State of AI Report Puts Software That Added AI Behind Companies With None

Air Street Capital published the ninth State of AI Report on October 8. Its revenue headline belongs to the frontier labs, with OpenAI and Anthropic at a combined run rate of about $105 billion by late summer, up from about $30 billion at the start of the year. The slide with more bearing on a software company's 2027 plan comes eleven slides later. The report uses preliminary second-quarter data from Standard Metrics, which collects the financials private companies report to their investors. It finds AI-native companies with $1 million to $20 million in annualized revenue growing 256% a year at the 75th percentile. Existing software businesses that have added AI grew 90%. Above $20 million the split is 172% against 53%.

That's the comparison the report headlines, and it's the one boards will repeat. The chart under it carries a third line the headline leaves out. Standard Metrics also plots companies with no AI in the product as a reference group, and at the 75th percentile they grew 148% in the $1 million to $20 million band and 88% above $20 million. Software that added AI sits below them in both bands. The next slide splits the data by founding year and by customer type. AI-enabled software trails the non-AI group in all six cuts, including 60% against 123% for companies selling to enterprises and 82% against 116% for SMB and mid-market sellers.

Nathan Benaich, who writes the report, is careful about what the numbers prove. The essay that launches it says the comparisons don't establish that adopting AI causes faster growth, "but I suspect that it does." The chart supports something narrower than either half of that sentence. Among the fastest-growing quarter of companies on the platform, adding AI to an existing product hasn't put a company ahead of peers that added none. A CRO at an AI-enabled company will have to answer that in planning.

Software that added AI trails the no-AI group in every cut the report shows

The pattern holds at every company age. Among companies founded before 2017, the top quartile of AI natives grew 49%, the non-AI group 45% and AI-enabled software 31%. For companies founded in 2020 or later, the figures are 487%, 272% and 199%. Age is the obvious confounder, since a business old enough to have existing software to add AI to is usually older than one built on AI. Splitting by founding year doesn't close the gap. Those panels pool four quarters, the third quarter of 2025 through the second quarter of 2026, for companies with at least $1 million in annualized revenue.

The figure with the most bearing on a forecast is about staying power. Among enterprise-focused companies in the top growth quartile in the first quarter of 2025, 42% of AI natives were still there a year later. For AI-enabled software the share was 23%, and for non-AI companies 28%. Of the three groups, a strong year at an AI-enabled company was the least likely to repeat.

A separate survey points the same way on the native-versus-enabled gap, though it can't speak to companies without AI. High Alpha's 2025 SaaS Benchmarks survey, with more than 800 responses, found AI-native companies growing faster than AI-enabled ones in every ARR band. The gap ran from 110% against 40% at $1 million to $5 million down to 38% against 16% above $50 million. Fewer than 1% of its respondents said their product had no AI at all, so having AI no longer separates anyone. High Alpha also reports identical median gross and net revenue retention for the two groups. If existing customers stay and expand at the same rates, the growth gap must be opening in new business, where AI-native products are winning more first purchases.

Survivorship and self-labeling could explain part of the gap

The strongest objection comes from Standard Metrics itself. Its 2025 Year in Review, published in April with a simple AI or non-AI split, found the median AI company above $20 million growing 30% in the fourth quarter of 2025. Non-AI companies grew 27%, which Standard Metrics called effectively parity. The gap at that size showed up in the tail, 262% against 211% at the 90th percentile. Below $20 million, AI companies did lead at the median, 73% to 39%. The State of AI chart reports only the 75th percentile, which is where the AI story is loudest, and it labels the second-quarter data preliminary.

The sample has its own lean. Standard Metrics draws on companies that report financials to their investors through its platform, which in practice means venture-backed companies. Its methodology notes that the mix shifts as companies join, leave, get acquired or shut down. An AI-native startup that dies stops reporting. AI status is partly self-declared at onboarding, too. A fast grower has every reason to call itself AI-native, and an incumbent whose growth has stalled has every reason to add an AI feature. That runs the causation backward, from slowing growth to AI adoption, and nothing in the report rules it out. The slides also don't define the non-AI reference group, which may include businesses that don't sell software at all.

Timing is another caveat in the April report. In the $1 million to $20 million band, the AI premium at the median disappeared for about 18 months before returning in late 2024. Standard Metrics read this pattern as an enterprise demand cycle rather than a permanent edge. A gap that tracks budgets can narrow when budgets turn.

Most of these biases inflate the AI-native line. The one that cuts against AI-enabled software is reverse causation, and it's no comfort to an incumbent either. It says companies added AI because growth was slowing, and the add-on hasn't lifted them back above peers that never added it.

What this means for CROs at AI-enabled software companies

Boards have a reason to ask. The same report records what public investors did to software this year. After Anthropic shipped Claude Cowork in January and added plugins for legal, marketing and finance, investors repriced the terminal value of SaaS revenue. Nearly $285 billion of software market cap went in early February. The S&P 500 software index fell 26% from its October peak before recovering 44% by September. The report's summary calls the episode one that "exposes uncertain winners." The private data above is an early measure of which winners those are, and it doesn't put AI add-ons among them.

Expect the 256% to show up in board decks this planning season, and get ahead of it by benchmarking against the right cell. A software company founded in 2018 that sells to enterprises belongs next to the cohort and segment figures for its own kind. The 487% belongs to AI natives founded in 2020 or later. Say which percentile you're quoting, because above $20 million the 75th percentile and the median tell different stories. Boards are hearing a related message from the people who buy whole companies, with private equity now paying barely more for SaaS than for other businesses.

Plan for regression if this year went well. The 23% persistence figure says a top-quartile year at an AI-enabled company repeats less often than one at a company with no AI. So a 2027 number that extrapolates this year's growth needs a stated reason the pattern won't apply.

Report AI as its own line. When AI features are bundled into the existing price, they can't show up in growth, and nobody can tell whether they're earning anything. Track AI revenue and the attach rate on new deals. Also track the win rate in deals where the buyer looked at an AI-native competitor. If High Alpha's retention parity holds, new business is where the gap opens, so that win rate is the number to put in front of the board.

Then decide whether the AI changes what the customer is buying. The report's split is AI as the product against AI as a feature of a product, and a roadmap made of features keeps a company in the second group, the one the data has growing slowest.

My bet is that Standard Metrics' final second-quarter figures, and the third-quarter figures after them, keep AI-enabled software below the non-AI group at the 75th percentile above $20 million. If the order flips, the add-on story deserves a second hearing. If it holds, "we shipped AI" stops working as an answer to a board's growth question.

Among the fastest-growing quarter of companies on the platform, adding AI to an existing product hasn't put a company ahead of peers that added none.

Report AI as its own line, with AI revenue, attach rate on new deals, and win rate against AI-native competitors, and benchmark growth against your own founding-year and customer-segment cohort.

By Business of GTM