The internet is full of hot takes claiming that “Claude code just killed this app category,” or that “the era of SaaS is over.” In fact, over the last 18 months, this narrative has been repeated so frequently that it is widely accepted as fact. But if SaaS were dying, then shouldn’t SaaS adoption numbers fall as well?
Instead, Torii’s recent 2026 SaaS Benchmark Annual Report shows that SaaS adoption is rising faster than anticipated. The report shows that organizations now run an average of 831 apps, with large enterprises averaging 2,191. The average employee touches 40 different applications for work. If SaaS were dying, this report should be an obituary, but instead it shows a category outgrowing its old label and definitions, which no longer fit.
Historically, we’ve associated SaaS with large applications and big valuations: Salesforce, Adobe, Hubspot, and Zoom. Adoption of these tools means lots of seats and lots of clicks in a UI. But today, we’re seeing valuations crumble as more and more employees turn to AI solutions and prompting to get more of their work done. What’s really dying is the old UI tax: the rule that value has to be delivered within a branded interface, by a human, one seat at a time. In fact, employees are now bypassing IT altogether, breaking our traditional models of governance.
Also Read: AiThority Interview with Glenn Jocher, Founder & CEO, Ultralytics
Torii’s data shows only 15.5% of apps in the modern stack are officially sanctioned tools; the majority, 61.3%, are shadow IT. Software is not disappearing; it’s escaping the tidy boxes we used to count it.
And this is where the story gets fuzzier. The popular narrative is that AI kills SaaS, but our research shows that SaaS is actually the primary vehicle for AI. Today, AI and shadow IT are often one and the same. Torii found that 26 of the top 50 shadow IT apps discovered in 2026 were pure-play AI tools. Additionally, nearly 700 new AI-native apps were adopted in enterprise environments in 2025. AI did not consolidate the modern stack; it makes the long tail explode. In many ways, the death of SaaS is a story of visibility. If you count only apps that identity and procurement can see, the stack looks tidy. Count the tools employees actually use through browser activity, direct signups, and unmanaged usage, and the number gets unruly fast.
That is why the seat-based model suddenly looks blunt. A seat tells you who can log in, but it says much less about what work is being done within, or outside of, that app. As software does more and more of the work, charging for a human presence inside a tool starts to look like billing a factory by how often the worker visits the boiler room.
The Democratization of Power
Today’s SaaS, AI, agentic world actually looks a lot like the distribution of energy during the second industrial revolution. In the late 1800s, most factories used a single central steam engine. It sat there, a soot-covered heart, beating in the basement and powering the whole operation through a series of belts and shafts. The building’s layout bent around it, the maintenance bent around it, work itself bent around it.
Then, electricity arrived. Suddenly, energy came from central sources through a series of cables. Rather than rely on basement behemoths, operations could “plug in to” a supply generated elsewhere. Power was more abundant than ever, but what changed was the delivery service. The steam engine did not lose because factories stopped needing energy; it lost because energy became so distributed, flexible, and easier to route to where work happened.
That is the SaaS shift now. The old model makes workers travel into the app, learn the menu tree, click through fields, and pay the seat toll. The new model pushes capability into the flow of work through agents, APIs, browsers, and embedded automation. Software is still here, it’s still cloud-based, and it’s still something you pay to access rather than own. What’s changed is what you pay for. No longer will companies pay for seats; now they pay for outcomes, whether through consumption-based pricing or another model.
The age of shelfware is coming to a close, but software-as-a-service is more critical than ever.
What Survives this Revolution?
Every major shift carries winners and losers. Gartner says that software spending is still expected to grow by 14.7% in 2026. So, this requires us to consider what characteristics make software valuable and worthy of its price tag.
First, mission-critical tools that are so deeply embedded in core operations that removing them would break the business. Think payroll software, ERP systems, or security tools. The switching cost isn’t just technical; it’s organizational. Jobs, audits, and legal obligations are so interwoven with these tools that they will survive the disruption.
Second, systems of record are arguably more important than ever. In a world where chat interfaces and AI wrappers can be copied in a weekend, the durable moat is the proprietary data, permissions, history, and workflow memory that an agent must query before it acts. The interface layer is getting cheap; the source of truth is not.
Third, specialization survives. Torii found that productivity, developer tools, design software, and sales and marketing combined for about 70% of shadow apps. That is what a market looks like when teams are shopping for fit, speed, and context. Not a single giant suite. As development costs drop, niche-industry and use-case apps will become more economically viable. A tool built for boutique law firms, dental practices, or vertical manufacturing workflows can beat a horizontal giant because it removes coordination cost, not just button clicks.
The Verdict
SaaS isn’t dead, but the lazy version is. The bloated, seat-metered, UI-obsessed version that relied on switching costs and user tolerance is in trouble. But software itself is more essential than ever.
The steam engine did not disappear because factories stopped needing power; it disappeared because power found a better delivery model. SaaS is at that moment now. The winners won’t brag about time in the tool; they will own the data, govern the workflow, and deliver the results.
Also Read: The Infrastructure War Behind the AI Boom
[To share your insights with us, please write to psen@itechseries.com]
Comments are closed.