Open vs Closed Platform

In 1981, IBM made a decision that would define the personal computer industry for four decades. Under pressure to ship a PC before Apple ran away with the market, IBM's engineers used off-the-shelf components and an open architecture — any manufacturer could build compatible hardware, any developer could write compatible software. The IBM PC became the standard. Within five years, Compaq, Dell, and dozens of clones had flooded the market with IBM-compatible machines at half the price. IBM's PC division haemorrhaged market share, and by 2005 the company sold what was left to Lenovo for $1.75 billion. The open architecture that won the market made it impossible to control the market. Meanwhile, Apple sold the Macintosh as a closed system — proprietary hardware, proprietary software, proprietary peripherals. Apple captured 8% of the PC market and 35% of the profits. IBM won adoption. Apple won economics. The pattern has repeated in every platform war since.

The fundamental tradeoff has not changed in forty years. Open platforms win adoption. Closed platforms win monetisation. Android holds 72% of global smartphone market share. Apple's iOS holds 27%. Android generates 30% of global app revenue. iOS generates 65%. Open: more users, lower margins, fragmented experience. Closed: fewer users, higher margins, controlled experience. Google makes its money not from Android itself — which it gives away for free — but from the services layered on top. Apple makes its money from the hardware, the App Store's 30% commission, and the services ecosystem that the closed platform enables. Two architectures, two business models, both wildly successful by different metrics.

The spectrum matters more than the binary. No successful platform is fully open or fully closed. Linux is nearly fully open — anyone can modify and redistribute the source code. The result: Linux runs 96% of the world's top one million servers and earns essentially nothing for any single company. Windows is moderately closed — proprietary code, controlled updates, but open to any hardware manufacturer and any software developer. The result: decades of dominance in enterprise computing with consistent 70%+ margins. iOS is highly closed — proprietary hardware, curated App Store, strict developer guidelines. The result: 27% market share, the most profitable smartphone business in history, and an ecosystem that generated $85 billion in App Store revenue in 2023. The degree of openness determines the adoption curve. The degree of closure determines the value capture.

Amazon Web Services won cloud computing by occupying a strategic position on the spectrum. AWS was more open than the proprietary alternatives — IBM, Oracle, SAP — that had locked enterprises into expensive, rigid infrastructure for decades. Enterprises could run any operating system, any language, any framework on AWS. But AWS was less open than pure open-source alternatives — it built proprietary services (Lambda, DynamoDB, SageMaker) on top of the open infrastructure that created switching costs and captured margin. The pattern: open enough to win adoption from customers fleeing proprietary lock-in, closed enough to build proprietary value that prevented them from leaving once they arrived. By 2024, AWS generated $90 billion in annual revenue at 30%+ operating margins.

Shopify's Tobi Lütke described the strategy as "arming the rebels" — building a platform open enough for any merchant to start selling in minutes, closed enough that Shopify controlled the economics of every transaction. The platform is open in the sense that matters for adoption: any product, any business model, any design template. It is closed in the sense that matters for monetisation: Shopify Payments processes the transactions, Shopify Shipping handles the logistics, Shopify Capital provides the lending. The merchant gets independence from Amazon's marketplace. Shopify gets a percentage of every dollar the merchant earns. Open at the top, closed at the bottom. The adoption layer is free. The infrastructure layer is owned.

The pattern that emerges across every platform war since IBM: winning platforms start open to build network effects, then selectively close to capture value. Facebook launched as an open platform — any developer could build apps, any user could join. Once the network effects compounded to two billion users, Facebook progressively closed: restricting API access, reducing organic reach for business pages, and channelling commercial activity through paid advertising. Google's Android started open to capture market share from iOS, then selectively closed: pre-installing Google apps, requiring hardware partners to meet compatibility standards, and making the open-source version (AOSP) progressively less functional without Google's proprietary services layer. The opening gets you the users. The closing gets you the revenue. The art is in the sequencing.