The obituary for enterprise software was written too soon.
Over the past year, roughly $2 trillion in software market value has been erased on fears that AI would render the software-as-a-service sector obsolete — a panic that analysts and commentators branded the 'SaaSpocalypse.' According to a commentary by Sonnenfeld, Tian and Henriques published in Fortune, that thesis was wrong from the start, and three companies are proving it in real time: Salesforce, Booking Holdings and IBM.
The bear case had a seductive logic. Large language model companies like OpenAI and Anthropic, the argument went, would deploy autonomous AI agents capable of managing entire customer relationships end to end. Salesforce — the dominant customer relationship management platform — would be demoted from central command center to a passive database that agents occasionally query, an interchangeable commodity at best, redundant at worst. The stock was priced accordingly: down roughly 20% this year and 40% from its high.
The market, according to the authors, read the dynamic backwards.
'Lower cost of intelligence increases value of incumbent data,' analysts at Wells Fargo declared. The insight is straightforward: AI agents are only as good as the data they operate on. An agent working to close a sale still needs somewhere to research the customer, log interactions, store contracts and draw on decades of relationship history. That repository is Salesforce — and there is no practical way to migrate that institutional memory into an LLM, nor any reason a corporation would trust a language model as the vault for its proprietary customer data.
The numbers support the thesis. Salesforce has processed over 216 trillion customer records this year alone. Its AI agent platform, Agentforce, has grown from $100 million to $1.5 billion in annual recurring revenue within 18 months of launch, with more than 30,000 Agentforce deals already closed. The platform now features a partnership with Anthropic's Claude as its premier agent. Every new AI agent deployed generates more data — data that, by necessity, flows back into Salesforce's infrastructure.
The authors argue that the original 'death' narrative has since softened into a secondary bearish claim: that even if SaaS companies survive, they will face higher customer acquisition costs and compressed margins. That revision, they contend, is equally mistaken for companies with entrenched data moats and proven AI monetization.
Booking Holdings and IBM are cited alongside Salesforce as additional examples of incumbent software firms positioned as AI beneficiaries rather than casualties — companies whose existing data assets and enterprise relationships give them structural advantages that new AI entrants cannot easily replicate.
---
CEO Times reads this as a textbook case of market panic overriding fundamentals. When capital stampedes toward a narrative rather than toward earnings, it creates opportunity — and it punishes companies that deserve a premium. The SaaSpocalypse thesis assumed that the value in AI would accrue entirely to the model layer, leaving data custodians with nothing. The opposite is proving true: clean, structured, proprietary enterprise data is the scarce input, and the firms that hold it are extracting pricing power, not surrendering it. Free enterprise rewards the companies that built real infrastructure over decades. The market is beginning to remember that.
