The iShares Expanded Tech-Software ETF dropped 22% year-to-date. In a single 48-hour window in February, $285 billion in market cap evaporated. The broader damage between January 15 and February 14, 2026: $2 trillion gone.
Atlassian reported its first-ever decline in enterprise seats. Then they laid off 1,600 people -- 10% of the company -- to "self-fund AI investments." Their stock lost more than half its value since January. Trading around $75 against a 52-week high of $242.
The market is telling you something. The per-seat subscription model is dying.
The Numbers Don't Lie
Per-seat pricing adoption collapsed from 21% to 15% in twelve months. Seventy percent of enterprise buyers now demand usage-based or outcome-based contracts. They're cutting software budgets by 30-40% and replacing dozens of tools with a handful of AI platforms.
This isn't speculation. Publicis Sapient cut their traditional software licenses by roughly 50%, replacing major creative platforms with generative AI tools. IBM Consulting reports that enterprises piloting AI orchestration agents saw 35-55% productivity improvements.
And the unit economics are brutal for incumbents. AI support agents cost $0.69 per ticket. Humans cost $37. That's a 98% reduction.
You're Funding Your Own Replacement
Here's what kills me about the small business side of this.
The average organization spends $55.7 million a year on software subscriptions across 305 applications. That's the enterprise number. But proportionally, small businesses feel it harder. You're paying per seat for a project management tool, a CRM, a help desk, an email marketing platform, a scheduling tool, a form builder, an analytics dashboard, and probably three things you forgot you're paying for.
Meanwhile, the companies selling you those subscriptions are simultaneously raising prices -- tacking on "AI premium" tiers -- while their own stock prices crater because investors know the per-seat model is broken.
You're funding the transition that makes your subscription obsolete.
One small business owner, John Sambrook, canceled 10+ subscription tools and built replacements with AI coding agents. His argument: when you can build a custom data migration tool in 15 minutes, the switching costs that kept you locked in drop to near zero. Costs down, workload down, capabilities up -- simultaneously.
The Build-vs-Buy Shift
Twenty-five percent of the current Y Combinator batch has codebases that are 95% or more AI-generated. Tiny teams are replicating features that took established software companies years and hundreds of engineers to build.
That's not a trend. That's a structural change in how software gets made.
The old math was simple: it's cheaper to buy a subscription than to build and maintain your own tool. That math assumed building was expensive and slow. It isn't anymore. Not for structured, repetitive workflows that follow clear rules.
Your invoicing flow. Your client onboarding sequence. Your content scheduling pipeline. Your lead tracking process. These aren't complex engineering problems. They're patterns. And AI agents eat patterns for breakfast.
Don't Panic-Cancel Everything
Now, the contrarian take, because I'm not going to pretend this is simple.
Gartner predicts only 35% of point-product subscription tools get replaced by AI agents by 2030. That means 65% survive. Jensen Huang called the "AI kills software" narrative "illogical."
Nearly 90% of C-suite executives said AI had no impact on workplace employment over the past three years. Sam Altman himself called out companies for "AI washing" layoffs they would have done anyway. A Wharton study found 95% of generative AI pilots failed to generate meaningful return.
So which is it?
Both. The stock crash is partly investor panic baking in a doomsday scenario. The actual seat erosion is real but more selective than the headlines suggest. The truth is somewhere between "everything changes tomorrow" and "nothing changes at all."
What Actually Dies vs. What Survives
Here's the filter I use.
Vulnerable: Any subscription tool that handles structured, repetitive workflows with clear inputs and outputs. Data entry. Report generation. Basic scheduling. Template-based communications. Status tracking. If the work follows a pattern, an AI agent can do it cheaper and faster.
Survives: Tools with deep data moats you actually need. Your accounting platform that's integrated with your bank and tax filings. Your industry-specific compliance system with years of regulatory data. Anything where the value isn't the interface -- it's the dataset underneath.
The practical move is an audit. Pull your credit card statement. List every monthly software charge. For each one, ask: does this tool handle a structured, repeatable workflow, or does it provide something I genuinely can't replace?
Most businesses I've talked to find that 40-60% of their subscriptions fall into the first bucket.
What We Did About It
We built 40+ custom internal tools. Not because we're obsessive -- because we got tired of paying monthly fees for software that did 10% of what we needed and nothing we actually wanted.
Our content pipeline runs daily: topic selection, research, writing, quality scoring, branded video generation, PDF creation, publishing, and distribution to six social platforms. No human in the loop. That replaced what would have been five or six separate subscription tools.
Our LTFI system handles everything from security assessments to business intelligence to client project management. Two people doing the work that would typically require a 10-14 person team. Not because AI is magic. Because most of those "team" roles were really just operating software that should have been automated in the first place.
Start With One
You don't need to build 40 tools. You need to build one.
Pick the subscription you resent most. The one where you're paying $30/seat/month for a feature you could describe in two sentences. Build a replacement. See what happens.
The switching costs that kept you locked into monthly software bills are collapsing. A trillion dollars in market cap already priced that in. Your subscription bill just hasn't caught up yet.
If you want to see how we consolidated our entire operation -- or if you want help figuring out which of your subscriptions are candidates for replacement -- become a free member at kief.studio. We publish resources, guides, and tools for exactly this kind of work. No commitment, no pitch. Just useful stuff from people who already did it.