Licensing

Licensing is not a universal model. It works brilliantly under specific conditions and fails badly when those conditions are absent. The model rewards companies that are better at inventing than at manufacturing, distributing, or selling.

The underlying logic is straightforward: licensing makes sense when you are the best in the world at creating something, but not the best at producing or selling it — and when the gap between creation and production is wide enough that both parties profit from the division of labor. ARM designs better processor cores than most chipmakers could design themselves, and it does so at a fraction of the cost of maintaining an in-house design team. The licensee gets world-class IP without the R&D burden; the licensor gets revenue without the manufacturing risk.

Licensing looks elegant on paper — high margins, low capex, infinite scalability. In practice, it is riddled with failure modes that can destroy value quickly.

The most dangerous failure mode is licensee vertical integration, because it is both predictable and nearly impossible to prevent. Every successful licensing relationship teaches the licensee what your IP does and how it works. Over time, the licensee's internal capabilities improve, and the gap between "build" and "buy" narrows. Apple's decision to design its own silicon — first for iPhones, then for Macs — was a direct consequence of years spent as ARM's and Intel's customer. The licensee learned enough to become a competitor. The only defense is to innovate faster than your licensees can replicate, keeping the IP frontier perpetually ahead of their internal capabilities.

Licensing businesses are deceptively simple to model at the top line — IP in, royalties out — but the unit economics reveal critical nuances about sustainability, leverage, and risk.

The key lever is operating leverage. Because R&D costs are largely fixed and marginal licensing costs are near zero, every incremental licensee or unit shipped drops almost entirely to the bottom line. A licensing business that grows revenue 20% while growing costs 5% is printing money. The danger is the reverse: if the addressable market contracts (fewer smartphones, fewer devices using your standard), revenue falls but R&D costs remain fixed, and margins compress violently. This is why the best licensing businesses obsessively expand the number of application domains for their IP — ARM moving from smartphones to IoT to automotive to data centers is a textbook example of expanding the denominator.

Licensing businesses compete on a fundamentally different axis than product companies. The battle is not for customers — it is for standards, ecosystems, and the right to be embedded.

The primary source of competitive advantage is standard-setting power. If your IP becomes the industry standard — or a mandatory component of one — licensing revenue becomes structurally embedded in the market. Qualcomm's CDMA and LTE patents are woven into the cellular standards that every phone manufacturer must comply with. ARM's instruction set architecture is the default for mobile computing. Dolby's audio and video codecs are required by Blu-ray, streaming services, and broadcast standards. Once your IP is in the standard, competitors cannot design around it — they must license or exit the market.

The second moat is ecosystem lock-in. When thousands of software developers write code for ARM's architecture, or millions of peripherals are designed for USB (whose patents are held by a consortium), switching to an alternative architecture imposes enormous costs on the entire ecosystem. The IP holder benefits from collective switching costs that no single licensee would bear. This is why RISC-V, an open-source instruction set, has taken over a decade to make meaningful inroads against ARM despite being free — the ecosystem cost of switching is measured in billions of dollars of existing software, tools, and trained engineers.

Licensing markets tend toward oligopoly or monopoly within a given standard, but fragmentation across standards. Within mobile wireless, Qualcomm dominates. Within chip architecture for mobile, ARM dominates. Within x86 computing, Intel and AMD share a duopoly built on cross-licensing agreements. But across these domains, different licensors control different standards, and the competitive dynamics are more about expanding your standard's territory than defeating a rival within the same standard.

Competitors respond to a dominant licensor in three ways: design-around (engineer a solution that avoids the patented claims — expensive and often inferior), vertical integration (build the capability in-house to eliminate the royalty — Apple's silicon strategy), or open-source alternatives (create a royalty-free standard that commoditizes the incumbent's IP — RISC-V versus ARM, VP9/AV1 versus H.264/H.265). The most dangerous competitive threat is the open-source alternative backed by a consortium of well-funded companies who are collectively motivated to eliminate the licensing tax.

Licensing manifests with dramatically different economics and dynamics depending on the industry. The royalty rate, enforcement complexity, and competitive structure vary widely.

Licensing rarely emerges fully formed. Companies typically arrive at it after discovering that their IP is more valuable than their products — or they evolve away from it when they realize they're leaving too much value on the table.

Coming from: Many licensing businesses start as product companies that discover their IP is the real asset. IBM spent decades as a vertically integrated hardware company before pivoting toward patent licensing and services — its patent portfolio generates over $1 billion annually. Qualcomm began as a satellite communications company, built CDMA technology, and realized that licensing the standard was more valuable than manufacturing handsets (it eventually sold its handset division). The pattern: build IP to support a product, discover the IP is worth more than the product, restructure around licensing.

Going to: The most common evolution is toward subscription or usage-based pricing, particularly in software. Microsoft's shift from perpetual Windows licenses to Microsoft 365 subscriptions is the canonical example — same IP, different monetization cadence. Adobe made the same transition with Creative Cloud. In hardware IP, the evolution is often toward ingredient branding — ARM's "Powered by ARM" and Dolby's logo on every device are licensing relationships that double as brand assets, commanding premium positioning for the licensee's product.

Adjacent models: Franchising shares licensing's DNA (monetizing a system rather than operating it) but includes operational control. White-label is licensing's less defensible cousin — the IP is generic enough that the licensee could replicate it. Revenue share aligns incentives more tightly than fixed royalties but introduces audit complexity.