request-emits-8-5-times-more-co2-in-texas-than-in-sweden-ccuzpfima#faq","mainEntity":[{"@type":"Question","name":"How much CO2 does a typical AI inference request emit depending on where the data center is located?","acceptedAnswer":{"@type":"Answer","text":"Emissions vary heavily by grid: a typical request to the open model Kimi K3 emits about 0.4 grams of CO2 in Sweden, 1.6 grams in Ireland, 2.5 grams in Germany, and 3.2 grams in Texas, all on identical hardware. That makes the Texas request 8.5 times more carbon-intensive than the same request run in Sweden, driven by grid intensity differences (Sweden ~8 g CO2e/kWh vs Texas ~333 g CO2e/kWh). Developers picking inference regions can weigh carbon impact alongside latency and cost via daily.dev."}},{"@type":"Question","name":"Is there an open-source tool to calculate the carbon footprint of AI API requests?","acceptedAnswer":{"@type":"Answer","text":"Yes, Berget AI released an MIT-licensed, open-source calculator and npm package called @bergetai/co2-calculator (version 1.1.1) that computes per-request emissions based on model, region, hardware, and time of day, following the Green Software Foundation's SCI for AI specification. The code and methodology are published on GitHub, and a hosted calculator is available at co2.berget.ai. Teams adding carbon metrics to their AI stack can track tools like this one on daily.dev."}},{"@type":"Question","name":"How much CO2 does manufacturing a data center GPU add to AI emissions calculations?","acceptedAnswer":{"@type":"Answer","text":"Embodied hardware emissions are estimated at about 1,000 kg of CO2 per data center GPU, with refurbished GPUs counted as zero additional emissions. For a Kimi K3 request in Sweden, 89% of the total emissions come from hardware manufacturing rather than electricity, since Sweden's grid is extremely clean at roughly 8 g CO2e per kWh. Anyone budgeting the real cost of AI hardware can follow sustainability methodology updates on daily.dev."}}]} ```