A patent is a capital allocation decision dressed up as a legal filing. You spend real money now — typically $10,000–$20,000 in attorney fees plus USPTO fees for one US utility patent — in exchange for a right to exclude that may or may not produce a return over the following two decades. Most inventors never run that analysis. They ask whether they can get a patent, not whether the patent will earn back what it costs.
Three different things people call patent investing
- Filing on your own inventions. You are buying an exclusion right around technology you already built. The return shows up as margin you keep, competitors you slow down, or leverage in a negotiation.
- Buying issued patents from wind-downs, brokers, or auctions in order to license or assert them. This is a specialist business with its own diligence discipline, litigation budget, and failure rate.
- Funding another inventor's filings for a share of future licensing revenue. Rare, and the economics only work across a portfolio of dozens, never a single asset.
This article is mostly about the first one, because that is the decision an operating company actually faces. If you are weighing one filing on a straightforward consumer item, the narrower question of whether a patent pays off on a simple product is worth reading alongside this.
The real cost is not the filing fee
The number people quote is the drafting quote. The number that matters is lifetime cost.
| Line item | Typical US range |
|---|---|
| Provisional application, drafted properly | $2,500–$6,000 |
| Non-provisional drafting and filing | $8,000–$15,000 |
| Prosecution, two to three office actions | $3,000–$9,000 |
| Issue and publication fees | $500–$1,500 |
| Maintenance fees at 3.5, 7.5, and 11.5 years | $4,000–$13,000 total, depending on entity size |
| One foreign family (PCT into three national phases) | $25,000–$60,000 |
Carried to full term, a single US patent usually lands between $20,000 and $40,000 all in. A modest international family triples that. Small and micro entity discounts change the arithmetic materially, so check whether you qualify for micro entity status before you budget, and read the full US patent cost breakdown for the line-by-line version. The recurring payments are set out in the USPTO maintenance fee schedule.
Where the return actually comes from
Licensing royalties
The classic dream, and the least common outcome. Running royalties on consumer hardware typically land in the 2–7 percent range of net sales, higher for pharma-style exclusivity, lower for components buried in a larger assembly. A licensee is not buying your paperwork; they are buying reduced risk and a working technology. Patents with no prototype, no data, and no market proof rarely license at all. How licensing deals are actually structured covers the terms that decide whether the royalty ever arrives.
Sale of the asset
Individual patents on the open market often trade for less than they cost to obtain. Prices climb sharply when the claims read on a product a large company already ships, because then the buyer is removing a threat rather than acquiring an option. Valuation methods before a sale or license explains the three approaches buyers use.
Exclusion and pricing power
This is where most patents earn their keep, and it never shows up as a royalty line. If a patent keeps two copycats off your shelf space for three extra years, the return is the gross margin on the units you sold instead of them. That is measurable, and it is usually the honest business case.
Financing and exit value
Acquirers and investors discount unprotected hardware. A granted patent whose claims genuinely cover the shipping product tends to raise the floor of a valuation rather than the ceiling — it removes an objection instead of creating a premium.
The base rate nobody quotes you
A very large share of US patents are abandoned for non-payment before the final maintenance fee comes due. Those are decisions made by the owners who know the asset best, and they are voting that a four-figure check is not worth writing. Only a small minority of patents are ever licensed for money, and a far smaller minority are ever asserted. Any pitch that treats a granted patent as an automatic asset is ignoring the distribution: patent returns are heavily skewed, with a few assets carrying entire portfolios.
What separates a valuable patent from an expensive one
- Claim scope that maps to how a competitor would build it, not to your specific prototype. Narrow claims drafted around one embodiment are trivially designed around.
- Detectable infringement. If you cannot tell from a teardown or a datasheet that someone is practicing the claim, you cannot enforce it economically. Internal manufacturing processes score badly here.
- Family depth. One patent is a target; a family with a live continuation is a moving target, because you can still write claims aimed at a competitor's actual product.
- Clean chain of title. Unrecorded assignments and unsigned contractor agreements destroy value faster than weak claims do.
- Remaining term. Value decays toward zero as the twenty-year term from the earliest filing runs out.
When the money is better spent elsewhere
Patenting is a poor investment when the product cycle is shorter than the grant timeline, when your advantage is manufacturing know-how rather than a novel mechanism, when you have no realistic budget to enforce anything, or when the addressable market cannot support the legal spend. In several of those cases secrecy is genuinely stronger, and the trade secret versus patent comparison lays out the tradeoff.
How to improve the odds
- Do a real prior art search before you commit to drafting, so you are not buying a narrow claim set you could have predicted.
- File a provisional, then use the priority year to test demand and gather manufacturing data before the expensive filing.
- Budget for the family, not the filing. If you cannot fund prosecution and one continuation, file fewer applications and fund them properly.
- Review the portfolio at each maintenance fee window and abandon what no longer covers your product. Discipline about killing assets is what makes the survivors pay.
Projects House is an engineering firm, not a law firm — the filing, prosecution, and portfolio strategy belong with a registered patent attorney. What we do is the part that decides whether a patent is worth owning: building the invention into a product whose claims describe something people will actually buy. If you want an engineering read on whether your concept is defensible and manufacturable before you spend on filings, send the details through our contact form.