The central question is no longer “Which SaaS do we buy?” but “where do we build, where do we buy, and how do we architect for an AI‑first world in which agents become the primary users of software?”¹
Median EV‑to‑revenue multiples for listed SaaS companies have fallen to roughly the mid‑single‑digit range by early 2026, down from peaks above 30x during the 2021 era of zero‑rate exuberance.⁴ In early 2026, around 300 billion dollars in software market cap was erased in a matter of days as markets reacted to AI agents, slowing growth, and the realisation that prior expectations had been mispriced.⁷
For years, SaaS investors behaved like farmers who kept planting more fields and assuming every harvest would be bigger than the last. Cheap capital and generous valuations made it seem that the land would never tire. Now the soil is depleted, yields are lower, and the market is repricing the land itself.⁶ In that environment, many large SaaS platforms begin to look less like growth crops and more like mature railroads: assets that still carry traffic and throw off cash, but whose business model is slowly dying even as investors focus on harvesting the remaining cash flows.⁶ You cannot keep paying premium prices for fields that no longer produce exponential harvests, even if they continue to deliver respectable, but finite, income.
Declining multiples are not just a stock‑market story; they are a signal that the assumptions underpinning the traditional SaaS model, including endless growth, seat‑based pricing, and perpetual expansion of software budgets, are breaking down.⁵ Median public SaaS EV/revenue multiples have fallen to around 3.4x, down from >30x at the peak.⁴
As a client preparing to make large SaaS investments or contemplating replacement of core systems, you are entering a market in which prices no longer fully reflect the new economics of AI‑assisted development.⁸ Vendors are under pressure to defend revenue models built for a different era.⁶ The option to build is now becoming structurally more attractive for firms.³
Build vs. Buy
The build‑versus‑buy equation for firms is adjusting to a more AI centric approach. Survey data and case studies show that enterprises are no longer defaulting to SaaS for every use case. A Futurum study of hundreds of IT leaders found a persistent preference for building most applications in‑house, supplementing with purchased solutions only where necessary.¹ For example, Retool’s build‑versus‑buy research and independent analyses report that roughly a third of enterprises have already replaced at least one SaaS tool with custom software, and a clear majority plan to build more internal tools in 2026.²
AI‑assisted development and app‑generation platforms have changed the economics of home-grown solutions. Modern tools allow teams to generate internal applications in weeks rather than months and at a fraction of historic cost, shifting the balance of power towards internal product and internal engineering teams.⁸
The Pattern Emerges
SaaS sprawl has become a material cost issue. One analysis estimated average SaaS spend at nearly $4,830 USD per employee per year, with tens of millions wasted annually on unused licences in large enterprises.¹² When AI makes it cheaper and faster to build tailored internal tools, those numbers no longer look sustainable. Meanwhile, 35% of enterprises have already replaced at least one SaaS tool with custom software and 78% plan to build more in 2026.²
Modern app‑generation platforms have shortened build times and lowered capital requirements. Research comparing build and buy scenarios suggests that building a moderately complex internal operations tool using AI‑assisted methods in 2026 can cost tens of thousands of dollars over a three‑year horizon, compared with substantially higher all‑in costs for a comparable SaaS product once licensing, integration, implementation, and annual increases are included.³
The quality of internal solutions can be higher because they are designed around the company’s actual workflows, data, and risk profile.³ Buyers are discovering that the real ceiling on SaaS is not just price; it is fit.
Case Studies: When Insourcing Works
Emerging cases illustrate how enterprises are replacing SaaS with internal, AI‑powered systems without sacrificing reliability or speed, though safety and security is still an issue to most firms.
- A logistics company facing high licence fees and rigid workflows on a transportation management SaaS platform rebuilt its orchestration layer using internal tools and AI‑driven routing logic. The new system integrated directly with existing data sources and delivered a meaningful reduction in fuel and operating costs, while eliminating a large recurring subscription.¹³
- In financial services, a digital bank that needed a new lending engine adopted a hybrid strategy. It used SaaS for the core ledger while building a custom AI credit‑scoring module on top. The result was a faster launch and a differentiated decision engine tightly aligned to its own risk models.¹³
- Palantir replaced an expensive third‑party CRM with an AI‑first solution built on its internal platform, explicitly framing this as a build‑over‑buy decision.¹⁴
- GoDaddy is testing the replacement of smaller third‑party SaaS tools in corporate functions with internally built applications to reduce cost and complexity.¹⁴
- SoundHound and others report insourcing software capabilities and using agents and internal platforms to consolidate legacy systems.¹⁴
Mid‑market firms are also shifting. One case highlights a team paying around $20,000 per year for a third‑party SaaS product whose support had become so slow that it was faster to rebuild the solution in‑house using an internal app‑building platform. The team rebuilt the tool and now runs dozens of internal applications off the same foundation, with greater control and faster product releases.¹⁵
Taken together, these cases show that the “SaaSpocalypse” is plausible. Enterprises are discovering that AI‑assisted development allows them to reclaim control over workflows that were previously outsourced to SaaS tools.²
Caution is Recommended
Not every build story is a success, and not every SaaS replacement is prudent.
