In the first week of February, "over $1 trillion in market capitalization was erased from software stocks" in seven days, according to Forrester. Forrester ties the sell-off to AI agent innovation and investor fears that per-seat pricing would become obsolete and that "vibe coding" would let startups copy SaaS features. The headlines called it the SaaSpocalypse.
For the people who pay for software, nothing of the sort happened. The same Forrester post projects global SaaS spending rising from $318 billion in 2025 to $512 billion in 2028. Gartner forecasts total software spending of about $1.44 trillion this year, up 15.1% (Gartner). Investors repriced the vendors. Buyers got price increases.
So the useful question for a buyer is not whether SaaS survives. It is which of your subscriptions still earn their renewal, and who is deciding that.
Your bill went up while the stocks went down
On July 1, Microsoft raised Microsoft 365 E3 from $36 to $39 and Office 365 E3 from $23 to $26, with Business Basic up 16% (Microsoft). The new packages add Copilot Chat enhancements and security features. Whether you wanted those or not, they are now in the price.
That is one vendor. Across the spend it processes, Vertice, a SaaS purchasing platform, puts SaaS price inflation at 16.4% in June, a record high, and 14.3% in early September on preliminary data (Vertice). It also warns that shrinkflation, list-price increases paired with less feature access at the same tier, may be adding to the real cost. That is a vendor's own data, not an independent index, but it points the same way as Microsoft's price list.
Meanwhile, much of what you already pay for goes unused. Zylo, a SaaS management vendor, analyzed more than 40 million licenses for its 2026 index and found organizations leave an average of 36% of licenses unused against recommended utilization (Zylo). In the same report, 78% of the 218 IT leaders it surveyed had been hit by unexpected charges tied to consumption-based or AI pricing, and 61% had cut projects because of unplanned SaaS cost increases. Business units, not IT, control 81% of SaaS spend.
Consolidation is already happening, by default
Forrester's advice to buyers is blunt: "The reaction shouldn't be 'buy more AI' — at least not right away." Rearchitect what you already have, cut the vendor count down to strategic partners, deal with redundancy, and don't buy new SaaS without knowing how it fits your AI strategy. It expects horizontal point-solution vendors to be challenged, and says those "with low switching costs and weak, non-embedded enterprise workflows won't scale beyond their current customer base."
Many organizations are consolidating, but not like that. In its September AI spending forecast, as reported by Digital Terminal, Gartner said that with GenAI "firmly in the Trough of Disillusionment in 2026," enterprises are using "simpler embedded AI features from their incumbent software providers." Gartner's John-David Lovelock added that "the risks associated with vendor lock-in, data sovereignty, and run-away costs are not deterring buyers" (Digital Terminal, reporting Gartner).
Put those together and you get consolidation by drift. The big platforms grow at each renewal because adding their AI tier is the easiest thing to approve. The long tail of point tools survives because nobody owns the decision to cancel them. Zylo found large enterprises still add an average of 21 applications a month, while overall app counts stay roughly flat. The portfolio churns, but it doesn't get smaller.
Gartner's own warning applies here: "adding more AI features often creates more cost, not better outcomes" (Gartner).
Building is cheaper. Owning is not free.
The other thing that changed this year is that "build" became a realistic answer for some renewals. In Retool's 2026 survey of 817 of its customers and builders, 35% said they had replaced at least one SaaS tool with a custom build, led by workflow automations and internal admin tools (Retool). That sample leans toward people who build for a living. The same survey found 60% had built software outside IT oversight in the past year, which is its own risk.
A useful worked example comes from Fivetran. It replaced a hosted status page product costing about $65,000 a year, plus at least $10,000 a year keeping its customizations working (Fivetran). Two engineers took about four months, roughly 1,050 hours. At an illustrative $100 an hour, that is around $109,500 to build, with AI tokens a small fraction of it. "Production hardening was 56% of the total effort." Running it costs an estimated $12,400 to $25,000 a year. Payback: "roughly 2.5 years."
That is a good result, and the author is careful about it. The savings are "real but not immediate." A status page is far simpler than a core data platform. And for a company with a much smaller bill, "the answer is probably still 'keep buying it.'"
That is the right shape for the decision. As one contributor to The Next Web put it, AI "has made software cheaper to build. It has not made it cheaper to run" (The Next Web). A build replaces a vendor dependency with an ownership commitment. It is worth taking on when the software is part of how you compete, or when the bill is large and the scope is narrow. It is not worth it to save a few thousand a year on something generic.
Make every renewal a decision
Here is a process a team can start on Monday.
Pull every renewal in the next 12 months into one list. Owner, annual cost, renewal date, notice period, licenses bought versus used. With business units holding most of the spend, finance and IT usually need to build this together.
Put a gate 90 days before each renewal. Before anyone signs, choose one of four outcomes:
- Cut. Low usage, a duplicate of something you already own, or nobody willing to defend it.
- Consolidate. The capability exists in a platform you already pay for. Check that the platform's version is actually good enough, not just present.
- Build. The software encodes how you work differently from competitors, or the bill is large and the scope narrow. Price the hardening and the run cost, not just the first version.
- Renew. It is generic, it works, and switching would cost more than it saves. Renew, but renegotiate on real usage.
Start with the long tail. Point tools with low switching costs are the easiest wins. Low switching cost works in the buyer's favor too.
Rationalize before you add AI tiers. For each AI add-on or bundled upgrade, ask which workflow it changes and how you will measure it. If nobody can answer, it is a price increase.
Check the pricing model, not just the price. Forrester expects contracts to move from seats to consumption and outcome-based pricing. Zylo's unexpected-charges finding shows what happens when nobody models usage first.
Keep exit costs in view. Consolidating onto fewer platforms is sensible. Consolidating without an exit plan is how you end up with the problem we wrote about in our post on vendor lock-in.
SaaS is the right answer for most of what a business runs. Payroll, email and ticketing rarely set you apart. Custom software earns its cost where it encodes how you work differently, and nowhere else. The job is to tell the two apart at every renewal.
We help teams make those build-or-buy calls and build the pieces worth owning. If your renewal calendar is filling up, let's talk.