🔍 Read the full analysis: Are AI Coding Tools Worth The Spend? Follow The Costs on ThorstenMeyerAI.com
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
TL;DR
A report based on interviews at 27 workplaces found AI coding budgets ranging from about $150 to $3,000 per developer per month. Output results vary, and many organizations cannot connect spending to business value; the source points to accepted changes, rework and delivery outcomes as more useful measures than tool usage.
Engineering teams are spending about $150 to $3,000 per developer per month on AI coding tools, but many companies still cannot show what that spending has delivered, according to an October 23 report by ThorstenMeyerAI.com. Drawing on interviews across 27 workplaces and data from code-review vendors, the report finds that output measures range from little change to substantial increases, making the business return difficult to compare as companies set 2027 budgets.
The cost range comes from interviews and reporting by The Pragmatic Engineer, which spoke with engineering leaders and senior engineers at 27 workplaces. Examples include a publicly traded company that caps spending at $150 per developer a month and a two-person European agency, Ix Technologies, that spends about $3,000 per developer. Some companies set higher caps, but actual use and budget controls differ.
The output figures do not point to one consistent result. CodeRabbit says weekly pull requests on its platform rose 113% year over year, a vendor measure drawn from teams using its AI code-review product. A Microsoft study of tens of thousands of engineers during an early-2026 rollout found adopters merged about 24% more pull requests over four months than they otherwise would have. Microsoft’s authors cautioned that a merged pull request does not by itself measure the value delivered.
Results reported by the 27 workplaces ranged from flat pull-request counts to increases of five to ten times. The source also describes cases where larger pull requests, rather than more frequent ones, accounted for the change. Controlled evidence is less settled: METR’s February follow-up estimated an 18% speedup for returning developers and 4% for new developers, but said both estimates had confidence intervals including zero and described the data as unreliable. A separate Veracode test cited by the source found 44% of AI coding tasks produced a known vulnerability, underscoring the need to account for 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
The Bill Needs an Outcome Measure
The immediate issue for engineering leaders is not simply whether developers use AI, but whether the cost produces changes that are accepted, maintainable and useful. A rise in generated code or pull requests can add work for reviewers and security teams if quality does not keep pace. The source recommends tracking cost per accepted change, rework, change failure rate and whether completed stories outnumber new incoming work.
That distinction matters as discounted subscriptions expire and usage-based billing makes consumption more visible. The report says some companies expect their costs to rise five- to tenfold when subsidies end, though the increase is based on accounts from the companies and is not a universal forecast. Without comparable measures of business outcomes, finance and engineering teams may struggle to tell productive use from higher token consumption.
The evidence also suggests returns may depend on the work and codebase. Interviewees identified clearer benefits in triage, code archaeology, prototyping and backlog cleanup on simple, uniform technology stacks. The report says benefits are less apparent in complex legacy systems, where understanding and checking changes can remain time-consuming.
As an affiliate, we earn on qualifying purchases.
Budgets Shift With Billing Models
The report’s timing reflects changes in both company budgets and vendor pricing. The Pragmatic Engineer published its workplace survey on October 20, 2026; ThorstenMeyerAI.com says its account uses the freely readable portion. According to the report, spending among companies on API plans peaked in June and July and had roughly halved since, a change Gergely Orosz attributed to the availability of Opus 5.5 and a move to open models.
Billing structures are changing as coding tools take on more agent-like tasks. The source says GitHub changed all Copilot plans to usage-based billing on June 1, charging tokens at each model’s API rate and removing the fallback to a cheaper model. GitHub said the change reflected higher compute and inference demands as agentic use became more common. The new pricing approach makes usage levels more directly relevant to a team’s bill.
The comparison evidence has limits. Vendor platform data describes users of those products, while company interviews and internal studies use different measures and settings. METR’s early-2025 experiment found experienced open-source developers took longer with AI; its later estimates differed but carried substantial uncertainty. The available findings do not establish one productivity gain that applies across teams.
“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.
Productivity Gains Remain Uneven
It is still unclear how much of the reported increase in coding activity translates into delivered business value. The studies and workplace accounts use different methods, teams and definitions of output, and some results come from vendors’ own platforms. Microsoft’s pull-request finding is not a direct measure of product quality or customer impact, while METR said its newer estimates were unreliable.
The report does not provide a standardized cost-per-outcome comparison across the 27 workplaces. Nor does it establish that AI tools caused every reported productivity change. Costs also depend on plan terms, model choices, usage limits and whether discounts remain in place. The Veracode result identifies vulnerability risk in the tested tasks, but the source does not provide enough detail here to generalize that rate to all AI-assisted development.
AI-powered developer productivity tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Companies Set 2027 Spending Plans
Companies are setting 2027 engineering budgets while some discounted subscriptions approach expiration and vendors move toward usage-based charges. The source does not identify a common deadline or policy change across employers, so the timing of individual budget decisions remains company-specific.
The next useful evidence will be whether organizations connect tool bills to measures such as accepted changes, rework, failures and delivery against incoming work. Until companies report those outcomes in comparable ways, the current data supports a varied picture: some teams describe clear gains on specific tasks, while the overall return remains unproven for many workplaces.
AI coding assistant for developers
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
How much do companies spend on AI coding tools?
The report describes budgets ranging from about $150 to $3,000 per developer per month across examples in its reporting. Spending varies by company and tool plan; the figures are not a universal average.
Do AI coding tools increase developer output?
Some reported measures rose, including Microsoft’s finding of roughly 24% more merged pull requests among adopters over four months and CodeRabbit’s 113% year-over-year increase in weekly pull requests on its platform. Other workplaces reported flat counts, and these measures do not establish a consistent gain in delivered value.
Where do teams report the clearest benefits?
Workplace accounts cited in the report point to issue triage, code archaeology, prototypes and backlog work, particularly on simpler, uniform technology stacks. The report describes less clear returns in complex legacy codebases.
How should companies measure whether the spend pays off?
The report recommends looking beyond tool consumption and counting cost per accepted change, rework, change failure rate and whether teams close more stories than arrive. Those measures are intended to connect spending with completed work and its quality.
Is the productivity evidence conclusive?
No. The cited studies use different methods, and METR said its February estimates had confidence intervals that included zero and that the data was unreliable. The source does not establish one productivity effect that applies to all teams.
Source: ThorstenMeyerAI.com
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
