A founder once refused to tell me what his product did until I’d signed an NDA. Fair enough. Then he spent the better part of a year trying to patent every angle of it before building a single unit. The invention was real. The fear was real too. But while he guarded an idea that existed only on paper, the market moved, and someone less cautious shipped something close enough. He protected the idea so carefully that he never gave it a chance to become a product.
I want to be upfront about my footing here. I’m a hardware engineer, not a patent attorney. What follows is how I’ve come to think about this from the product and engineering side, not legal advice, and the details vary a lot from one country to the next. For anything specific to your situation, talk to a patent attorney. Treat the specifics below as my best understanding, worth confirming, not as settled law.
What a patent actually is, and what it isn’t
A patent is a territorial, time-limited right to stop other people from making, using, or selling your invention, granted by one country and valid only there. Two words in that sentence do most of the work.
Territorial: a US patent does nothing in China, and a Canadian patent does nothing in Germany. You get protection country by country, and each one costs money and time.
Right: a patent is a right to sue, not a force field. It doesn’t stop a copy from existing; it gives you standing to go to court over it. Enforcement is on you, and litigation is slow and expensive. As I understand it, a patent runs about twenty years from its filing date, but those twenty years are yours to defend, at your cost.
To be patentable, an invention generally has to be new, non-obvious, and useful. And here’s a shift a lot of people my age or older still get wrong: most of the world, including the United States since 2013, is now first-to-file. The date you file matters more than the date you invented. Sitting on an idea to “establish” you had it first is, as best I understand it, no longer how it works.
Where the provisional patent fits
In the US, a provisional patent application is a cheaper, simpler filing that locks in a priority date and lets you legitimately say “patent pending.” It isn’t examined, and it never becomes a patent on its own. You get twelve months to file the full application claiming that early date, or the provisional lapses and does nothing.
The catch most people miss: its value depends entirely on whether it actually describes the invention well enough. A vague provisional written in a weekend protects very little, because you can only later claim priority to what you genuinely disclosed in it. Cheap to file, easy to file badly.
“Provisional” in that exact form is a US mechanism. Other countries reach a similar twelve-month head start through the Paris Convention and instruments like the PCT (the international application that holds your place in many countries at once). Canada, where I’m based, doesn’t have provisionals as such, but a first regular filing can serve as your priority date. Again: this is the part to confirm with an attorney for your jurisdiction, because I am genuinely not certain of the mechanics in every country.
There’s one trap worth flagging even from a non-lawyer, because it collides directly with how engineers work. In the US there’s a grace period, roughly a year, for your own public disclosures. In much of the rest of the world, Europe in particular, any public disclosure before you file can destroy patentability outright. That means showing the product at a trade show, or running an open field test, before filing can quietly cost you patent rights abroad. If foreign patents matter to you, file before you go public. I’m fairly confident about the broad shape of this, less so about every exception, so confirm it.
What the internet and globalization changed
The classic patent playbook was built for a world where copying was slow, physical, and local. That world is gone.
Today a product can be bought, photographed, reverse-engineered, and quoted by an overseas contract manufacturer in a matter of weeks. Your design files, your bill of materials, your Gerbers pass through several hands just to get built at all. The information wants to travel, and it travels instantly.
So you get the mismatch at the center of all this: a patent is territorial, but copying is global. You might hold a solid US patent while an identical unit gets made and sold in markets where you will realistically never litigate. Filing broadly enough to matter internationally, a PCT application followed by national-phase filings in each country, turns into real money spread over years. For most early hardware companies, that money buys more when it’s spent on getting to market.
None of this makes patents worthless. They still signal something real to investors and acquirers, they’re a genuine deterrent against a well-funded domestic competitor, and they’re a defensive asset if someone ever comes after you. But as a moat for a small team, a patent is weaker and slower than founders tend to imagine.
The protection that actually holds: speed and field testing
Here’s the reframe that helps most of the founders I work with. The idea is rarely the hard part, and rarely the thing that’s defensible. Making it work reliably, in the real world, at a cost that makes sense, is the hard part. And that knowledge does not fit inside a patent or a teardown.
Someone can buy your product and copy the visible design. What they cannot lift from it: the months of field data that told you which failure modes actually happen versus the ones you feared, the firmware tuning that only came from real users doing unexpected things, the supplier relationships, the calibration process, the second and third revisions you shipped while they were still cloning your first.
By the time a copyist reverse-engineers version one, a team that keeps building is on version three, with customers, a brand, and a support relationship. That lead compounds. A patent is a moat you rent from a legal system and have to defend; a shipping lead built on field experience is one nobody can file for.
This is where field testing early does double duty. It’s how you actually build the product, and it’s how you build the lead that protects it. The speed of your iteration is the defense.
Let me be careful not to turn that into recklessness. This is not “skip IP.” If there’s a genuinely novel core to what you’re doing, a cheap early provisional, or your country’s equivalent, is smart insurance, and it’s entirely compatible with moving fast. Use NDAs with your manufacturers. Mind the public-disclosure trap before you demo widely. The point is proportion, not neglect. Protect the one truly novel thing cheaply, then get moving.
The honest version
In my experience the founders who lost their idea mostly didn’t lose it to a patent-savvy thief studying their filings. They lost it to their own delay, guarding an unbuilt concept for months while someone more willing to ship simply got there first. An idea locked in a drawer, undisclosed and unbuilt, is the least protected idea of all. Protect the genuinely novel core cheaply, then go build the lead that no filing can give you and no copy can take.