Wall Street's idea that a software pure play will “vanish” into an LLM is “nonsense”: we need more software, and AI-enabled software moves up the product stack
Too much of what is going on is a race to get to a theoretical end state of a whole new world of business and technology.
@stevesiSteven Sinofsky
Context & Ripple Effects
The argument pushes back on a software-market narrative that has persisted long enough to weigh on valuations: fears of an AI-driven software extinction event have damaged software stocks even as the thesis remains contested. It instead treats LLMs as an enabling layer, not the endpoint for every product.
That view extends the earlier case that LLMs lower the barriers to creating software. The meaningful question becomes where applications retain control: in the workflow, distribution, and higher-level product experience built around the model.
First-order effects
Software pure plays gain a clearer strategic test: add AI in ways that improve the end product and move the offering up the stack, rather than treating the model itself as a substitute for the application.
Investors evaluating software businesses must distinguish exposure to model commoditization from the value of the application layer the company controls.
Second-order effects
Application vendors face pressure to show durable workflow ownership and differentiated distribution as AI features become easier to build.
LLM providers may compete more intensely for placement inside software products, while software companies seek to avoid being reduced to interchangeable interfaces.
Third-order effects
If this framing holds, AI value capture will be distributed across a stack rather than concentrated solely in foundation models; the strongest positions may sit with products that own deployment and customer workflows.
The persistence of the extinction narrative despite contested assumptions suggests that software-market repricing will remain sensitive to evidence of whether AI replaces applications or expands what they can deliver.
The trend: This is one data point in the shift from viewing AI as a replacement for software to viewing it as a layer that changes where software differentiation and value accrue.
@stevesi “The number of processes and experiences in work and life that are not yet fundamentally improved by software is far greater than the number that have been improved by software.” 💯
@stevesi To add, I suspect software companies are better equipped to be AI-powered and agent-ready than, say, retail companies were to adopt internet sales. Constant change and R&D is norma; in software, and so much of AI is programming-adjacent. It is software and hardware, afte…
@JI Maybe but why wasn't IBM equipped to capitalize on the PC they invented or Kodak on the digital camera they invented? These were companies that invented more stuff than just about any other company.
Box CEO @levie's defense of software over vibe-coded, n-of-1 internal tools: “If you're Ford, and you're doing your supply chain on an ERP system, you want that to work the exact same way every single time.” “The billions of transactions going through that ERP system, you [video]
The “death of software” thesis commits the same category error as “death of retail”... it assumes a fixed pie of economic activity that gets reallocated. The reality is that software expands to fill the available capacity for automation, and AI doesn't reduce that capacity, it
Really solid articulation on the death of software, or not, from somebody who has a unique depth of experience on the software industry changes over time:
Excellent article. This is actually a bull case for software because TAM will explode and never before addressed usecases will be taken up by ai/human driving ai.
A+ post - “what is absolutely part of this whole arc are people who are certain we are less than five years away and are in a rush to build with absolute belief in where things are heading, and people who support them with their labor or dollars.”
“The most important thing about the PC is that the first predictions were de minimis, followed by the prediction that it would eliminate mainframe computing and the data center. HAHA. Everyone was wrong all around.” Great, great piece by Steven.