AI-native vs AI-retrofit: the growth gap incumbents should study

At a glance: The State of AI Report 2026, published on 8 October, cites preliminary second-quarter data from Standard Metrics showing that, above $20 million of annualised revenue, AI-native companies grew 172% year on year at the 75th percentile, against 53% for companies adding AI to existing software. The report states that the comparison does not establish causation. Bain & Company puts current productivity gains from AI at 20% to 27%. For incumbent CEOs, the question is what to copy, and what not to conclude.

The numbers

The State of AI Report, written by Nathan Benaich, draws on preliminary Q2 data from Standard Metrics, a platform that collects financial data from private companies. It compares two groups: AI-native firms, built around AI from the start, and firms adding AI to existing software, which this article calls AI-retrofit firms.

The comparison uses the 75th percentile, meaning the growth rate achieved by a company that outperformed three-quarters of its peers. It is a measure of strong performers, not of the typical company.

Revenue band (annualised)	AI-native, YoY growth (75th percentile)	AI-retrofit, YoY growth (75th percentile)
$1m to $20m	256%	90%
Above $20m	172%	53%

Source: State of AI Report 2026, citing preliminary Q2 2026 Standard Metrics data.

The gap holds in both size bands. In the larger band, which is more relevant to established businesses, AI-native firms grew a little over three times as fast as retrofit firms.

What the data does not prove

The report is explicit on the limits: "These comparisons do not establish that adopting AI causes faster growth." Benaich adds that he suspects it does, but the data alone cannot show it.

WCRC analysis: there are at least three alternative explanations a board should weigh before drawing lessons.

• Selection. Investors have poured capital into AI-native start-ups. Better-funded companies can hire and sell faster, whatever their architecture.
• Market timing. AI-native firms often sell into new budgets that did not exist two years ago. Retrofit firms are frequently selling into mature categories where customers already own a product.
• Base effects. Even in the above-$20m band, AI-native firms may be younger on average than retrofit firms of similar revenue, and younger companies often grow faster.

The data is also preliminary, and drawn from one provider's sample of private companies. It is a signal, not a settled finding.

What incumbents can still take from it

None of those caveats makes the gap irrelevant. They change the lesson. The question for an incumbent is not "should we become AI-native?", which is rarely possible for a business with existing customers, contracts and systems. It is "which parts of an AI-native operating model can we adopt?"

Bain & Company's 7th Global Technology Report, published on 29 September, offers useful reference points. Bain says the productivity gains captured from AI so far range from 20% to 27% across key metrics. Its survey of 293 senior technology leaders anticipated far larger improvements within one to two years: a 148% gain in release-cycle speed and a 95% uplift in developer productivity.

WCRC analysis: the distance between the 20% to 27% achieved and the expectations in Bain's survey is where AI-native firms appear to be pulling ahead. Retrofitting AI into an existing workflow tends to deliver incremental gains. Redesigning the workflow around AI is where larger gains are expected, and that is the default for a company that started with AI.

Bain also notes that leading AI labs are investing upwards of $9.75 billion in forward-deployed engineering, placing engineers inside customer organisations to make the technology work in practice. The signal for incumbents is that even the companies building the models do not expect value to arrive by licence alone.

The pressure on software incumbents

The State of AI Report documents how markets are already pricing this shift. It says nearly $285 billion of software market value was wiped out in two weeks in February, an episode it labels the "SaaSpocalypse". It also reports that Anthropic's internal index shows Claude leading 26% of measured model R&D work in August, up from under 1% in February, an indication of how quickly AI is changing work inside AI companies themselves.

WCRC analysis: for incumbent software and services firms, the risk is not a single competitor but a change in the expected growth rate of the category. If investors and customers come to expect AI-native growth, a 53% top-quartile growth rate, strong by historical standards, may start to look slow.

What it means for leaders

• Do not read 172% versus 53% as proof. The State of AI Report says the comparison does not establish causation. Treat it as a reason to examine your model, not a forecast for your business.
• Redesign one workflow end to end. The difference between Bain's 20% to 27% captured gains and leaders' expectations suggests bolt-on AI underdelivers. Pick one process and rebuild it around AI.
• Fund deployment, not just licences. Leading labs are investing heavily in engineers who work inside customer businesses. Budget for the same internal capability.
• Benchmark against AI-native peers. Track how fast AI-native competitors in your category are growing, and what they charge, not only your traditional rivals.
• Report productivity in board terms. Measure release speed, cost to serve and revenue per employee before and after AI changes, so the evidence is your own.

Quick answers

Q: How much faster are AI-native companies growing? A: The State of AI Report 2026, citing preliminary Q2 Standard Metrics data, shows 172% year-on-year growth at the 75th percentile for AI-native firms above $20m revenue, against 53% for firms adding AI to existing software.

Q: Does the data prove AI causes faster growth? A: No. The State of AI Report states that "these comparisons do not establish that adopting AI causes faster growth", although its author says he suspects it does.

Q: What productivity gains are companies getting from AI today? A: Bain & Company's 7th Global Technology Report says current productivity gains from AI range from 20% to 27% across key metrics, below the gains surveyed technology leaders expect within one to two years.