🔍 Read the full analysis: Is AI Coding A Smart Investment? Measure The Costs And Gains on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- 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 finds AI coding spend ranges from about $150 to $3,000 per developer each month, with no consistent link between company size and spending. Output measures vary, and teams still lack reliable ways to connect tool use to delivered value; reported gains are strongest in triage, code archaeology, prototyping and backlog work on uniform systems.
Engineering teams are paying from about $150 to $3,000 per developer per month for AI coding tools, but many cannot show what that spending has produced, according to reporting by Gergely Orosz of The Pragmatic Engineer. His October 20 report, based on interviews at 27 workplaces and data from two code-review vendors, arrives as companies set 2027 budgets and some discounted subscriptions near expiration.
The spending figures vary widely and do not track simply with company size. One publicly traded company with 1,500 engineers caps spending at $150 per developer monthly; a two-person European .NET agency reported spending about $3,000 per developer. Other companies described caps between $1,000 and $4,000, with typical monthly use around $1,000. The report says spending caps are common, but can be bypassed.
Reported output measures also differ. CodeRabbit, an AI code-review vendor, says developers on its platform produced 113% more pull requests per week than a year earlier. Microsoft research covering tens of thousands of engineers during an early-2026 rollout found adopters merged about 24% more pull requests than they otherwise would have over four months. Microsoft’s authors cautioned that a merged pull request does not necessarily represent delivered value.
Other companies in Orosz’s reporting saw flat pull-request frequency or increases of several times. In one case, the number of pull requests stayed flat while their size tripled. Those figures do not establish a common productivity effect across teams: they come from different organizations and measurement methods, and vendor platform data describes users of that vendor’s product.
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
Budgeting Beyond AI Usage
The reporting matters to companies deciding whether to renew tools, expand access or accept higher bills. Usage and output counts are not the same as business value: more code or more pull requests may also mean more review work, rework or risk. Teams need to connect spending to work that is accepted, maintained and useful.
The article points to measures such as cost per accepted change, rework, change failure rate, and stories closed compared with stories arriving. These do not prove AI caused an improvement on their own, but they can give managers a more practical basis for comparing costs and results than token consumption alone. The stakes are higher where discounted plans are ending and usage-based billing can make costs less predictable.
Quality is another part of the calculation. The source cites Veracode testing in which 44% of AI coding tasks produced a known vulnerability. That finding is a test result, not a measurement of vulnerabilities in every company’s code. It nonetheless underlines why review and security costs must be counted alongside speed or volume.
As an affiliate, we earn on qualifying purchases.
Why AI Coding Bills Are Shifting
Orosz’s October 20 report combines interviews with engineers and engineering leaders at 27 workplaces with aggregate information from CodeRabbit and Gitar. The source article says its cited material comes from the report’s freely readable section. The companies’ varied results make it difficult to treat any one productivity number as a general forecast.
Spending reportedly peaked in June and July and has roughly halved since among companies on API plans, according to Orosz, who attributes the change to Opus 5.5 and movement toward open models. Separately, GitHub changed its Copilot plans on June 1 to usage-based billing, charging for tokens at each model’s API rate and removing a fallback to a cheaper model. GitHub said agentic use brings higher computing and inference demands.
The source also describes a quant trading firm whose AI spending rose from $500,000 a month in January to $8 million in September. Its largest spenders were traders, and the firm evaluates AI as part of desk-level profit and loss. That is a specific example, not a typical company spending pattern; most workplaces discussed in the report do not have an equivalent direct scoreboard.
“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 and Quality Still Vary
The evidence does not settle how much AI coding increases delivered value across the industry. The report’s company examples range from flat pull-request counts to much larger increases, while the underlying tasks, teams and measurement periods differ. Pull-request volume is an imperfect proxy, particularly when changes grow larger or require additional review.
Controlled research cited in the source is also 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 a 4% speedup for new developers, but both confidence intervals included zero. METR called the data unreliable, in part because 30% to 50% of developers held back tasks they did not want to do without AI.
The source does not provide a single industry-wide cost-benefit estimate, nor does it establish that the reported gains caused improvements in revenue, customer outcomes or software quality. Company-level spending and results may also change as subscriptions, model choices and billing plans change. Whether gains persist after human review and security work remains a key question.
As an affiliate, we earn on qualifying purchases.
What Teams Should Track Next
The immediate test for engineering leaders is whether their 2027 budgets can be tied to outcomes rather than activity. The report recommends tracking cost per accepted change, rework, change failure rates and backlog flow, then comparing results across teams and work types. These measurements should include review and security effort, not just time spent generating code.
Companies facing the end of discounted subscriptions will need to decide whether to limit usage, switch models, renegotiate plans or fund broader access. Results from triage, code archaeology, prototypes and backlog cleanup may not transfer to complex legacy systems. The next useful evidence will be sustained, team-level measurements that show whether changes are accepted and maintained, and what they cost after review.
As an affiliate, we earn on qualifying purchases.
Key Questions
How much are companies spending on AI coding tools?
The workplaces in The Pragmatic Engineer’s reporting described spending from about $150 to $3,000 per developer per month. These are examples from 27 workplaces, not an industry-wide average.
Does the reporting prove AI coding increases productivity?
No. Some studies and companies report higher output, while others report flat measures or uncertain results. Pull-request counts are not the same as delivered value, and the evidence uses different methods and teams.
Where did teams report the clearest benefits?
The source describes benefits in triage, code archaeology, prototyping and backlog cleanup, particularly on simple, uniform technology stacks. These are reported examples, not a guarantee that every team will see similar results.
What should a company measure before expanding its AI budget?
The article suggests tracking cost per accepted change, rework, change failure rate, and stories closed compared with new stories arriving. Teams should also account for review and security work when evaluating costs.
What remains uncertain about AI coding costs?
There is no single industry-wide estimate of return on investment in the reporting. The durability of productivity gains, their effect on software quality, and the full cost of reviewing AI-generated work remain unclear.
Source: ThorstenMeyerAI.com
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
