06-reference

stratechery microsoft software survival

Mon Feb 02 2026 19:00:00 GMT-0500 (Eastern Standard Time) ·reference ·source: Stratechery ·by Ben Thompson
microsoftsaasai-codingazuresoftware-economicsaggregation-theory

Microsoft and Software Survival

Thompson argues Microsoft’s stock hammering after Azure growth missed by a point was misplaced — the company deliberately allocated scarce GPU capacity to its own products (M365 Copilot, GitHub Copilot, R&D) over Azure resale, a rational portfolio decision even if Wall Street punished it.

The core framework: AI-written code won’t kill software companies, it will intensify competition between them. Software vendors remain valuable because most companies don’t want to maintain custom code — they want products with support, compliance, and integrations. But the cost of code trending toward zero means every SaaS company can now attack adjacencies, collapsing the neatly siloed SaaS ecosystem. Thompson frames this as shifting from “growing the pie” to “fighting for it.”

The piece introduces the “token foundry” concept — pure-play cloud providers (neoclouds, Oracle) may have an underappreciated advantage over hyperscalers who prioritize their own products. Microsoft’s capacity allocation made this tension explicit: Azure customers are third in line behind Microsoft’s own products and OpenAI.

On agents and identity: Microsoft’s “Work IQ” layer built on Active Directory makes strategic sense but faces a structural headwind — per-seat licensing shrinks as human headcount shrinks. This creates urgency for Microsoft to absorb more business functions.

RDCO Mapping

The SaaS consolidation thesis and “adjacency wars” framing is directly relevant to how we think about vertical software opportunities. The observation that AI makes code cheap but products remain valuable reinforces RDCO’s positioning as a services/product hybrid. The token foundry concept is worth tracking for infrastructure decisions.