The New IT Bottleneck: Why U.S. Companies Can’t Scale as Fast as Their Software

A ten-minute demo, and the whole room is sold. The new tool drafts, routes and summarizes things in seconds. Six months later, half the team still keeps a spreadsheet on the side "just in case," and the old approval chain is still running next to the shiny new one.

Sound familiar? Software has never been easier to buy, build or switch on. That's the good news. The bad news is that the company around the software hasn't sped up with it, and that gap is where growth quietly stalls.

Software got faster. Companies didn't.

McKinsey's 2026 State of AI survey shows the split clearly. About eight in ten respondents say AI has made them more productive personally. Yet only 37 percent say it has added anything to company earnings, the same share as a year earlier, even though more organizations are scaling the technology. The report also found that nearly three-quarters of the top performers had fundamentally redesigned their workflows around AI, versus about a quarter of everyone else.

Read that again. The tools work. What separates the winners is how much of the surrounding work they were willing to change.

Workflows: the part nobody budgets for

Buying software takes a purchase order. Fixing the process around it takes meetings, arguments and a few uncomfortable decisions about who does what. So, most companies bolt the new tool onto the old routine and wonder why the results look the same. If a task still passes through four approvals and two spreadsheets, a faster tool just delivers it to the bottleneck sooner.

People: change has a limit

Your employees are absorbing all this too. In an HBR piece drawing on Gartner research, the average worker went through 10 planned enterprise changes in 2022, up from two in 2016. Over the same stretch, willingness to support change fell from 74 percent to 43 percent. That data predates the current wave of AI rollouts, so it's reasonable to assume the squeeze hasn't eased.

Every new platform asks people to learn something, unlearn something and trust something. Do that too often and they start nodding in the training session and ignoring it afterward.

Hiring: you can't buy your way out

The usual fix is to add specialists. The Bureau of Labor Statistics projects software developer employment to grow 10 percent from 2025 to 2035, against 3 percent for all occupations, with a median wage of $135,980 as of May 2025. Strong demand and high pay mean good people are expensive and slow to land. Most mid-sized companies won't out-hire the problem. Many bring in a managed IT services provider to cover rollout planning, user access and day-to-day support, which frees a small internal team to work on the process itself.

Ownership: the build-it-yourself surprise

There's a newer wrinkle. McKinsey found that 32 percent of respondents had passed on buying at least one software product or feature because coding agents let them build it in-house. That can be smart. It also means more homegrown tools that somebody has to patch, secure, document and fix at 2 a.m. Every one of them needs a named owner on day one, not after the first outage.

What to do before the next rollout

Cap the change. Limit each team to one major rollout per quarter, even when three good ideas are waiting.

Retire something. Every launch should name the old process or tool it replaces, with a date. If nothing is being replaced, ask why.

Measure use, not purchase. Sixty days in, check what share of the intended users do the task in the new tool each week. That number tells you more than any vendor dashboard.

Assign owners. Each tool, bought or built, gets one person responsible for access, updates, security review and support.

The speed limit is human

Software will keep getting faster, and the vendors will keep promising the moon. The companies that scale won't be the ones with the most tools. They'll be the ones that know how much change their people can take on, and plan their IT roadmap around that number.