Users can now prompt an AI to build a custom version of a SaaS tool, tailored to their exact needs. This marks a shift towards personal, disposable software, which increases software's abundance while simultaneously eroding the moats of traditional SaaS businesses.

Patrick Collison suggests AI fundamentally changes software economics. Instead of a fixed-cost product sold at scale, software will become bespoke, created on-demand for individual users at the moment of consumption, similar to ordering a custom pizza. This introduces variable inference costs.

Just as YouTube lowered media distribution costs, AI is lowering software development costs. This could shift the SaaS market away from large, one-size-fits-all platforms toward a model where small, elite teams deliver highly customized software solutions directly to enterprise clients.

The dominant per-user-per-month SaaS business model is becoming obsolete for AI-native companies. The new standard is consumption or outcome-based pricing. Customers will pay for the specific task an AI completes or the value it generates, not for a seat license, fundamentally changing how software is sold.

Software has long commanded premium valuations due to near-zero marginal distribution costs. AI breaks this model. The significant, variable cost of inference means expenses scale with usage, fundamentally altering software's economic profile and forcing valuations down toward those of traditional industries.

AI tools aren't just making employees more efficient; they are replacing human labor. This allows software companies to move from cheap per-seat pricing to a new model based on outcomes, like charging per support ticket resolved, capturing a much larger share of the value.

AI is moving beyond enhancing worker productivity to completing entire projects, like drug discovery or engineering designs. This shift means software will be priced like a services business, based on the value of the outcome delivered, not the number of users with access.

The traditional SaaS model—high R&D/sales costs, low COGS—is being inverted. AI makes building software cheap but running it expensive due to high inference costs (COGS). This threatens profitability, as companies now face high customer acquisition costs AND high costs of goods sold.

The next major business model shift in software is from seat-based pricing to outcome-based pricing (e.g., paying per task completed). This favors AI-native newcomers, as incumbents will struggle to adapt their GTM and financial models.