🔍 Read the full analysis: AI Coding Payoff: What Costs To Watch And Results To Measure on ThorstenMeyerAI.com
Get business pricing on monitors, keyboards and dev gear
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
A report drawing on interviews at 27 workplaces describes wide variation in AI coding costs and mixed results across productivity measures. Some teams report gains in triage, code research and backlog work, but companies have not established a consistent link between tool usage, more pull requests and delivered value.
Engineering teams surveyed across 27 workplaces spend between about $150 and $3,000 per developer per month on AI coding tools, but many cannot say what measurable value the spending produced, according to a report published October 20 by The Pragmatic Engineer. The findings arrive as companies set 2027 budgets and some discounted subscriptions are due to expire.
The figures vary widely, including a publicly traded company with 1,500 engineers that caps spending at $150 per developer monthly and a two-person European software agency that spends about $3,000 per developer. Other firms reported caps of $1,000 to $4,000, with typical use around $1,000. The survey does not establish that company size predicts spending.
Output figures also differ by source and measure. CodeRabbit, an AI code-review vendor, says weekly pull requests on its platform rose 113% year over year. In a study of tens of thousands of engineers, Microsoft found adopters merged about 24% more pull requests over four months during an early-2026 rollout of Claude Code and GitHub Copilot CLI. The Microsoft researchers cautioned that a merged pull request does not by itself show the value delivered.
Companies interviewed reported results ranging from flat pull-request counts to increases of five to ten times. At one firm, pull-request frequency was unchanged while the average pull request grew to three times its former size. These are different measures and settings, not a single comparable productivity estimate.
Infographic 01 · Where the money goes
From a $150 cap to $3,000 a month per developer.
Anecdotal data from 27 workplaces, mostly anonymized. It shows how wide the range is, not an industry average.
Spend per developer per month
Examples from the free part of the article · scale to $4,000
actual or typical spendbudget cap
What changed in 2026
Why the 2027 budget needs a range
- Peak in June and July: spend has roughly halved since for companies on API plans (Opus 5.5 and open models).
- Caps are common at medium and high spenders, and most people say they are easy to get around.
- Subsidies are ending: some companies must optimize or see the bill rise 5–10×.
- GitHub Copilot bills by token since 1 June 2026, at each model’s API rate, with no fallback to a cheaper model.
- Company size does not predict spend.
ROI is “a question that few companies seem able to answer – at least for now.”Gergely Orosz, The Pragmatic Engineer, 20 Oct 2026
Sources: The Pragmatic Engineer, “What’s the ROI of software engineers’ spending on AI?” (20 Oct 2026, free part); GitHub (27 Apr 2026).
Powered by Thorsten Meyer AI
Budget Decisions Need Outcome Measures
The gap between tool spending and demonstrated outcomes makes budget decisions harder as companies face usage-based pricing and the possible end of discounts. The source report says several companies expect discounted subscriptions to expire, which could raise bills five- to tenfold if usage and pricing are unchanged. Those increases are expectations reported by the companies, not a confirmed outcome across the industry.
The report points to work such as issue triage, code archaeology, prototyping and backlog cleanup as areas where teams describe concrete time savings. But higher activity can also mean larger or more numerous pull requests without proving that software is safer, easier to maintain or more useful to customers. Veracode’s latest test, cited in the source, found known vulnerabilities in 44% of AI coding tasks; the provided material does not specify enough about the test design to generalize that rate to all AI-generated code.
For leaders, the practical question is whether AI changes delivery outcomes rather than simply increasing usage. The report recommends tracking cost per accepted change, rework, change-failure rate and stories closed versus stories arriving. These measures can help expose costs that raw token or pull-request totals miss.
As an affiliate, we earn on qualifying purchases.
Pricing and Productivity Signals
The Pragmatic Engineer’s October 20 report draws on interviews with engineering leaders and senior engineers at 27 workplaces, alongside aggregate information from AI code-review vendors CodeRabbit and Gitar. The source article says its cited findings come from the freely readable portion of that report. Vendor data reflects activity on those platforms and should not be treated as a controlled comparison of all software teams.
Pricing has also shifted. The source says GitHub moved every Copilot plan to usage-based billing on June 1, charging tokens at each model’s API rate and removing the fallback to a cheaper model. GitHub described the change as a response to greater use of agents and their higher computing and inference demands. The article also reports that API-plan spending peaked in June and July and had roughly halved since among some companies, which its author attributes to Opus 5.5 and a move to open models.
Controlled productivity research cited in the source is mixed. METR initially found experienced open-source developers took 19% longer with AI in early 2025. A February follow-up estimated an 18% speedup for returning developers and 4% for new developers, but both confidence intervals included zero. METR also said the data was unreliable, in part because 30% to 50% of developers held back tasks they preferred to do without AI.
“ROI is “a question that few companies seem able to answer – at least for now.””
— Gergely Orosz, The Pragmatic Engineer
As an affiliate, we earn on qualifying purchases.
The ROI Evidence Is Still Uneven
The available figures do not provide a single, directly comparable estimate of AI coding’s effect. The studies and company accounts use different measures, populations and time periods; the reported pull-request increases cannot by themselves establish improvements in quality, delivery speed or business value. The 27-workplace survey is not described as a representative industry sample.
It is also unclear how much reported output growth comes from AI rather than other changes in staffing, workflows, project mix or review practices. The source author’s view that individual output is probably up by double digits is an interpretation, not a settled finding. METR’s estimates have confidence intervals that include zero, while its survey of 349 technical workers found a median self-reported increase in work value of 1.4 to 2 times; METR said participants had overestimated AI’s effect in an earlier period.
The source does not provide full details for the cited Veracode test or the specific methods behind every company estimate. It also remains uncertain how much bills will change when discounts expire, since costs depend on tools, models and usage.
As an affiliate, we earn on qualifying purchases.
Companies Set 2027 Scorecards
As teams plan 2027 budgets, the next step is to set a baseline and compare AI-assisted work against outcomes over a defined period. The measures proposed in the report include cost per accepted change, rework, change-failure rates and whether completed user stories outpace incoming work. These indicators need consistent definitions if companies want to compare teams or periods.
Organizations will also need to watch how usage-based billing, model choices and expiring discounts affect monthly spend. The source does not name a common industry deadline for those changes or a standard evaluation framework. Results from future company tracking and more controlled studies will be needed to clarify whether reported activity gains translate into reliable, valuable software delivery.
As an affiliate, we earn on qualifying purchases.
Key Questions
How much are companies spending on AI coding tools?
The 27-workplace survey cited by The Pragmatic Engineer reported budgets from about $150 to $3,000 per developer per month, with some other firms setting caps between $1,000 and $4,000. These are reported examples, not a representative industry average.
Do the figures prove AI coding increases productivity?
No single figure proves a general productivity increase. CodeRabbit reported 113% more weekly pull requests on its platform, while Microsoft found adopters merged about 24% more pull requests in its study. Other companies reported flat counts, and pull-request volume alone does not establish quality or value.
Where do teams report the clearest benefits?
The source describes reported gains in triage, understanding existing code, prototyping and backlog work, particularly on simple, uniform software stacks. These are company accounts and should not be assumed to apply equally to complex legacy systems or every team.
What should companies measure to judge returns?
The report recommends tracking cost per accepted change, rework, change-failure rates and stories completed compared with stories entering the backlog. It argues these measures are more informative than tool usage or raw pull-request counts alone.
Source: ThorstenMeyerAI.com
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
