A 3 trillion dollar selloff in response to Claude shipping paralegal-adjacent capabilities is an over-reaction.
This is not the invisible hand declaring an indefinite moratorium on your Quickbooks/Salesforce/XYZ license. It's much more a symptom of information asymmetry, cognitive lethargy, and an opportunity for serious re-branding by software companies.
Capital Flows and Cognitive Lethargy
Capital flows are largely controlled by non-operators. This is more significant than people realize. Jobs today are rightfully specialized. Accordingly, fund managers and investors rarely use 95% of enterprise software themselves.
The result — it becomes cognitively easy to treat automation and disintermediation as immediate, binary, and irreversible.
"Oh, AI can do even this now" is too simplified an explanation. Yes, agents are incredibly value additive, but they don't yet have access to decision traces, non-verbal context, and still hallucinate in catastrophic ways. Buying into a strawman when you're not in the weeds yourself — using this stuff in specific workflows — is inevitable.
Great SaaS is SPECaaS
Treating TurboTax as just tax calculation software is equivalent to treating OpenAI or Anthropic as just an LLM provider. Yes, the model is table stakes, but so is the RLHF environment, compute capacity, public policy, trust and safety, taste, judgement, and the myriad other aspects needed to transform from a product to an actual company.
Great SaaS companies are actually SPECaaS — Software, People, Expertise, and Collaboration/Community.
The market will bounce back, but it's a good opportunity for all of us to move beyond thinking of code as the only, or even primary, moat.
Meanwhile, can some of the labs just go public already? That way we can rob Peter to pay Paul all we want.
My S&P 500 ETF will be eternally grateful.
For two compelling (and far more elegantly written) essays on this topic, check out Dan Hockenmaier's LLMs vs Marketplaces and The Software Shakeout.