Semiconductor Cross-Licensing: How Chipmakers Clear the Patent Thicket

Semiconductor cross-licensing is the quiet machinery that lets modern chips exist at all. A leading-edge processor can read on tens of thousands of patents held across dozens of companies, and no single firm owns a clear path to market. Rather than sue each other to a standstill, the major chipmakers trade access to their portfolios — each licensing the other under broad swathes of patents — so products can ship without an infringement suit landing on every design decision. Understanding how those deals work is essential to reading a chip company’s IP, valuing its portfolio, or diligencing it in a transaction.

What Semiconductor Cross-Licensing Actually Is

A cross-license is a two-way agreement in which each party grants the other a licence under some or all of its patents. In semiconductors it is rarely about a single invention; it is a portfolio-for-portfolio deal, where two companies that both hold thousands of patents agree not to assert them against each other’s products within a defined scope. The point is freedom to operate: each side buys the certainty that it can build and sell without the other’s patents blocking the way.

Semiconductor cross-licensing exists because the alternative is unworkable. A company shipping a complex system-on-chip cannot practically clear every one of the tens of thousands of patents its product might touch through individual negotiations. Cross-licensing collapses that impossible clearance problem into a handful of broad agreements with the other large patent holders, priced through a net balancing payment when one portfolio is stronger than the other.

The practice is old and deeply embedded. Semiconductors are the textbook example of an industry that runs on cross-licences precisely because innovation is cumulative — each generation of chips builds on patented work from many prior contributors — and because the products are so complex that overlapping rights are unavoidable rather than exceptional.

The Patent Thicket: Why Chips Cannot Ship Without It

Economists have a name for the condition semiconductor cross-licensing solves: the patent thicket. In his influential analysis Navigating the Patent Thicket, Carl Shapiro described it as ‘a dense web of overlapping intellectual property rights that a company must hack its way through in order to actually commercialize new technology’ — and named semiconductors, alongside software and biotech, as the fields where the thicket is thickest.

The scale is hard to overstate. Industry analyses estimate that a modern, AI-grade GPU or system-on-chip can implicate well over 100,000 patents, spanning everything from lithography and transistor structures to packaging, interconnects and the software instructions the chip executes. No company owns all of them, and many are held by direct competitors.

In a thicket that dense, litigation is mutually assured destruction: any large player can find some patent the other is infringing, so a suit invites a counter-suit, and both sides risk an injunction that halts their own products. Cross-licensing is the negotiated peace — the mechanism that converts a standoff of overlapping rights into a workable freedom to build. It is why two fierce rivals can each ship competing chips without the courtroom deciding who gets to sell.

Cross-Licensing, Patent Pools and SEP/FRAND: Not the Same Thing

Three mechanisms clear overlapping rights in chips, and they are often confused. A bilateral cross-license is a private, two-party trade of portfolio access. A patent pool aggregates many owners’ patents into a single bundled licence that any implementer can take on set terms — a ‘one-stop shop’ that is re-emerging in AI, data-centre and software-defined-vehicle silicon. And standard-essential patents (SEPs) licensed on FRAND — fair, reasonable and non-discriminatory — terms cover patents that cannot be avoided if a product implements a technical standard.

The distinction matters commercially. A cross-license is voluntary and negotiated on whatever terms the two parties accept. A FRAND commitment, by contrast, is an obligation the patent holder made to a standards body — it constrains what the holder can charge and bars it from discriminating between licensees, which is why standards-heavy parts of a chip (cellular, Wi-Fi, video codecs) are licensed differently from the proprietary logic.

For a chip company, the three interact. A cross-license may exclude the other side’s SEPs, which are handled separately on FRAND terms; a pool may sit on top of both. Reading a portfolio means knowing which patents are traded privately, which carry a FRAND leash, and which sit in a pool — because each category behaves differently in a licence, a lawsuit and a deal. We unpack the FRAND side in our guide to standard essential patents and FRAND licensing.

What a Semiconductor Cross-License Must Pin Down: 6 Rules

A cross-license lives or dies on its scope. Two agreements can look similar and grant wildly different freedom depending on how these six terms are drafted. Anyone negotiating — or diligencing — a semiconductor cross-licensing arrangement should read every one of them:

• Portfolio scope — does the grant cover the whole patent portfolio, a technology field, or a named list? A whole-portfolio grant buys the most certainty; a field-limited one leaves gaps.
• Capture period — does the licence reach patents filed after the signing date, and for how long? A frozen capture date means next year’s patents are not covered.
• Field of use — is the freedom limited to certain products, nodes or markets? A cross-license for one product line does not clear another.
• Have-made rights — can the licensee have chips manufactured by a third-party foundry under the licence, or only make them itself? This is decisive in a fabless/foundry world.
• Assignment and change of control — does the licence survive an acquisition, and does it travel to a buyer? This is where cross-licences collide with M&A.
• Balancing payment — where portfolios are unequal, the net royalty or lump sum that settles the difference, and how it is calculated.

Each of these is a place where a confident summary and the enforceable text diverge. A licence that a target describes as ‘a full cross-license with a major supplier’ may, on the page, be limited to one field of use, frozen at a 2019 capture date, and non-assignable — three facts that change what a buyer is actually acquiring.

How Cross-Licensing Shows Up in Valuation and M&A

Because semiconductor portfolios are so densely cross-held, cross-licensing is one of the first things a credible valuation or diligence exercise has to map. A patent family that reads on a rival’s product looks valuable in isolation — until you find it is already cross-licensed to that rival, at which point it excludes nobody and its assertion value is gone. The encumbrance, not the count, sets the worth.

This is exactly why cross-licences belong on an acquirer’s red-flag register. An outbound cross-license can mean the ‘exclusive’ technology a buyer is paying for is already in a competitor’s hands; a non-assignable one can evaporate at change of control, stripping freedom-to-operate the day the deal closes. Our M&A IP due diligence case study shows how a single overlooked licence-out reshaped a purchase price.

Reading cross-licences is also part of building a proper semiconductor patent landscape: the map is not just who owns what, but who has already traded rights with whom. Two rivals bound by a broad cross-license will not litigate each other no matter how their portfolios overlap, and a landscape that misses that reads the competitive dynamics backwards. The same discipline underlies any cross-licensing agreement, in chips or beyond.

Approaching a Semiconductor Cross-Licensing Negotiation

The leverage in a cross-license comes from portfolio strength and relevance, not raw count. A smaller portfolio of patents that genuinely read on the other side’s flagship products can command a better balance than a large portfolio of peripheral filings. So the work before a negotiation is analytical: map which of your patents the counterparty actually practises, and which of theirs you do, because that overlap is what sets the net payment.

Scope discipline matters as much as leverage. It is tempting to sign the broadest possible peace, but a grant that is too generous on capture period or field of use can give away future patents for nothing. The sharpest cross-licences are precise about what is traded and for how long, leaving room to negotiate again as portfolios and products evolve.

For most companies the practical trigger is a threat letter or a product launch into contested ground. Either way, the sequence is the same: read your own portfolio and the counterparty’s against the products in play, decide whether the cleaner route is a cross-license, a pool or a FRAND take, and negotiate the six scope terms above with the enforceable text — not the headline — in view.

How PerspireIP Can Help

At PerspireIP, our team helps innovators and businesses protect what they build. Whether you need a patent or trademark search, prior-art analysis, or an IP strategy tailored to your goals, we can help. Contact our team to discuss your next step.

Frequently Asked Questions