Will SaaS Survive the AI Shockwave?
Animation by Grok
Will AI crush SaaS? This month Gerry Gimenez, our recently joined Director of Software & AI coverage, and Andrew Birmingham of Mi3, duke it out over the future of SaaS in an AI world. In this article, AB makes the case that SaaS is in deep peril. You can find Gerry’s bull case ‘SaaS Isn't Dead, It's Getting a Brain’ here.
When George Colony founded Forrester Research in 1983, software meant shrink-wrapped disks and boxed manuals. Forty years later, the industry that replaced them, Software-as-a-Service, now faces its own mortality, says the storied firm’s CEO and chairman.
The cause is not another delivery model but an entirely new logic of computation: systems that think, act, and transact on behalf of users who speak to software via prompts, using natural language.
The most famous proponent of the move off SaaS and into AI is Swedish Buy Now Pay Later firm Klarna, whose CEO Sebastian Siemiatkowski, has been vocal on the scale of the opportunity, while annoying firms like Salesforce and Workday along the way. According to Business Insider, Klarna claims to have eliminated over 1,200 external SaaS tools, and used AI to cull headcount by 40 per cent. (although Siemiatkowski recently told Reuters he may have focused a bit too much on cost-cutting.)
The shock caused by the disruption of generative and now agentic AI has also reached the capital markets. Colony says those markets face a period of great disruption and chaos, because many venture capitalists are holding companies that are not AI companies. “They're going to have to shed them or sell them before they can begin to renew their portfolios.”
Article continues below.
Like this article? Join the thousands of tech founders, board members and investors who subscribe to our free monthly newsletter, Tech Round-Up. Sign up below!
Spaghettification
There is also the issue that many contemporary SaaS stacks are not the gracefully architected cathedrals the glossy sales brochures make them out to be. Instead, most are messy amalgamations of acquisitions bolted together on the fly. Year after year, the integration challenge gets harder, and downstream threats like cybersecurity get worse. Nobody wants to get caught holding onto the next generation of technology debt.
SaaS consolidation can also be found in the numbers, if you know where to look. Scott Brinker, the editor-in-chief of Chiefmartec, has been plotting the rise of the marketing and customer tech category within SaaS for over a decade. The number of solutions has increased from barely a hundred at the start of the last decade to over 15,000 now.
But in this year’s report, Brinker noted something new – while generative AI has rapidly fuelled the emergence of an emerging class of solution providers, it is traditional SaaS vendors from the pre-ChatGPT era who are churning out of Brinker’s list the fastest.
According to the State of Martech 2025 Report, which he and Frans Riemersma produce each year, “The majority of the 1211 products removed – 774 – were born before the ChatGPT inflection point of November 2022.” That number is expected to grow.
Brinker, whose day job for the last eight years was running HubSpot’s partner ecosystem, envisages a very different future for business software.
“Popular AI assistants such as ChatGPT, Claude, and Gemini – that have millions of non-technical subscribers – are often creating software programs behind the scenes to do the bidding of their users without those users even knowing that software was built and run on their behalf.” Per the report, it’s a future that promises “Not millions but billions of custom software programs proliferating companies’ digital operations. Probably trillions, many blinking in and out of existence on demand.”
That’s a very different world from the one that heralded the rise of SaaS. Shrink wrap may have moved to the cloud, and capex may have shifted to opex, but the predominant pricing remained anchored around seat-based models.
That is now under threat because with hyperscaled automations, there will be far fewer seats required. In some parts of the market – such as for data analytics platforms – transaction pricing is more common. But there’s a belief that with the rise of agentic AI, pricing could become more outcome-focused – where customers pay for things like call centre resolutions. In financial systems, the measure could be reconciliations completed.
It’s also worth noting that where Colony is a revolutionist when it comes to the impact of agentic AI on SaaS, Brinker is more of an evolutionist. But both see huge disruption. They just disagree on the timing.
Pricing breaks down
The maths that underpinned SaaS margins (some of which run as high as 80 per cent, according to Brinker) no longer works. SaaS models built on per-seat pricing are heading for collapse as autonomous agents take over tasks at a fraction of the cost – and promise near infinite scalability. The more effective AI becomes, the faster seat-based pricing erodes.
Colony warned that the next wave of AI will do two things that spell trouble for incumbents: it will solve business problems at a fraction of today’s cost, and it will do so with far simpler, more intuitive tools. That combination of cheaper and easier poses an existential threat to software giants whose models depend on hefty enterprise subscriptions. A system that once cost a bank millions a year to license could, under agentic and generative AI, be delivered for a tenth of the price, he believes. The technology, in other words, won’t just disrupt products, it will upend the very economics that have sustained the SaaS era.
Perhaps it should. Studies by Zylo, a platform that helps companies manage the licences for their application portfolio, reveal that software utilisation rates for some of the most popular enterprise software solutions like ServiceNow and Atlassian, are running at only 25 per cent or less. Such profligacy will not survive the agentic era.
Meanwhile, the Forrester chief contends that today’s SaaS giants who are all well-versed in the lessons of The Innovator’s Dilemma, are already deploying a familiar defensive playbook: acquire the upstarts, obstruct the rivals, imitate the revolution, and fortify their dominance by tethering new technologies to old empires.
He just doesn’t think it will work.
And indeed that’s informing his firm’s advice to its clients – all those giant global enterprises that fund the global SaaS industry: slow the burn. On a podcast in August, Colony said the advice now is: “If you are looking to buy a traditional SaaS platform in the next two years, wait. Or sign very short-term deals. Do not make big commitments.”
SaaS vendors, of course, are all frantically adding agentic capabilities to their existing stacks. And it's not like SaaS will evaporate overnight. Bain and Company, for instance, argues that because AI automates tasks and replicates workflows, disruption is mandatory but obsolescence is optional. Goldman Sachs, meanwhile, says AI agents will expand software markets by unlocking productivity, but that any expansion will favour agent-centric designs over traditional UI/seat paradigm.
Each imagines a world where tech budgets are reallocated from SaaS modules and agencies toward AI inference and orchestration.
It’s those kinds of assessments that lead many to believe that SaaS’s milk and honey moment may have already passed.
Median SaaS multiples have fallen by about 60 per cent from their 2021 peaks, having now returned to pre-COVID levels according to analysis by firms such as SaaS Capital and Aventis Advisors. The agentic transformation is only just starting, so the outlook is unlikely to get any easier.
Where cloud computing redistributed cost structures, generative and agentic AI collapse them. Three forces are converging to make that collapse inevitable: capital reallocation, architectural obsolescence, and a pricing model that no longer fits the way digital work gets done.
________________
Andrew Birmingham is the technology editor of Mi3 Australia. He is the current recipient of the Kester Lifetime Achievement Award for services to IT Journalism in Australia, and was awarded the Best Business Technology Journalist prize at the 2025 IT Journalism Awards.