Platform strategy

Platform strategy is the set of decisions that determine what a product or company offers as a foundation others build on, how that foundation creates switching costs or network effects, and where the real defensibility sits. The sources here circle that question from several angles.

The clearest structural argument comes from internal developer tooling. Platform Engineering End-to-End frames internal developer platforms as the canonical example: a team owns a curated abstraction layer that other engineering teams consume, and the value is in reducing cognitive load and enforcing consistency. The platform team’s job is not to ship features for end users but to shift what counts as “normal” infrastructure inside the organization.

At the infrastructure level, Building a Cloud argues that the dominant cloud platforms are built on wrong abstractions — VMs, remote block storage, expensive networking — and that fixing those primitives is itself a platform strategy bet. Who controls the abstraction layer controls the economics downstream.

For AI tooling specifically, The Orchestrator Isn’t Your Moat makes a pointed claim: teams that invest in custom orchestration frameworks are building on sand, because frontier model providers will absorb that layer. The defensible move is to own domain-specific APIs and context that extend the platform, not to replicate it. This maps cleanly onto the broader logic that moats live in what the platform cannot easily internalize.

Pricing is increasingly a platform strategy variable. The AI Model Pricing War shows a 75x spread in token costs across providers, which means platform decisions about which model to route to directly determine unit economics and what business models are viable. Provider lock-in is a platform risk, not just a technical inconvenience.

Startup positioning intersects here too. The Founder’s Playbook argues that architectural decisions made at MVP stage determine what’s possible at scale — a quieter form of platform strategy where the platform is your own codebase. And Startups.RIP implies that failed platform bets leave the market education behind even when the company dies, meaning timing and positioning within an ecosystem often matter as much as the idea itself.

Defensibility that survives platform shifts tends to rest on what platforms cannot replicate. The Competitive Moat That AI Can’t Replicate locates that remainder in human trust relationships. SpaceX and the Sentient Sun illustrates the same logic at infrastructure scale: Starlink revenue and Falcon 9 reusability are the platform economics that fund everything else, compounding advantage across each layer.