🔍 Read the full analysis: How To Assess AI Coding’s Costs Before Counting The Gains 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 published by The Pragmatic Engineer on October 20, based on interviews at 27 workplaces and data from two code-review vendors, found AI coding budgets ranging from $150 to about $3,000 per developer per month. Reported output gains vary sharply, so teams setting 2027 budgets are being urged to track accepted changes, rework and delivery outcomes rather than tool usage alone.
A report published October 20 by The Pragmatic Engineer found that companies surveyed spend from $150 to about $3,000 per developer per month on AI coding tools, while reported productivity results range from flat pull-request counts to substantial increases. The findings arrive as companies set 2027 engineering budgets and face questions about whether tool costs translate into delivered value.
The report’s author, Gergely Orosz, interviewed engineering leaders and senior engineers at 27 workplaces and included aggregate data from AI code-review vendors CodeRabbit and Gitar. The public portion of the report describes wide spending differences that do not appear to track company size. Examples include a publicly traded company with 1,500 engineers that caps spending at $150 per developer monthly, and a two-person European agency, Ix Technologies, that spends about $3,000 per developer.
Output measures also differ. CodeRabbit reported that developers on its platform created 113% more pull requests per week than a year earlier. That is vendor data from teams using its AI review product. A Microsoft study covering 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. The study’s authors cautioned that a merged pull request does not by itself show the value delivered.
Company accounts in the survey ranged from unchanged pull-request frequency to increases of five to ten times. In one case, frequency stayed flat while pull requests became three times larger. The source also cites Veracode testing in which 44% of AI coding tasks produced a known vulnerability. That result underscores a potential cost beyond subscriptions: review, remediation and security work when generated code is not safe or maintainable.
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
What Teams Should Measure
The findings matter because tool consumption is not the same as engineering value. A rise in tokens or pull requests can reflect more activity without showing whether changes are useful, accepted, reliable or costly to maintain. Teams that budget by usage alone risk paying more without understanding the return.
The source recommends tracking cost per accepted change, rework, change-failure rates and whether completed user stories outnumber incoming work. These measures connect spending to software delivery and quality. They also expose trade-offs: a faster first draft may add review burden, security fixes or later maintenance.
Some benefits may be easier to identify in specific tasks than through an overall productivity figure. The report describes gains in triage, code archaeology, prototypes and backlog cleanup, particularly on simple, uniform technology stacks. That suggests companies may need to compare outcomes by workflow and team rather than expect one productivity percentage to apply across the organization.
Popular right now
Ninja Foodi DualZone Smart XL Air Fryer, 10 Qt, 6-in-1, Black (DZ550)
Vitamix VX1 Professional-Grade Blender, 64 oz, Black
Roborock RockNeo Q110H Wire-Free Robot Lawn Mower
As an affiliate, we earn on qualifying purchases.
Why Budget Questions Are Rising
The October 20 report comes as companies reassess the costs of AI coding tools. According to Orosz’s account, spending among companies on API plans peaked in June and July and had roughly halved by the time of the report. He attributed that change to the arrival of Opus 5.5 and a shift toward open models. Some companies also said discounted subscriptions would soon end, potentially making their bills rise sharply if usage remained unchanged.
Pricing arrangements are changing as well. 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 explained: “Agentic usage is becoming the default, and it brings significantly higher compute and inference demands.” The shift makes forecasting harder when engineers use agents that generate more requests and consume more compute.
Evidence of productivity remains mixed. 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 a 4% change for new developers, but both confidence intervals included zero. METR also said the data was unreliable because participants held back some tasks they preferred to do without AI. Self-reported productivity gains were higher, but METR said its 2025 participants overestimated AI’s effect by 40 percentage points.
““A question that few companies seem able to answer – at least for now.””
— Gergely Orosz, The Pragmatic Engineer
Where the Evidence Falls Short
The available figures do not establish a single, reliable productivity gain for AI coding across the industry. The survey covers 27 workplaces, vendor numbers describe users of particular products, and company examples may not represent other teams. The source notes that some of the underlying report is paywalled; its cited details come from the freely readable portion.
It also remains unclear how much higher pull-request volume turns into customer value after review, testing and maintenance. Results may depend on the codebase, task, developer experience and team process. The METR estimates are uncertain, while self-reported gains may overstate measured effects. No common, independently verified measure in the source resolves these differences.
The 44% Veracode figure applies to the tasks in its latest test; the source does not provide enough detail here to generalize that rate to all AI-generated code or production software. Nor does it specify a shared baseline for comparing every company’s spending and output. Budget caps, discounts and pricing plans can also change the cost picture over time.
How 2027 Budgets Will Be Tested
Companies now setting 2027 budgets will need to decide whether to renew discounts, impose usage limits, change models or accept higher bills. The report says spending caps are common but can be easy to get around, leaving finance and engineering teams with a gap between nominal limits and actual usage. Teams facing the end of subsidies will have to identify which work justifies continued spending.
The next useful evidence will come from organizations comparing AI-assisted work with their own baselines: accepted changes, time to resolve issues, rework, failures and backlog movement. Those measures should be tracked alongside direct tool costs and the added time needed for review and security. The report does not set a universal evaluation period or threshold, so companies will have to state their methods and limitations when presenting results.
Until those internal comparisons are available, the survey offers a map of possible outcomes, not a general return-on-investment verdict. The practical question for each organization is whether its AI tools improve delivery or quality enough to justify their full cost on the work its teams actually do.
Key Questions
How much are companies spending on AI coding tools?
The workplaces in The Pragmatic Engineer report described budgets ranging from $150 to about $3,000 per developer per month. The figures are examples from a 27-workplace survey, not an industry-wide average.
Do AI coding tools increase developer productivity?
Some studies and companies report more pull requests or faster work on particular tasks, while other results are flat or uncertain. The source does not establish a single productivity gain that applies across teams.
What should companies measure before renewing AI coding budgets?
The report recommends tracking cost per accepted change, rework, change-failure rates and whether completed stories exceed incoming work. Tool usage or pull-request counts alone do not show delivered value.
What kinds of work appear to benefit most?
Accounts in the report point to triage, code archaeology, prototyping and backlog cleanup, especially on simple, uniform stacks. These are reported examples, not guarantees that the same gains will occur elsewhere.
What does the Veracode result show?
Veracode’s latest test, as described in the source, found that 44% of AI coding tasks produced a known vulnerability. The source does not provide enough detail to treat that rate as representative of all AI-generated code.
Source: ThorstenMeyerAI.com
