Software loses seat-count revenue when AI reduces the humans who use it. Software loses point-solution revenue when AI agents replicate single-workflow tools. Software retains — and may gain — value when it owns the system of record, the compliance layer, or the proprietary data context that makes AI work better.
Hypothesis Log
2026-03-31
v3
Software loses seat-count revenue when AI reduces the humans who use it. Software loses point-solution revenue when AI agents replicate single-workflow tools. Software retains — and may gain — value when it owns the system of record, the compliance layer, or the proprietary data context that makes AI work better.
Split into three sub-claims after seat compression data and PwC M&A analysis — read post
2026-03-30
v2
Software isn't losing value because it's easier to build. It's losing budget share because enterprises are prioritizing AI spend over software expansion spend.
Refined after Q1 earnings data — mechanism is budget cannibalization, not replacement — read post
2026-03-23
v1
Software loses its value because AI can create it fast and efficiently, commoditizing the product and collapsing the SaaS business model.
Opening hypothesis — intentionally broad — read post
Evidence
Microsoft open-sources Agent Governance Toolkit addressing all OWASP Top 10 agentic risks. Governance infrastructure arriving, which could slow agent deployment velocity — partially counteracting disruption timeline.
Challenger: 25% of March 2026 layoffs attributed to AI by employers (15,341 of 60,620). Labor market impact now measurable.
Salesforce Agentforce hits $800M ARR (Q4 FY26), up 48% QoQ. Agent-revenue-model company growing while seat-model companies compressed. Bifurcation pattern holds.
IGV forward EV/Sales drops to 3.6x — lowest since 2011. Trailing P/E from 45x to 32x in one quarter. Software category-wide repricing, not individual company issue.
ServiceNow drops 10.4% in one day; stock down 38% over 12 months. Stifel slashes target citing federal spending cuts and agentic AI threat to per-seat model. NOW AI product growing while stock falls — market pricing in seat cannibalization.
a16z: real moats (network effects, cornered resources, process power) aren't threatened by AI; code was never where value lived
Sequoia: copilot tools race against the model; autopilots capture work budgets
AI-native companies commanding 83% higher M&A premiums; domain depth and workflow gravity are durable moats
Snowflake revenue grew 30% YoY; Palantir 70%; ServiceNow 21% — AI-driven platforms growing despite compression
Atlassian first-ever enterprise seat count decline; Workday down 40%; Salesforce down 37%
AI budgets up 100%+ YoY while total IT budgets up only 8%
SaaS valuation multiples compressed 34% (5.6x to 3.7x) while fundamentals held