How LLMs are dismantling the moats that made vertical SaaS defensible, and why the market selloff is structurally justified but temporally exaggerated
In the past few weeks, nearly $1 trillion was wiped from software and services stocks. FactSet dropped from a $20B peak to under $8B.
@nicbstmeNicolas Bustamante
Context & Ripple Effects
The article frames the software rout as a reassessment of whether vertical SaaS advantages remain durable when LLMs can reproduce more product functionality. That view conflicts with the earlier argument that AI-enabled software can move up the product stack rather than disappear into an LLM.
The debate has already spread beyond equities: software debt returns in CLOs had lagged other sectors amid AI concerns, and subsequent coverage showed the software-stock selloff extending across major names. FactSet's drop from roughly $20 billion at its peak to under $8 billion makes the valuation reset concrete.
First-order effects
Public software and services companies face an immediate repricing as investors discount the durability of product-level differentiation; FactSet is among the named examples of that pressure.
Vertical SaaS management teams face sharper demands to show which assets—such as proprietary workflows, data, distribution, or customer integration—remain defensible beyond a general-purpose LLM.
Second-order effects
Competitors are pushed to add AI capabilities faster, making baseline functionality less differentiating and increasing pressure on pricing and product roadmaps.
The equity selloff can feed into financing conditions: the related weakness in software CLO returns indicates that the risk reassessment is relevant to lenders as well as shareholders.
Third-order effects
If LLMs consistently lower the cost of building vertical features, SaaS moats may shift from standalone software functionality toward embedded workflows, proprietary data, and distribution—an instance of moat recomposition rather than automatic industry elimination.
The market may increasingly separate companies whose AI adoption strengthens their product position from those whose revenue depends on features customers can source more cheaply elsewhere; the speed and scale of that separation remain uncertain.
The trend: AI is forcing a re-rating of subscription software around whether its differentiation lies in code alone or in harder-to-replicate customer and data advantages.
Super interesting piece - in part because it uses the financial data industry as an example Over 12 (!) years ago, I wrote a post called “Can the Bloomberg Terminal be toppled?” My point then was that the Terminal's durability was not mainly about feature sprawl or even the
Great rundown of the vertical SaaS reckoning. I'd add that proof-of-control tooling coming from web3 companies/protocols allow for even further erosion when agent activities and identities are verifiable
Required reading for anyone in fintech saas especially Bloomberg and Factset. Nicolas lays out the risks that are coming for the incumbent terminals from the growth of LLMs
@marsnine At some point your agent knows you so well that is solves the blank page problem. @fintool recommends me the top things for me right now + works in the background on tasks to keep me updated on what's happening. [image]
Excellent and detailed guide from @nicbstme about the hard earned lessons he's learned from building, observing, and evaluating agents at Fintool. If you're building in finserv or really any vertical, this is a must read.
An excellent, nuanced take on the future of vSaaS. Would just add that if barriers of entry crumble and number of competitors (even if they only cover a low % of your feature set, they are a competitor) increases, CAC will structurally increase — you need to cut through the