32% of Companies Now Build Software Instead of Buying It — Here's Why That Matters

Build vs. Buy: The Ultimate Guide to Your Company's Software Strategy

A new global survey shows AI coding agents are quietly rewriting the build-vs-buy decision that has favored SaaS for over a decade.

For most of the last ten years, the answer to "should we build this or buy it?" was almost always the same: buy it. SaaS subscriptions were cheaper than in-house engineering, faster to deploy, and easier to justify to a CFO.

That assumption is now being tested — and in a growing number of companies, it's losing.

A new global survey from McKinsey puts a real number behind a trend many founders and product teams have only sensed anecdotally: AI coding agents are making "build it ourselves" a genuinely competitive option again.

The Headline Number: 32%

McKinsey's State of AI: Global Survey 2026 surveyed 1,719 business leaders across 97 countries between May and June 2026. One finding traveled faster than the rest of the report.

According to McKinsey's own summary, nearly a third of organizations — 32% — report they decided against buying one or more software products or features because they could be built internally with agentic coding tools instead.

Coverage from Yahoo Finance adds useful context: this trend isn't spread evenly. It's most pronounced among the "high performers" McKinsey identifies — the roughly 6% of respondents who say AI already contributes at least 5% of their EBIT. Nearly half of these high performers have skipped a software purchase in favor of building, compared with 31% of everyone else.

Which Industries Are Leading The Shift

The build-vs-buy shift isn't uniform across sectors. Some industries are moving far faster than others.

Per the same Yahoo Finance breakdown, the technology sector leads at 41%, followed by healthcare payers and providers at 39%, with professional services and energy close behind at 38%.

On the other end of the spectrum, a separate analysis from THE D*AI*LY BRIEF notes that insurance and the public sector sit far lower, at 19% and 17% respectively — sectors where buying software often means buying audit trails and liability coverage, not just code.

That distinction matters. Build-vs-buy isn't just a cost question anymore. In regulated industries, "buying" still comes with compliance guarantees that an internal build doesn't automatically include.

Why This Is Happening Now

The short answer is that agentic coding tools crossed a usability threshold. Writing a working internal tool no longer requires a dedicated engineering sprint — for a growing number of use cases, it takes a capable coding agent and a clear brief.

Reporting from Startup Fortune quotes McKinsey senior partner Lieven Van der Veken describing this as a deliberate strategic pivot — leaders who once assumed AI work required outside partners are now asking what their own teams should build themselves.

Retool's own 2026 Build vs. Buy report backs this up independently: 35% of enterprises say they've already replaced a SaaS tool with something they built in-house, according to the same Startup Fortune coverage.

This isn't a fringe experiment anymore. It's becoming a standard part of how technology budgets get evaluated.

The Catch: Building Is Easy, Owning Is Not

Here's where the story gets more complicated — and more useful for anyone actually making these decisions.

The same McKinsey survey found that the share of companies attributing any EBIT impact to AI held flat at 37% year-over-year, even as build activity rose. Analysis from Under30CEO frames this gap directly: individual teams are shipping more, but the profit line hasn't caught up yet — a warning about long-term maintenance costs, not proof that the trend is hollow.

THE D*AI*LY BRIEF makes the point even sharper: your engineering team can now build the thing you were about to renew a subscription for — but that is not automatically the same as saving money. An internally built tool still needs updates, security patches, and someone accountable when it breaks at 2 a.m.

In other words, agentic coding tools have solved the "build" cost. They haven't solved the "own" cost.

A Wider Pattern: Build, Buy, or Borrow

McKinsey isn't the only firm tracking this shift. According to KPMG's global research, more than 40% of companies have already integrated AI agents into their workflows, up from just 11% in early 2025 — and the build-vs-buy-vs-borrow decision is now treated as a genuine strategic question rather than a purely technical one.

That framing — build, buy, or borrow — is a more useful mental model than a binary build-or-buy choice. "Borrowing" covers things like embedding a vendor's agent inside your own workflow rather than either building it from scratch or paying for a full standalone product.

How Big Enterprises Are Scaling This Differently

Company size changes the shape of this trend too, not just industry.

Large enterprises with more than $1 billion in revenue are scaling agentic tools noticeably faster than the average. Reporting from Yahoo Finance notes that 40% of these large enterprises are now scaling agents in one or more business functions — a jump from 27% just a year earlier.

Smaller organizations, by contrast, have stayed roughly flat year over year according to the same coverage. That gap suggests the build-vs-buy shift is currently being driven disproportionately by companies with the internal engineering capacity to actually operationalize agentic coding at scale — not by resource-constrained smaller teams, at least not yet.

For smaller businesses, that's actually useful context: the tools are becoming more accessible, but the organizational muscle to use them well is still catching up, even at large companies with dedicated engineering budgets.

What This Means If You Run a Small Business or Blog

You don't need to be a Fortune 500 company for this shift to matter. If anything, it matters more for small teams and solo founders, because the labor cost of "just build it" has dropped the furthest for people who previously couldn't afford to build anything at all.

A few practical takeaways:

• Before renewing a SaaS subscription, ask whether the specific feature you actually use could be built as a small internal tool with a coding agent.
• Factor in maintenance, not just build time. An agent can write the first version in an afternoon — someone still has to own it long-term.
• Watch for regulated or compliance-heavy tools. Those are the cases where buying still tends to win, based on the industry data above.
• Treat "build vs. buy" as a live decision, not a one-time one. The economics are shifting fast enough that a tool worth buying in 2025 might be worth building by 2027.

The Bigger Picture For SaaS And Software Vendors

If you sell software rather than buy it, this data is a signal worth taking seriously. As one analysis from Digital Applied puts it bluntly: agentic AI has collapsed custom-build costs, so more firms now win by owning tools rather than renting SaaS.

That doesn't mean SaaS is disappearing. Complex, compliance-heavy, or deeply integrated products still carry real value that a quick internal build can't replicate. But the pool of "simple enough to build ourselves" features is growing every quarter — and vendors whose entire value proposition lives in that shrinking pool are the most exposed.

Final Thoughts

The build-vs-buy decision hasn't been this genuinely open in over a decade. Agentic coding tools have made "build it ourselves" a real option for a much wider range of companies — not just tech giants with in-house engineering teams.

But the data also comes with a clear warning: shipping fast and owning software long-term are two different challenges, and the second one is where most of the real cost lives.

Whether you're evaluating a $50-a-month tool or a six-figure enterprise contract, the right question in 2026 isn't just "can we build this?" It's "are we prepared to own it after we do?"