PCB Trade Secrets vs. Patents: How Hardware Founders Protect Electronics IP

Reasons for obtaining a utility patent on circuit topology

A patent gives you a 20-year right to exclude others from making, using, or selling the claimed invention, counted from the filing date. That's the headline benefit, and it's a real one. A few things follow from it.

You can enforce it even against someone who invented the same circuit independently. This is the single biggest practical difference from trade secret protection. If a competitor's engineer sits down cold and designs the same topology without ever seeing your board, a patent still covers them. A trade secret does not; independent invention is a complete defense to a misappropriation claim.

You can license it, and licensees will pay for the certainty. A trade secret is much harder to license since it is essentially a nondisclosure agreement and the licensee has to believe they and others are unlikely to reverse engineer it on their own. A patent license is a defined, recorded right. Investors and acquirers tend to value patented technology more highly for exactly this reason: it shows up on a balance sheet in a way a trade secret does not.

Patents are public, and publicity itself has value. A published patent application or granted patent is prior art the moment it publishes. That blocks competitors from patenting the same idea later, even if they came up with it on their own after seeing your product on a teardown site. Even if only narrow claims get allowed, a published patent can help prevent competition from obtaining patents in the field by acting as prior art against competitors.

Downsides to obtaining a patent

Circuit topology is hard to patent well. A lot of circuit designs, especially in power management, signal conditioning, and analog front ends, are combinations of well-known building blocks. Examiners will often reject claims as obvious over prior art, citing textbooks, app notes, or old patents showing the same op-amp configuration or filter topology. Getting allowable claims frequently means narrowing them to a specific combination of values, a specific topology detail, or a specific problem the design solves, and narrow claims are easier for a competitor to design around. Change one component value or restructure a feedback path and you may be outside the claim scope entirely.

Disclosure is mandatory and irreversible. To get a patent you have to describe the invention in enough detail that someone skilled in the art could build it. Once that application is published (generally 18 months after filing, sometimes sooner), the schematic-level detail is out there for anyone to read, whether or not you get the patent granted. If your real competitive advantage is in layout tricks, manufacturing tolerances, or firmware calibration rather than the topology itself, you may end up disclosing your best ideas while patenting a claim that's narrow enough to walk around.

It's slow and it's not cheap. Prosecution in the US typically runs one to three years and costs somewhere in the low tens of thousands of dollars once you include drafting, office action responses, and issue fees, more with foreign filings. Hardware products often have shorter market windows than that. By the time the patent issues, the product generation it covers may already be obsolete.

Enforcement against hardware infringement is its own project. Proving that a competitor's board practices your claims usually requires physical teardown, X-ray or decapping, and sometimes expert testimony on circuit equivalents. That's expensive litigation-support work layered on top of the underlying suit.

Reasons for treating layouts and Gerbers as trade secrets

Trade secret protection costs nothing to obtain and there's no waiting period. Protection starts the moment you take reasonable steps to keep the information secret and lasts indefinitely, as long as secrecy holds. Coca-Cola's formula has been a trade secret for over a century. Patents don't last anywhere near that long.

Trade secrets cover things patents can't. Trade secrets can protect the layout optimizations, the component sourcing that makes yield acceptable, the thermal tricks, the exact stackup, and every other bit of hard-won manufacturing knowledge that isn't inventive enough to patent but is still valuable enough that a competitor would kill for your Gerber files. A lot of what actually makes a board manufacturable at scale falls into this bucket rather than into “novel circuit topology.”

Trade secrets do not require any public disclosure. You never have to explain how it works to anyone outside the people you choose to tell. That matters most for designs where the “how” would be genuinely hard for a competitor to reconstruct even if they wanted to.

But trade secret protection has a weakness that patents do not, which is that trade secrets only stop misappropriation, not independent discovery. If a competitor reverse-engineers your board through lawful means, such as buying it on the open market and tearing it down, they're free to use what they learn. If they design the same circuit independently, they are also welcome to practice what they design. The law protects the secret, not the idea, and the secret is only as good as your ability to keep it out of other people's hands or heads.

To get trade secret protection requires maintaining “reasonable secrecy measures” which is a real ongoing job, not a checkbox. Courts look hard at whether a company actually did anything to protect the information, which includes NDAs with contractors and fabs, access controls on design files, marking documents confidential, limiting who touches all the files, including the Gerbers. Skip this and a court may decide you never had a protectable trade secret to begin with, no matter how sensitive the information was.

Trade secrets offer nothing against a competitor who gets there without stealing from you. This has always been the tradeoff. What's changed recently is how easy reverse engineering and discovery has become.

