McKinsey Survey: 32% of Respondents Say Their Organization Skipped Buying Software It Could Build With AI — Run the Build-vs-Buy Math Before You Renew

McKinsey Survey: 32% of Respondents Say Their Organization Skipped Buying Software It Could Build With AI — Run the Build-vs-Buy Math Before You Renew

2026-10-11

Takeaway: In McKinsey’s State of AI survey, published in August 2026 (1,719 respondents, fielded May 4 to June 8, 2026), 32% of respondents said their organization had decided against buying at least one software product or feature because it could be built in-house with agentic coding tools. That’s a self-reported purchase decision, not proof that custom builds paid off. Before you cancel a subscription, price the upkeep.

What the survey found

McKinsey published “The state of AI in 2026: On the road to ROI” on August 25, 2026 (McKinsey, Aug 25, 2026; Michael Chui, McKinsey senior fellow and report co-author, LinkedIn, Aug 25, 2026). That’s about seven weeks ago, so it isn’t breaking news, but if you’re planning next year’s software budget, one number in it deserves a hard look.

The key line: “Nearly a third of respondents (32 percent) report that their organizations have decided against buying one or more software products or features because they could be built internally with agentic coding tools” (McKinsey, Aug 25, 2026).

“Agentic coding tools” are AI assistants that don’t just suggest code but write, run, and fix it across multiple steps with limited supervision.

How big is the study, and what does the 32% measure?

  • Size: an online survey of 1,719 participants in 97 countries, fielded May 4 to June 8, 2026, with results weighted by each country’s share of global GDP (McKinsey, Aug 25, 2026).
  • Who answered this question: McKinsey asked it only of respondents whose organizations regularly use AI in at least one business function, so the 32% is a share of those respondents (McKinsey, Aug 25, 2026).
  • The unit: “one or more software products or features” (McKinsey, Aug 25, 2026). As an analysis by Digital Applied points out, a company that built one dashboard instead of buying a license counts the same as one that replaced a whole platform (Digital Applied, Sep 1, 2026).
  • No trend line yet: the report gives only a 2026 figure for this question, with no prior-year comparison (McKinsey, Aug 25, 2026). As of October 2026, this August edition is the latest McKinsey State of AI survey, and we found no newer data on this question.
  • By industry: respondents in technology (41%) and healthcare payers and providers (39%) reported it most often, while insurance (19%) and the public and social sector (17%) were lowest. Some industry groups are small (as few as 40 insurance respondents), so treat the gaps as rough (McKinsey, Aug 25, 2026).
  • By AI maturity: among AI “high performers” (the roughly 6% of respondents who attribute at least 5% of EBIT to AI and call its impact significant), nearly half reported such a decision, compared with 31% of other respondents (McKinsey, Aug 25, 2026).

The part that should slow you down

The same survey found that 37% of respondents attribute at least some EBIT (earnings before interest and taxes, roughly operating profit) impact to their organization’s AI use, about the same share as last year (McKinsey, Aug 25, 2026). It also found that about 20% of respondents said AI-related operating costs, including token costs, constrained their AI use (McKinsey, Aug 25, 2026).

And size matters. Among respondents from organizations with $1 billion or more in revenue, the share reporting that their organization is scaling AI agents rose from 27% last year to 40%; for organizations under that size it stayed essentially flat at 22% (McKinsey, Aug 25, 2026). The report doesn’t break the 32% out by company size, and its “smaller” group covers everything under $1 billion in revenue, so a 40-person firm should not assume the same math.

A build-vs-buy checklist for owners and operators

Build when…

  • The tool is narrow and internal (a report, an intake form, a simple workflow).
  • The SaaS product makes you pay for 50 features to use three.
  • Your process is a genuine edge that off-the-shelf tools force you to abandon.

Buy (or keep buying) when…

  • It handles money, payroll, compliance, or customer data at scale.
  • It needs uptime, security patches, and support you can’t staff.
  • The vendor carries certifications your customers or auditors require.

Price the whole life, not the first version

AI makes the first version cheap. The ongoing costs don’t disappear:

  • Ownership: who fixes it at 9 p.m. when it breaks?
  • Security: who reviews AI-written code for access control and secrets handling?
  • Running costs: hosting plus AI usage fees, which about one in five McKinsey respondents already flag as a constraint.
  • Exit plan: if the person who built it leaves, can someone else maintain it?

Start with one renewal

Pick the next subscription coming up for renewal, list what you actually use, and get a rough build estimate that includes a year of maintenance. Compare that, not the sticker price, to the license.

The bottom line

McKinsey’s number says many organizations now treat building as a real option, not that it’s always the right one. The winners will be the businesses that make the call one tool at a time, with eyes open about upkeep and security.

If you’re weighing a custom build against a renewal, Yellow Coop’s fractional CTO and technology project team can help you scope it, estimate the full cost, and build it safely if it makes sense.

Internal links: Innovate, What We Do, How We Engage, Contact, Insights.

Sources