AI is inflationary for software prices
Questioning yet another popular narrative about SaaS
One of the more persistent misconceptions I keep hearing is that, because AI makes software easier to create, software prices should come down—or that AI hurts SaaS companies’ pricing power because customers can vibe-code your SaaS and cut you out.
That’s not what I’m seeing in the wild.
In fact, AI is inflationary for software prices. Across startups, public SaaS companies, and hyper-scalers, software companies have been raising prices throughout 2024 and into 2025. If anything, price hikes will continue through 2026 and probably into 2027.
This matters because it breaks a popular narrative: that higher software supply (due to AI) kills enterprise SaaS pricing power.
That hasn’t been the case. Software vendors have found many ways to tap into AI budgets, particularly by bundling subscriptions with AI features. So far, this has more than offset slowing growth in the number of licenses.
So I’d argue we’re in some kind of temporary “AI windfall” phase for software, where price increases haven’t been met with enough new supply to bring prices down. But first, here are some representative examples of price hikes:
Microsoft effectively raised prices around Copilot by adjusting Microsoft 365 pricing and billing mechanics, resulting in roughly a ~5–6% increase for many customers once Copilot was bundled.
Notion increased its Business plan from $15 to $20 per user per month shortly after launching AI features, including Notion Agents, then gated full AI functionality behind that tier.
Salesforce announced a ~6% increase to base subscription pricing across several core products, explicitly tied to continued innovation and AI investment.
Google Workspace raised prices in early 2025, Atlassian increased 5-10% for Jira, Bitbucket, etc.
What about vertical / industrial SaaS? Well, AutoDesk, OnShape (CAD), ProCore, and practically every vertical incumbent has raised prices also.
The list goes on and on. In fact, the majority of public software companies have hiked prices in the last 18 months, sometimes multiple times.
Which brings me back to my original point: AI has become a legitimate excuse for price hikes, and customers are ok with accepting them. This is also a testament to pre-AI software companies’ pricing power, suggesting that price hikes can continue at least into 2027.
Ok, but why is this happening?
The most surface-level explanation for this pricing power is value. AI is now responsible for a meaningful share of work in domains like coding, creative production, and writing. In high-value domains like coding, customers can tolerate massive price increases without materially impacting churn.
But what’s under-appreciated is that progress in foundational models can create a windfall for SaaS applications themselves (for example, Cursor, Photoshop, AutoCAD). When coding models get better, users often attribute some of that improvement to Cursor due to a halo effect. That gives Cursor additional pricing power “for free.”
Still, the value-based argument is only half the story. There are deeper, more structural reasons why software companies can raise prices—and why customers, at least so far, are willing to pay them.
And of course, this “AI windfall” phase won’t last forever. Markets adapt, and arbitrage eventually closes.
So in this post, I’ll talk about:
the three structural reasons for price increases, aside from the value based argument, aka “AI makes software more valuable, so you can charge more”
how long I think this price hiking regime will persist,
how much “room” software companies have for further price hikes, and
when this AI windfall phase might end
🎄🎄🎄 This holiday season, gift this publication to your colleagues or friends.🎄🎄🎄
Three Structural Reasons for Price Increases
Aside from the “value” argument, there are three understated reasons why AI is causing software prices to go up.
First, aggressive bundling of AI features is effectively laundering subscription revenue into “AI revenue”.
In plain English: SaaS companies are stuffing AI features into traditional SaaS tiers (free, pro, business, etc), and raising prices 5-10%, while attributing some fraction of that as “AI”.
By bundling, I mean packing multiple AI features into a tier, and removing the option for you to 1) buy the AI feature separately, 2) or opt out of AI. This is great for optics, because bundling allows companies to reclassify existing SaaS revenue as AI revenue.
You can see this clearly with Notion, which raised prices in mid-2025 while undergoing a rebrand exercise as an AI-native product: