TL;DR
A report by The Pragmatic Engineer surveyed engineering leaders and senior engineers at 27 workplaces about AI-tool spending and pull request (PR) output. Some platform datasets show PR volume more than doubling over a year, but the report says few companies can yet establish whether that output increase delivers a measurable return on AI spending.
The Pragmatic Engineer has reported wide variation in how much companies spend on AI tools for software engineers, alongside platform data showing sharp increases in pull request (PR) volume. But its survey of 27 workplaces found that few companies can clearly say whether higher output has produced a measurable financial return.
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
In data shared with the publication, CodeRabbit reported a 113% increase in weekly PRs among data engineers using its platform, compared with a year earlier. Its monthly PR measure rose 127%; a separate tracking group of 20,000 developers showed a 128% increase over a year. These figures describe PR counts, not verified gains in productivity, software quality or revenue.
Gitar, an AI code-verification agent, shared data from hundreds of customers showing that some teams categorized as AI-native shipped more than 60 PRs per developer per week. The report does not establish that AI tools alone caused the differences in output: team practices and company culture may also affect how frequently engineers submit PRs.
Spending also varied widely. The report describes a two-person .NET company spending $3,000 per developer per month, while a publicly traded company with 1,500 engineers capped monthly AI spending at $150 per developer. At one quantitative trading firm, reported total AI spending reached $8 million a month in September, up from $500,000 in January; average spending was reported at $1,600 per developer monthly, although traders were the largest spenders.
Why PR Growth Is Not ROI
The findings show why companies are tracking engineering output as they increase AI budgets, but also why PR volume alone cannot establish ROI. A rise in submitted changes does not by itself show that code is better, reaches users faster, reduces labor costs or creates more revenue. Those outcomes require measures beyond counts of PRs.
For engineering teams and company leaders, the practical issue is whether added spending produces benefits that justify its cost. That is harder to assess when tools are bundled in subscriptions, usage varies sharply between employees, or subsidized plans hide the full price. The report says some subsidies are ending, which could leave companies facing bills several times higher unless they change usage or pricing arrangements.
Products Worth Considering
Anker SOLIX C1000 Gen 2 Portable Power Station with 200W Solar Panel
Jackery Explorer 2000 v2 Portable Power Station with Two 200W Solar Panels
LG UltraGear 32-inch Curved QHD Gaming Monitor (2560×1440, 180Hz, 1ms, 1000R, HDR10)
As an affiliate, we earn on qualifying purchases.
How Companies Track AI Costs
The report was prompted by a software engineer asking how other companies spend on AI per engineer and whether PR throughput changed after adoption. The author spoke with engineering leaders and senior engineers at 27 workplaces and collected aggregate figures from CodeRabbit and Gitar. Most company names were withheld at their request.
Across the companies surveyed, the report found no apparent relationship between company size and per-engineer spending. Spending caps were common among medium and higher spenders, though the report says employees could often avoid those limits. June and July were peak spending months for many companies; among firms using API-based plans, costs had roughly halved from those peaks, which the report associated with model changes and greater use of open models. Claude models, including Opus 5.5 as named in the report, were frequently cited, alongside Cursor, Codex and GitHub Copilot.
““Lots of traders are using Claude. Engineering is learning the importance of rate limiting for traders who accidentally run expensive queries without realizing it.””
— An engineer at the quantitative trading firm, as quoted in The Pragmatic Engineer
What the Output Figures Cannot Show
The figures do not prove that AI caused PR counts to rise. The data covers users or customers of particular platforms, and the report does not provide a controlled comparison with otherwise similar teams that did not use AI. It also does not establish whether more PRs meant faster delivery, fewer defects or more useful features.
Company spending figures were gathered from a limited set of workplaces, with many organizations anonymized. The report does not give a standardized accounting method for comparing subscriptions, API usage, employee subsidies and spending by non-engineers. It remains unclear how many companies can connect AI costs to quality, delivery time or business outcomes.
Companies Face a Measurement Test
The report says companies are preparing to review costs as subsidies end and AI use grows. Some have spending caps or are considering cheaper models; the trading firm cited is building internal GPU and CPU capacity and developing models, according to the report. The source does not provide a completed return-on-investment assessment for these investments.
Further evidence would compare AI spending with consistent measures of delivery speed, code quality and business value over time. Until companies publish or produce those comparisons, the reported PR gains and spending totals provide a snapshot of adoption but do not establish whether AI coding tools pay for themselves.
Key Questions
What ROI did the report find for AI coding tools?
The report does not establish a clear ROI. It says few companies it spoke with could answer whether AI spending was producing a measurable return.
How much did PR output increase?
CodeRabbit reported a 113% rise in weekly PRs among data engineers using its platform and a 128% increase for a tracked group of 20,000 developers over a year. These are platform-reported output measures, not proof of equivalent productivity gains.
How much do companies spend per engineer?
The surveyed examples ranged from a $150 monthly cap per engineer at one large company to $3,000 per developer monthly at a two-person startup. These are individual examples from a limited survey, not a representative industry average.
Does higher PR volume show that AI made engineers more productive?
Not on its own. PR counts do not measure code quality, delivery time, business value or whether AI caused the increase. The report says company culture and team practices may also influence output.
Source: rss
