Predictium · Bubble Watch
SaaS module · weekly
As of 17 Aug 2026
Drill-down · the AI disruption front

The SaaSpocalypse

Did AI break the model — or is this 2016 again?
55%
45%
◄ Cyclical panic — re-rates & recovers
Structural decline ► — the “newspapers” path
Our read · editorial estimateSentiment: max pessimismSoftware crowding 22.8% vs semis 99.3%

The whole sector already de-rated below the market — the cheapness isn't the question anymore. The only thing that matters now is whether AI agents permanently broke per-seat software, or whether this is another panic that recovers in months.

Market cap erased
~$2T
Jan–Feb 2026
IGV software ETF
-22%
YTD trough · worst vs S&P since 2000
Software fwd P/E
22.7×
below the S&P — first in the cloud era
EV / revenue
7×→3.4×
trough multiple reset
Two layers of panic

1 · Seat compression

Per-seat pricing — the bedrock of SaaS — breaks if one AI agent does the work of five licensed users. Fewer seats, less revenue per customer.

2 · Interface bypass

Agents execute workflows directly, demoting the app to “rails” behind a chatbot. The UI that justified the subscription disappears.

Who AI eats — sub-sector disruption risk
Sales automation / CRMSalesforce
85
Project mgmt / collaborationAtlassian
80
HR / back-officeWorkday
66
Securitymixed signals
52
Design / creativeAdobe, Autodesk
40
DevOpsstickiest contracts
30
Data infrastructureSnowflake, Databricks
25
Name by name — the carnage & the survivors
CompanyFrom peak*Fwd P/ERead
Salesforce CRM~-45%~20×"Seat economy → task economy"; 20-yr data moat is the floor
ServiceNow NOW~-50%28× (67× avg)Beat earnings, fell 11% — multiples compress on a clean quarter
Snowflake SNOW~-50%Consumption model is both the threat and the shield
Workday WDAY~-40%Back-office seats squarely in the agent crosshairs
Atlassian TEAM~-70%Task tracking is exactly what agents automate first
Adobe ADBEde-rated12× (30× avg)AI augments creative — a relative safe haven, cheap
Microsoft MSFT~-25%~27×Even the blue-chip got hit — shed $357B on Jan 29
Palantir PLTRresilientrichFramed as the "OS for AI agents" — the rare winner
The bull / bear fork

Cyclical · 55%

  • JPMorgan & Goldman: selloff has gone too far; fundamentals intact
  • Switching costs on systems of record (CRM, ERP) are brutal to rip out
  • PE floor: Thoma Bravo / Vista circling take-privates at these prices
  • 2016 rhyme: the last SaaS panic recovered within months

Structural · 45%

  • Goldman's Snider: could be a permanent decline, like newspapers
  • Growth was already dying: SaaS growth fell every quarter since 2021
  • Per-seat pricing genuinely breaks if agents replace human users
  • AI-native upstarts undercut incumbents on cost and automation

Get the weekly read

Monday mornings — the number + what moved. Free, never gated.
*Drawdowns & multiples are approximate, mid-2026, blended across sources and moving daily; the sector has clawed back somewhat since the March trough.
Sources: SaaStr (fwd P/E 22.7×), iShares IGV, multiples.vc (EV/Rev), Bulloak Capital ($2T), Oliver Wyman (disruption framework), JPMorgan crowding model, Goldman Sachs. Widely attributed to agentic-AI product launches (notably Anthropic's Claude Cowork). Kept factual and sourced.
Methodology: the 55/45 fork is our editorial estimate, not a model output — whether AI permanently impairs per-seat software has no historical ground truth yet, so we publish a transparent judgment rather than imply false precision. Sub-sector scores rank AI-substitution risk 0–100.
Methodology & sources: how every score is built →
Not investment advice. A high reading signals fragility and weak long-run expected returns — not a near-term crash call.