🔍 Read the full analysis: What Does AI Coding Cost, And How Can You Track Its Value? on ThorstenMeyerAI.com
Get hardware and tech essentials delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Engineering teams in a survey of 27 workplaces reported AI coding budgets ranging from $150 to about $3,000 per developer per month, while companies reported sharply different output changes. The evidence points to gains in specific tasks, but does not establish a universal return on investment; teams can track value through accepted changes, rework and delivery outcomes rather than tool usage alone.
Engineering teams at 27 workplaces reported AI coding budgets ranging from $150 to about $3,000 per developer per month, but the data does not show a consistent link between spending and delivered value. The findings, published October 20 by The Pragmatic Engineer, arrive as companies set 2027 budgets and face usage-based charges that can make AI bills harder to predict.
The survey and interviews, reported by The Pragmatic Engineer’s Gergely Orosz, found that company size did not predict spending. One publicly traded company with 1,500 engineers caps monthly spending at $150 per developer; a two-person European software agency reported spending about $3,000 per developer. Other companies described caps between $1,000 and $4,000, with typical developer spending around $1,000. The findings reflect the workplaces surveyed, not an industry-wide average.
Reported productivity results also varied. Microsoft’s study of tens of thousands of engineers during an early-2026 rollout of Claude Code and GitHub Copilot CLI found adopters merged about 24% more pull requests than they otherwise would have over four months. CodeRabbit, an AI code-review vendor, said developers on its platform produced 113% more pull requests per week than a year earlier. Those figures use different methods and populations; neither by itself establishes how much additional business value the work created.
Among the companies interviewed, some saw little change in PR frequency, while others reported doubling or larger increases. One team’s PRs became three times larger without an increase in frequency. The Microsoft study’s authors cautioned that “a merged PR is not the same as the value it delivers.” The report also cites a Veracode test in which 44% of AI coding tasks produced a known vulnerability, underscoring the need to count review and security work alongside output.
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
Why AI Coding Bills Need Outcome Measures
The gap between tool usage and delivered value matters as firms decide whether to expand, limit or change AI coding plans. A higher PR count can reflect faster delivery, but it can also mean larger review burdens, more rework or changes that do not improve a product. If subscriptions or discounts end, the cost of those tools may rise while the benefit remains difficult to quantify.
The report describes a quant trading firm whose monthly AI spending rose from $500,000 in January to $8 million in September. Its largest spenders were traders rather than engineers, and the firm measures spending against each desk’s profit and loss. Most companies lack that direct scoreboard. For them, tracking costs against accepted changes, rework, failures and completed work can give leaders a more useful picture than token consumption alone.
Task and codebase also appear to shape returns. Interviewees described clear gains in customer-issue triage, understanding unfamiliar code, prototypes and backlog work on simple, uniform stacks. The report’s examples do not establish that those gains will transfer to every team, particularly teams maintaining complex legacy systems.
As an affiliate, we earn on qualifying purchases.
How Costs and Productivity Shifted
The October 20 report combined interviews with engineering leaders and senior engineers at 27 workplaces with aggregate data from CodeRabbit and Gitar, two AI code-review vendors. The source says its cited material comes from the freely readable portion of Orosz’s article. Its sample is useful for showing the range of company experiences, but it is not presented as a representative survey of all software firms.
According to Orosz, API-plan spending at some companies peaked in June and July and had roughly halved by the time of his report. He attributed that change to the availability of Opus 5.5 and a move to open models. Several companies also told him that discounted subscriptions were due to end, potentially increasing bills if they did not adjust usage. These are reported company experiences, not a forecast that every AI coding bill will rise.
Pricing has also shifted. GitHub changed all Copilot plans to usage-based billing on June 1, charging for tokens at each model’s API rate and removing the fallback to a cheaper model. GitHub said: “Agentic usage is becoming the default, and it brings significantly higher compute and inference demands.” That change makes monitoring usage and model choice more relevant to budgeting.
Research findings on productivity remain mixed. METR’s early-2025 study found experienced open-source developers took 19% longer with AI. Its February follow-up estimated an 18% speedup for returning developers and a 4% speedup for new developers, but both confidence intervals included zero. METR said the data was unreliable in part because 30% to 50% of participants held back tasks they did not want to attempt without AI.
““A merged PR is not the same as the value it delivers.””
— Microsoft study authors, as quoted in The Pragmatic Engineer
As an affiliate, we earn on qualifying purchases.
What the Available Evidence Cannot Show
The available findings do not establish a single productivity or return-on-investment figure for AI coding across the industry. The workplaces in the report vary in size, tasks, technology and measurement methods, while vendor platform data reflects users of those products. Reported output increases are not directly comparable, and PR counts do not show whether changes were valuable, safe or durable.
It is also unclear how much of the reported improvement is attributable to AI rather than other changes in staffing, processes or workload. METR’s results have wide uncertainty, and the source says its survey participants’ self-reported gains may overstate effects. The source excerpt ends during a discussion of a DORA report and does not provide the full details of that evidence. The 44% Veracode figure is specific to its test and should not be treated as the rate for every tool or company.
Finally, actual bills will depend on model choice, usage, pricing plans and whether discounts continue. The report does not offer a representative cost average or a reliable forecast for companies’ 2027 spending.
As an affiliate, we earn on qualifying purchases.
How Companies Can Track 2027 Spend
The report recommends measuring outcomes rather than treating tokens or generated code as proof of value. Teams can track cost per accepted change, rework, change-failure rates and whether completed user stories outpace incoming work. Those measures should be considered alongside review and security demands, so faster production does not hide a rise in defects or maintenance.
As companies set 2027 budgets, they will need to compare those outcome measures with actual AI-tool costs, including usage-based charges and any expiring discounts. A useful comparison will separate tasks and teams rather than relying on one company-wide productivity figure. The source does not identify a common measurement standard or report a future industry-wide cost outlook; whether firms can build a credible financial case remains an open question.
As an affiliate, we earn on qualifying purchases.
Key Questions
How much are companies spending on AI coding tools?
The 27 workplaces in The Pragmatic Engineer report described budgets from $150 to about $3,000 per developer per month. These are reported examples, not a representative industry average.
Does more AI-generated code mean higher productivity?
Not necessarily. Studies and companies reported different changes in pull-request output, and the Microsoft study authors cautioned that a merged PR does not automatically represent value. Teams also need to account for quality, rework, review and delivery outcomes.
Which engineering tasks showed reported gains?
Companies described gains in customer-issue triage, code archaeology, prototypes and backlog cleanup, particularly on simple, uniform technology stacks. The report does not establish that these gains apply to every organization or complex legacy codebase.
What metrics can teams use to assess AI coding value?
The report points to cost per accepted change, rework, change-failure rate and stories closed versus stories arriving. These measures can help distinguish useful delivery from higher tool consumption or raw code volume.
Source: ThorstenMeyerAI.com
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
