Why software services companies must prepare for a pricing reset
A Rule I Lived By for Two and half Decades
Early in my career, I spent weeks buried in estimation spreadsheets. The gospel back then came from Capers Jones the researcher who gave us cost-per-line-of-code models and language productivity tables. Every estimation exercise I did referenced his work in some form.
One number stayed with me for over twenty years: a good developer should not write more than 25 lines of production code per day.
Not because developers are slow typists. But because those 25 lines had to be thought through the logic validated, edge cases considered, tests written, code refactored for clarity. Quality software was expensive precisely because human thinking is slow and careful.
That single number 25 lines a day was the invisible foundation under every proposal I ever wrote, every rate card I ever negotiated, every project I ever estimated. Multiply the lines needed by the cost per developer-day, layer in testing and management overhead, and you had your price.
Simple. Reliable. And as of last year, completely irrelevant.
A Conversation That Changed My Thinking
A few months ago, I was having coffee with a friend who leads delivery at a major Indian IT services firm. Casually, almost as an aside, he mentioned something that stopped me mid-sip.
“We revised our recent proposals down by a third,” he said. “And the teams are still delivering the same scope for less than the revised cost.”
Let that sink in. Not a small trim. Not a 5–10% efficiency gain. The proposals were cut by a third and the actual delivery came in under even that reduced number.
The reason? AI coding assistants had changed the math. His developers were using Copilot, Claude, and other tools to generate boilerplate, write tests, refactor legacy code, and scaffold entire modules. The 25-line-a-day developer was now part of a human-AI team producing multiples of that and the quality was holding up.
I walked out of that coffee meeting and started digging. What I found confirmed that my friend’s experience was not an outlier it was the new normal.
I’ve Seen This Movie Before , With Mobile Phones
In 2003, I bought a very basic mobile phone for ₹15,000. It could make calls, send SMS, and had a tiny monochrome screen. That’s it.
Today, for the same ₹15,000, I can buy a phone with a high-resolution camera, GPS, AI assistant, 128GB storage, and more computing power than the machines that sent Apollo to the moon. The price stayed the same. The value exploded.
That’s what automation does. It doesn’t just make things cheaper it resets what buyers expect for their money.
AI-generated code now costs a fraction of a cent per line compared to the dollars-per-line economics that the entire IT services industry was built on. Whether you agree with lines-of-code as a metric or not (and many don’t), the directional message is undeniable: the cost of producing software is collapsing.
The Market Isn’t Waiting for a Memo
If my friend’s anecdote felt like one data point, the market has provided thousands more.
Indian IT stocks saw their steepest monthly decline since the 2008 crisis earlier this year. Jefferies downgraded most large Indian IT firms and cut price targets by up to a third. The five biggest firms including TCS, Infosys, HCLTech, Wipro, Tech Mahindra have shed nearly 58,000 employees over two years, after adding over 350,000 in the year before that.
Analysts now project a 14–16% drop in IT services pricing, with up to 40% of current revenue vulnerable to AI-driven deflation. Vinod Khosla went as far as predicting that IT outsourcing could be “almost completely gone by 2030.”
You can dismiss that as hyperbole. But when Goodfirms reports that 91% of software companies are already using AI to cut development costs, and 61% expect project budgets to drop by 10–25%, this isn’t a forecast anymore. It’s a weather report.
“But More Software Means More Work for Us, Right?”
This is the counterargument I hear from optimists, and there’s real truth in it. Cheaper code production will expand the market. More software will be built. AI services spending alone is projected to reach $985 billion by 2030. The pie is undeniably getting bigger.
But a bigger pie doesn’t guarantee bigger slices for everyone at the table.
Go back to my mobile phone example. The smartphone market grew from near-zero to billions of users. But that growth destroyed Nokia and BlackBerry. The new value went to Apple, Google, and the app economy not to the incumbents who built the previous generation of phones.
The same pattern is unfolding in IT services. The companies that capture the AI-era value will not necessarily be the ones who dominated the billable-hours era.
The Quality Objection, Real, But Temporary
Skeptics raise a fair point: AI-generated code isn’t perfect. Studies show quality concerns, higher code churn, and security vulnerabilities. One rigorous trial even found experienced developers were slower when using AI tools on complex, familiar codebases.
I take these findings seriously. But here’s what they don’t change: even if AI code needs human review and refinement, you still need far fewer people to deliver the same project. Where you once staffed 10 developers for six months, you might now need 4 developers and AI agents for three months. The pricing impact is the same whether AI writes perfect code or merely decent code that humans polish.
And the models are improving at a pace that makes today’s limitations feel like complaining about the camera quality on a 2007 iPhone.
So What Should Services Companies Do?
TCS recently told its employees something remarkable: use AI to deliver work faster and cheaper — even if it cannibalizes near-term revenue. That takes courage. It also shows that the smartest leaders in the industry already see where this is going.
The playbook, as I see it, has three moves:
Accept the deflation — companies that hide productivity gains to protect billing rates will lose clients to those who pass savings through.
Move up the value chain — shift from selling developer-hours to selling outcomes, platforms, and AI-augmented delivery models. And
Reshape the talent model — fewer junior coders writing boilerplate, more architects and AI-supervisors designing systems that agents execute.
The firms that make this pivot won’t just survive the pricing reset — they’ll define the next era.
Back to Those 25 Lines
I spent two and half decades building estimates on the assumption that a developer writes 25 lines of production code a day. That assumption shaped proposals worth millions. It shaped careers. It shaped an entire industry.
That assumption is now a relic. AI agents work around the clock, don’t need onboarding, and produce code at a fraction of the old cost. Software services companies that are not preparing for significant price cuts over the next 3–5 years are in denial.
The question isn’t whether software pricing will fall. It’s whether your organization will be the one setting the new price — or the one scrambling to match it.
• • •
Are you seeing this pricing pressure in your organization? Have your proposals changed? I’d love to hear from leaders navigating this shift.
First published on LinkedIn.