One in Three Companies Now Choose to Build Rather Than Buy — McKinsey Measures How Coding Agents Are Reshaping the Procurement Decision
McKinsey's annual survey found that about 30% of respondents passed on buying software because they could build it in-house with coding agents. We unpack the procurement shift from buying to building — and the current reality that productivity is up while profits stay flat.
Should you buy software or build it yourself? Coding agents are quietly rewriting this age-old decision. In its annual survey "The state of AI in 2026," published by McKinsey in August, roughly 30% of respondents said they had passed on buying a software product or feature at least once because "we can build it in-house with agentic coding tools." The survey drew on more than 1,700 respondents across industries and company sizes.
Procurement Now Starts With "Can We Build It?"
For many companies, business tools have long been something you search for, choose, and buy. Building in-house rarely made sense on cost or time, and paying a monthly SaaS fee was the more rational option.
That assumption is starting to wobble. Coding agents like Claude Code, Codex, and Cursor have lowered the barrier to implementation, letting internal teams prototype the features they need in a short time. McKinsey describes this as the "build-versus-buy" decision shifting. Rather than waiting on what a vendor offers, building something tailored to your own operations has become a realistic path.
The Gap Between Industries
The share of companies that "built instead of buying" varies by industry. Technology leads, followed by regulated industries and professional services. The figures are based on the source's tabulations.
| Industry | Share that passed on a purchase to build in-house |
|---|---|
| Technology | 41% |
| Healthcare (payers and providers) | 39% |
| Professional services | 38% |
| Energy and materials | 38% |
| Financial institutions | 36% |
| Media and telecom | 34% |
| Pharma | 33% |
Adoption momentum also splits by company size. Among large enterprises with over $1 billion in annual revenue, the share reporting a full-scale deployment of agents rose from 27% the previous year to 40%. Smaller organizations, by contrast, held flat at 22%. The companies with deeper pockets are the ones steering toward building in-house.
Productivity Is Up, but Profits Haven't Moved Yet
That said, the same survey carries numbers that make it hard to celebrate this shift unreservedly. Eighty percent of respondents said their personal productivity had improved. Yet that gain doesn't necessarily show up in company profits.
The share reporting that AI "contributed at least somewhat" to operating profit (EBIT) was 37%, unchanged from the previous year. And the "high performers" who can attribute more than 5% of profit to AI came to just 6% — also flat. There is still a gap between what individuals feel and what lands on the earnings statement. McKinsey itself frames this cautiously, saying companies are "at the stage of building a new discipline for optimizing ROI while ramping up investment."
Monthly SaaS Fees Give Way to Usage-Based Token Charges
Shifting toward building in-house also comes with a cost swap that's easy to overlook. The fixed spending once paid out as license fees turns into the token consumption behind AI inference — a highly variable, hard-to-predict expense. Just because you stop buying doesn't mean the payments disappear.
Indeed, 20% of organizations said AI operating costs were holding back their adoption. Lieven Van der Veken, a senior partner at McKinsey, argues that "operating cost is not something to defer; it should be treated as a design constraint." Precisely because building has become possible, the point is that you need the discipline to estimate — at the design stage — how much to build and what it costs to run.
The Homework for Developers and Executives
What this survey shows is a shift in which coding agents have started to reach beyond "speed of development" and into procurement decisions themselves — what to buy and what to build. On the ground, teams will increasingly find that a proposal to adopt an external tool gets reframed as "could we build this in-house first?" For leadership, the flip side of trimmed SaaS costs is that the company now shoulders variable token expenses and maintenance responsibility, and that needs to be factored in. It isn't a flashy announcement, but it's the kind of change that will make itself felt in everyday budget approvals.
References: CIO Dive: Enterprises bet on agents to build in-house software / The Register: McKinsey says enterprise AI is finally 'on the road to ROI' / India Gazette: Nearly 32% of organizations decide to build in-house software using AI coding agents / TechTimes: Record AI Spending Can't Move Earnings Needle for 94% of Enterprises