Internal software comes with governance, talent, and maintenance burdens. Surveys of enterprise builders note that shadow IT is proliferating; many teams are creating internal tools outside traditional guardrails as app‑generation platforms spread.²⁰ Without disciplined architecture, enterprises risk recreating SaaS sprawl internally, with less visibility and more security exposure.²⁵
Case studies also show many failed builds where organisations underestimated the complexity of these builds, the ongoing effort required for security, compliance, and performance, and the need for robust product management and user experience for internal tools.²⁵
Enterprise research frames the emerging best practice as “custom core, SaaS periphery”. Build where the use-case is differentiating, such as primary workflow tools, customer‑facing portals, and domain‑specific intelligence. Keep SaaS for commoditised functions like communications, payments, email, monitoring, and generic analytics.²⁸ This is not a retreat from software as a service; it is a more exacting and nuanced standard for where SaaS belongs in the stack.
SaaS Disruption is Inevitable
Valuation compression signals that investors expect slower growth, greater competitive pressure from AI, and shifts in customer behaviour. Multiples have come down sharply because markets are aggressively discounting the future cash flows of SaaS businesses whose pricing and product strategies are misaligned with an AI‑first world.⁴ As a buyer, you should assume that some vendors in your shortlist will not look the same in five years. Some will be acquired, some will pivot, and some will fail.⁵
Every major SaaS category is now under replacement pressure. Build‑versus‑buy research finds that workflow automation tools, internal administration applications, CRMs, BI platforms, project management, and customer support systems all face growing risk of being replaced by custom builds.²⁴ If you are contemplating a major investment in one of these staples, you should assume that AI‑powered alternatives, internal or external, will mature rapidly during the life of your contract.²⁴
The economics of AI‑assisted building have shifted faster than SaaS pricing. Analysts highlight a mismatch in which SaaS pricing has not fallen in proportion to the reduced build cost, making custom builds increasingly attractive over a three‑to-five‑year horizon.⁸ When you lock into large multi‑year SaaS commitments at current price points, you may be overpaying relative to what AI‑enabled alternatives will cost even two years from now.⁸
Large enterprises themselves are signalling the direction of travel. When companies such as Palantir and GoDaddy publicly report replacing external SaaS with internal AI platforms, they are not just cutting costs. They are repositioning software as a strategic capability, not a collection of external subscriptions.¹⁴ Boards and executives should treat that signal as a prompt to rethink their capabilities and technology stacks¹⁴
How Buyers Should Respond
For clients considering large SaaS investments or significant system replacements, a disciplined response to this landscape requires a playbook.
- Treat every major SaaS decision as an architectural choice. Start from a map of core workflows, data domains, and risk boundaries. Determine where you must own, where you can rely on external platforms, and where AI agents are likely to operate. Let this architecture, not vendor marketing, drive your decisions.¹⁶
- Run a serious build‑versus‑buy analysis with AI‑adjusted numbers. Use contemporary cost data to compare what it would take, given modern app‑generation tools, to build and maintain internal solutions over three to five years against the full economic cost of SaaS, including licence costs, integration, support, and annual uplifts.³ For strategic workflows, that calculation increasingly favours building or hybrid strategies.³
- Design for custom core and SaaS periphery. Reserve internal builds for capabilities that truly differentiate your business, such as orchestration of proprietary processes, domain‑specific decision engines, and cross‑system intelligence. Use SaaS where the function is a utility and where vendors can credibly claim long‑term resilience, for example in payments, communications, observability, and commodity analytics.²⁸
- Require an AI‑ready integration posture from any SaaS you do buy. Tools that do not offer clean APIs, robust event models, and clear policies on agent access will feel like legacy systems long before their contracts expire. Your agents will need to act across systems; vendors that cannot accommodate that will slow your transformation.²⁵
- Manage internal builds with the same rigour you apply to external spend. Establish standards for security, architecture, and lifecycle management. Treat internal platforms as products with roadmaps, service levels, and clear ownership, not as one‑off projects.²⁵
- Maintain a sceptical stance towards size as a proxy for safety. In a market where multiples have compressed and AI is eroding traditional moats, large incumbents are not automatically safer than smaller, focused platforms or than your own internal solutions. The farming analogy applies here. The largest fields are the most exposed when the soil is exhausted.⁷ The crops that survive are those planted in the right terrain, with realistic expectations about yield.