AI is eroding the assumptions around how practically hard it is to reverse-engineer

For years, a lot of PCB trade secret protection rested on an unspoken assumption, which is that someone could theoretically reverse-engineer your board, but it would take a skilled engineer weeks of manual tracing, X-ray inspection, and delayering to reconstruct the schematic and layout. That difficulty functioned as a practical shield even where the legal protection of a trade secret was thin.

That shield is weakening due to recent developments which are important to know when deciding whether to patent or keep a PCB as a trade secret.

There is now automated netlist extraction from images and X-ray/CT scans. Companies and researchers have built machine-learning pipelines that take photographs, X-ray images, or CT scans of a populated board and automatically identify traces, vias, and component footprints, then reconstruct a netlist. Work that used to require a technician tracing copper by hand under a microscope can now be substantially automated, particularly for two- and four-layer boards.

There is improved component and package recognition from images. Vision models can now identify component families, and sometimes exact part numbers, from photographs or die images with reasonable accuracy, including in cases where the markings have been sanded off or blacked out. That undercuts a design obfuscation technique that used to be fairly effective on its own.

Delayering plus reconstruction is getting faster and cheaper. Combining automated optical delayering with image-stitching and ML-based trace recognition means multilayer boards, which used to be much more resistant to reverse engineering than simple ones, are becoming tractable targets too. The cost of getting a usable schematic out of a competitor's product is dropping, and the skill floor required to do it is dropping with it.

These tools are increasingly available off the shelf. This kind of reverse engineering used to require a specialized failure-analysis lab and years of institutional expertise. Some of it is now approachable with commercial or open-source tooling and a service bureau, which puts it within reach of smaller competitors who never could have justified building that capability in-house.

None of this means trade secret protection is dead for hardware. Plenty of the real value in a board (manufacturing know-how, test procedures, calibration data, supplier relationships, firmware) still isn't recoverable just by staring at the copper tracings, however good the imaging gets. But if your trade secret theory depends on “nobody will bother reverse-engineering this,” that assumption deserves a second look for anything valuable enough to attract attention. AI-assisted reverse engineering also works in the other direction and can increase the benefit of a patent. These same techniques can help you or a forensic expert prove that a competitor's product falls within your patented PCB layout.

How you should decide

There's no universal answer, but a few questions tend to point clearly in one direction or the other.

Is the innovation in the topology itself, or in how you built it? If you've invented a genuinely new circuit arrangement that solves a known problem in a non-obvious way, a patent is worth pursuing, especially if the design will show up in a product a competitor can buy and study. If the real edge is in layout, tuning, and manufacturing know-how sitting on top of a fairly conventional topology, trade secret protection probably fits better, since a patent claim on that kind of improvement is often hard to get and easy to design around anyway.

How long is the product's real competitive life? A design that will be obsolete in 18 months gets little value from a patent that takes two years to issue. A platform architecture you expect to sell for a decade is a much better patent candidate.

How exposed is the board? Something that ships inside a sealed, potted, or otherwise physically protected enclosure is harder to reverse-engineer than a board sold as a standalone product or one that's easy to access inside a consumer device. Exposure level should factor into how much weight to put on trade secret protection alone.

Can you actually keep it secret? If the design has to be shared with multiple contract manufacturers, assembly houses, and component suppliers across different countries, ask honestly whether “reasonable secrecy measures” are realistic to maintain. A trade secret that leaks through the supply chain is not protected.

What's the budget and timeline? Patents cost real money and take real time. For an early-stage company with one shot at a design before running out of runway, trade secret protection plus solid NDAs and access controls may be the only practical option, with patenting reserved for the specific innovations that matter enough to justify the spend.

A hybrid approach usually wins

Most hardware companies end up doing both, and that's generally the right call. Patent the genuinely novel circuit-level inventions that will still matter years from now and that a competitor could plausibly design around anyway once your product ships. Keep the layout files, manufacturing process details, test fixtures, and calibration data as trade secrets, protected by NDAs with every fab, assembler, and contractor who touches them, access controls on the design repository, and clear internal policies about what counts as confidential.

Given how much easier automated reverse engineering has become, it's worth treating the “assume it will eventually be extracted” scenario as the default when deciding what to patent versus what to keep secret, rather than the exception. If a board is valuable enough that a competitor would put in the effort to reverse-engineer it, assume someone eventually will, and structure your protection accordingly. For the filing-side traps that come up once you do decide to patent, see our companion guide on eight patent pitfalls hardware and PCB engineers must avoid.

If you're working through this decision for a specific design, bring the schematics and a sense of your product roadmap to a call. The right answer usually becomes clear once we know what's actually novel, how long the product needs to stay ahead of competitors, and how it will physically reach the market.

This article is intended as general information for clients and does not constitute legal advice on any specific design or fact pattern. Talk to counsel before making patent filing or trade secret decisions for a particular product.