In February 2026, in a single 48-hour window, $285 billion in market value vanished from corporate software stocks. Since the October 2025 peak, the S&P 500 Software & Services index has lost roughly $2 trillion, according to Fortune. Wall Street has named the event the SaaSpocalypse — and AI agents sit at the center of the diagnosis.
The question is no longer academic: if an agent does the work of five analysts, no one needs five software licenses. The per-seat pricing model that powered two decades of SaaS growth is under direct attack. But it is important to separate consolidated thesis from market panic — especially for Brazilian companies, who live a reality quite different from the American one.
Where the panic comes from
The trigger, according to Fortune, was Anthropic's launch of Claude Cowork in February. For the first time, investors watched agents execute complex knowledge-work flows without continuous supervision. The reaction was immediate: Thomson Reuters fell 15.83% in a single trading day, LegalZoom dropped 19.68%, and banks like Jefferies downgraded Workday and DocuSign citing AI disruption risk explicitly.
This is not just a short-term reaction. Deloitte projects 75% of global companies will invest in agentic AI by year-end 2026, with the market jumping from $8.5 billion to $45 billion by 2030 — a 53% compound annual growth rate. IDC goes further: estimating 1 billion AI agents in operation by 2029, 40 times the 2025 volume.
The strong thesis: software-as-tool becomes utility
The logic behind the SaaSpocalypse is coherent. Traditional SaaS systems exist so humans can execute tasks — fill in fields, query dashboards, move data between spreadsheets. If an agent does this alone, the visual interface loses value. What matters is the outcome.
According to Gartner, 40% of corporate SaaS spend is expected to shift by 2030 to usage-, agent-, or outcome-based pricing, leaving fixed per-user billing behind. The industry has already started reorganizing: SaaS companies are experimenting with charging a percentage of the savings generated or additional revenue captured by agents running over their systems.
The counterpoint: 65% of SaaS will survive
Here lives the part headlines ignore. The same Gartner forecasting disruption is explicit: only 35% of so-called point tools — single-purpose, narrow-function software — are expected to be replaced by agents through 2030. The other 65% don't disappear. They evolve.
The segmentation matters:
- Systems of record (ERPs, CRMs, data platforms): remain the source of truth on which agents operate, and tend to strengthen their moats
- Network-effect platforms: Salesforce, ServiceNow, Slack and the like benefit when more agents need them as interface
- Complex, regulated software: sectors like healthcare and finance carry audit requirements no autonomous agent fulfills alone
- Applications with proprietary data: the deeper the data moat, the lower the risk of agent commoditization
The soberest reading is that the SaaSpocalypse is, in fact, a process of selective unbundling. Point tools commoditize; platforms with data moats come out stronger.
The Brazilian angle: disproportionate opportunity
In Brazil, the equation shifts on at least three dimensions. First, the national SaaS market reached $7.9 billion in 2025 and is projected to hit $25.5 billion by 2034, according to IMARC Group — still growing on a curve below international peers, which dampens the impact of a value compression.
Second, IDC estimates that AI agents will attract R$3.4 billion in investment in Brazil in 2026 alone, entering the top three most-demanded IT categories among local companies, with growth above 30% over 2025. Third, only about 5% of Brazilian small and medium businesses use any SaaS today — meaning that, for this segment, agents aren't replacing software; they're arriving before traditional software was even adopted.
The consequence is strategic: Brazil can leapfrog. Companies that haven't yet migrated to SaaS can go straight to agent architectures, avoiding two decades of legacy enterprise software technical debt.
What Brazilian leaders should do now
For companies that buy software, the path begins with renegotiating based on value delivered, not seats occupied. Audit how many licenses your organization actually uses, identify repetitive flows that could be executed by agents, and open outcome-based conversations with vendors.
For companies that sell software, three moves protect the portfolio:
- Build agent layers on top of the existing product, turning SaaS into infrastructure for automation rather than a human-use interface
- Migrate part of revenue to usage- or outcome-based pricing before the customer demands it
- Invest in differentiators agents don't replicate: proprietary data, certified regulatory integrations, user communities
For professionals and managers, the recommendation is to study agentic architectures now — MCP, tools, RAG, orchestration — before that fluency becomes a basic market prerequisite.
Conclusion
The SaaSpocalypse is real as a market phenomenon, but the headline overstates it as a prognosis of annihilation. The $2 trillion that vanished reflects a legitimate repricing of business models that depended on human friction to generate revenue. What comes next isn't the end of corporate software, but its reorganization around agents.
Brazilian companies are in a privileged position to choose where they want to stand in this transition — as long as they start deciding now.
This article was published on May 14, 2026. Follow Entercast to stay updated on the next developments.