In Summary
The SaaS model is undergoing a structural test. AI agents, compressed multiples, and insourcing trends are shifting advantage towards organisations that treat software as a strategic capability rather than a catalogue of subscriptions. For buyers, the real risk is not that SaaS disappears, but that procurement and architecture decisions remain anchored in assumptions from the last decade. Executives who rebuild their software stack around clear workflows, disciplined build‑versus‑buy choices, and AI‑ready integration will be positioned to use this disruption to deepen capability and reduce structural waste.
For SaaS vendors, survival now depends on becoming outcome‑platforms rather than “just” seat‑priced tools, and on restructuring both product and pricing around AI agents and proprietary value rather than generic workflows.
- You must assume seat‑based pricing will compress. As AI agents do more of the work, customers will need fewer human licences, and CIOs are already pushing vendors toward consumption‑ and outcome‑based models.
- Thin, horizontal point solutions are most exposed. Vendors that offer narrow, easily replicable workflows with low switching costs are at real risk of being displaced by AI‑assisted development and internal tools.
- Depth, data, and domain focus are the new moats. Vendors that anchor themselves in complex, regulated, or industry‑specific processes, own proprietary data, and integrate cleanly into agentic architectures have a far higher chance of enduring.
- You need an explicit AI agent strategy. It is not enough to bolt on a chatbot. The product roadmap has to assume agents as primary users, with robust APIs, telemetry, and controls that let customers shape agent behaviour and outcomes.
- Expect consolidation and harder customers. Buyers are shortening contract terms, cutting overlapping tools, and renegotiating aggressively; mid‑market vendors without a clear AI story or defensible economics are likely acquisition targets or casualties.
In short, the SaaSpocalypse is closing the window on generic, seat‑priced SaaS. Vendors that re‑architect around agentic workflows, differentiated data, and outcome‑linked pricing can still thrive; vendors that cling to the old model should plan for consolidation, not independence.
To hear more about how the “SaaSpocalypse” might disrupt your business, listen to our podcast. Listen to “The Insurgent Mindset” on Apple, Spotify and Youtube.
Or contact us at mtice@insurgencegroup.com.
References and Links
- Futurum Group, “Enterprises Maintain 56% Build‑In‑House Preference Despite SaaS Strengths.”
Enterprises Maintain 56% Build-In-House Preference Despite SaaS Strengths
- Retool / BusinessWire, “Retool’s 2026 Build vs. Buy Report Reveals 35% of Enterprises Have Already Replaced SaaS With Custom Software.”
https://www.businesswire.com/news/home/20260217548274/en/Retools-2026-Build-vs.-Buy-Report-Reveals-35-of-Enterprises-Have-Already-Replaced-SaaS-With-Custom-Software
- EmphasisTech, “Why 35% of Enterprises Already Replaced Their SaaS.”
https://www.emphasistech.com/blog/post/build-vs-buy-2026
- L40, “SaaS Multiples 2026: The Real Private Range (4x to 9x ARR).”
https://www.l40.com/insights/saas-multiples
- BVResources, “SaaS multiples slide as growth slows and AI concerns mount.”
https://www.bvresources.com/articles/bvwire/saas-multiples-slide-as-growth-slows-and-ai-concerns-mount
- Aventis Advisors, “SaaS Valuation Multiples: 2015–2026.”
- Forbes, “$300 Billion Evaporated. The SaaS‑Pocalypse Has Begun.”
https://www.forbes.com/sites/donmuir/2026/02/04/300-billion-evaporated-the-saaspocalypse-has-begun/
- AI Navigate News, “AI lowered the cost of building software. Enterprise governance has not caught up.”
https://ai-navigate-news.com/en/articles/23c51b5c-848a-4712-8cd0-831f1ddc0de5
- Viviscape, “Build vs. Buy in 2026: Why 35% of Enterprises Are Replacing SaaS.”
https://viviscape.com/news/build-vs-buy-2026-enterprises-replacing-saas
- Tectome, “When to Build Custom Internal Software Instead of Buying SaaS.”
https://www.tectome.com/blogs/build-custom-software-vs-saas
- LastRev, “How do companies build AI‑powered internal tools instead of buying more SaaS.”
https://lastrev.com/blog/how-do-companies-build-ai-powered-internal-tools-instead-of-buying-more-saas
- Lynton, “What $2 Trillion in SaaS Losses Means for Your Business.”
https://www.lyntonweb.com/library/saaspocalypse
- Velsof, “When SaaS Stops Scaling: Why 35% of Companies Are Building Custom Software in 2026.”
When SaaS Stops Scaling: Why 35% of Companies Are Building Custom Software in 2026
- Hudson Labs, “AI Replacing SaaS: 2026 Software Insourcing Trends.”
https://www.hudson-labs.com/research/ai-saas-replacement-2026
- LI Solutions, “Should you replace SaaS with custom software?”
https://li.solutions/blog/replace-saas-with-custom-software/
- WildnetEdge, “Enterprise Software Development vs SaaS: Key Differences.”
Enterprise Software Development vs SaaS Platforms: Choosing the Right Enterprise Solution