🔍 Read the full analysis: Is Your AI Coding Investment Delivering Value? 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 published by The Pragmatic Engineer on Oct. 20 surveyed engineering leaders and senior engineers at 27 workplaces and found AI coding budgets ranging widely, with many unable to quantify returns. Some teams report faster work in triage, code discovery and backlog cleanup, but output measures vary and security and review costs can offset gains.
Companies are spending from $150 to about $3,000 per developer per month on AI coding tools, but many still cannot quantify the return, according to a survey of 27 workplaces published Oct. 20 by The Pragmatic Engineer. The findings arrive as firms set 2027 budgets and some discounted subscriptions are due to expire, putting pressure on engineering leaders to distinguish useful output from higher bills.
The survey, by The Pragmatic Engineer’s Gergely Orosz, drew on interviews with engineering leaders and senior engineers and aggregate data from AI code-review vendors CodeRabbit and Gitar. The source article says its figures come from the portion of Orosz’s report available without a subscription. Spending did not track company size in a consistent way: a publicly traded company with 1,500 engineers capped monthly use at $150 per developer, while a two-person European .NET agency, Ix Technologies, spent about $3,000 per developer.
Measures of coding output differed sharply. CodeRabbit reported that developers on its platform created 113% more pull requests per week than a year earlier, while a Microsoft study of tens of thousands of engineers found adopters merged about 24% more pull requests over four months than they otherwise would have. Those figures use different populations and methods. Microsoft’s researchers cautioned that a merged pull request does not necessarily represent value delivered.
At the workplaces surveyed, pull-request frequency ranged from flat to increases of five to ten times. One company reported unchanged frequency but pull requests that were three times larger. The source also cites Veracode testing in which 44% of AI coding tasks produced a known vulnerability. That result concerns the tasks in Veracode’s test; it does not establish the vulnerability rate across all AI-generated code.
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
Budgets Need Outcome Measures
The gap between spending and measurable returns matters because rising usage can increase costs without showing that teams ship more useful or reliable software. The source article says some companies’ API spending peaked in June and July and later fell by roughly half, which Orosz attributes to Opus 5.5 and a shift to open models. Those are his explanations for the reported change, not a controlled comparison of model costs or productivity.
Engineering leaders need to account for more than tool subscriptions. Larger pull requests can take longer to review, and security defects or rework can reduce the value of faster code production. The article recommends tracking cost per accepted change, rework, change failure rate and whether stories are closed faster than new ones arrive, rather than treating tokens or raw pull-request counts as success.
Some teams report concrete benefits in narrower tasks. At a software company with about 600 engineers, a staff engineer said customer-issue triage that could take a day sometimes became minutes of agent work plus up to an hour of human checking. The engineer also described finding the history behind code changes in minutes rather than a day or two. These are reported examples, not independently measured results across the company.
As an affiliate, we earn on qualifying purchases.
Mixed Evidence Behind Productivity Claims
The Pragmatic Engineer report comes amid changes to both company budgets and vendor billing. The source says several companies expect discounted subscriptions to end, with some anticipating bills five to ten times higher if they do not adjust usage. It also reports that GitHub changed all Copilot plans to usage-based billing on June 1, charging tokens at each model’s API rate and removing the cheaper fallback model.
GitHub explained the change by saying, “Agentic usage is becoming the default, and it brings significantly higher compute and inference demands.” The statement describes the company’s rationale for its pricing change; it does not establish how much customers’ bills will rise in practice.
Independent research cited in the source offers a less settled picture than vendor and workplace reports. METR’s early-2025 study found experienced open-source developers took 19% longer with AI. A February follow-up estimated an 18% speedup for returning developers and 4% 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 do without AI. The source also says METR participants’ self-reported productivity gains exceeded measured effects in its 2025 research.
““Agentic usage is becoming the default, and it brings significantly higher compute and inference demands.””
— GitHub, explaining its June 1 Copilot billing change
As an affiliate, we earn on qualifying purchases.
Companywide Returns Remain Unmeasured
The available figures do not provide a common measure of productivity across the 27 workplaces. The survey combines interviews, company examples and vendor data, while the cited studies cover different developers, tasks and time periods. Higher pull-request counts cannot by themselves show whether software quality, delivery speed or customer outcomes improved.
The source article does not give a comparable accounting of subscription costs, human review time, security fixes and rework for each company. It also does not establish whether reported savings persist as teams take on more complex tasks or as discounted plans expire. Veracode’s 44% figure is a result from its test, and the article does not provide enough detail here to generalize it to all AI-assisted development.
As an affiliate, we earn on qualifying purchases.
Track Results as 2027 Budgets Set
Companies setting 2027 budgets will need to compare AI tool costs with outcomes using measures that include review, rework and failures, not just code production. The source offers no single forthcoming industry benchmark or common audit that will settle the return-on-investment question. The next useful evidence will come from companies publishing comparable results over time, including the cost of human oversight and the quality of shipped changes.
For now, the report suggests leaders should examine where their own teams see measurable gains: issue triage, understanding unfamiliar code, prototypes and backlog work on simple, consistent software stacks. Whether those gains justify higher usage or new billing plans remains a company-specific question.
As an affiliate, we earn on qualifying purchases.
Key Questions
How much are companies spending on AI coding tools?
The survey cited by The Pragmatic Engineer found budgets ranging from $150 to about $3,000 per developer per month across 27 workplaces. The figures describe surveyed examples, not a representative industry average.
Does the report show that AI coding increases productivity?
Some measures show higher output, including CodeRabbit’s report of 113% more weekly pull requests on its platform and Microsoft’s finding of about 24% more merged pull requests among adopters over four months. The measures are not directly comparable, and a higher pull-request count does not prove greater business value.
Where did teams report the clearest benefits?
The source describes reported gains in customer-issue triage, code archaeology, prototyping and backlog cleanup, especially on simple, uniform software stacks. These examples are company accounts rather than a controlled measure of results across all engineering teams.
What should a company measure to assess value?
The article recommends tracking cost per accepted change, rework, change failure rate and whether teams close more user stories than arrive. Those measures can give a fuller picture than token use or raw pull-request volume alone.
What does the Veracode vulnerability figure mean?
The source says 44% of AI coding tasks in Veracode’s latest test produced a known vulnerability. It is a test result, not a confirmed vulnerability rate for all AI-generated code or every development team.
Source: ThorstenMeyerAI.com
Evergreen bestsellers Picks
bestsellers
As an affiliate, we earn on qualifying purchases.
