How Much Can You Sell a Patent For? The Real Numbers

Ask how much you can sell a patent for and every answer is a range. Five thousand dollars to millions. Twenty thousand to five hundred thousand. Tens of thousands to seven figures. All of them are technically true and none of them helps, because a range that wide is just the question restated with dollar signs.

A distribution is more useful than a range. So: what the distribution looks like, what sets the ceiling inside it, and what is left once everyone standing between you and the buyer has taken their share.

The number the ranges hide

Across US patent transactions, the median sits at roughly 77,000 USD, the top quartile of deals averages 280,000 to 650,000, and elite assets in AI, autonomous vehicles and semiconductors routinely transact above a million, on figures drawn from Richardson Oliver Insights IP3 reports covering 2016 to 2024.

Read those three numbers in order and the shape of the market appears. Most completed sales are five figures. The comfortable six-figure outcomes every guide quotes are the top quarter. Seven figures is rare enough to be news.

That distribution also counts only the patents that sold. Underneath it sits the base rate: on a market basis, the value of roughly 99 percent of issued patents is zero, because an acquirer has no reason to take on maintenance and annuity costs for an asset nobody is infringing. If you are holding a report that says otherwise, that gap has its own explanation in why a valuation report and a real sale price rarely match.

So the honest answer arrives in two parts. Most likely nothing. And if it does sell, anchor on the median and treat every step above it as something you have to earn, input by input.

What sets the price is the buyer's alternative, not what you spent

One sentence does most of the work in patent pricing and it is usually left undeveloped. A buyer weighing an acquisition asks whether it can monetize the patent, whether it blocks a competitor or fills a gap in its portfolio, and what its alternative is: design around the patent, or acquire it.

That last clause is the pricing model. Your patent is worth what it saves the buyer against its next best option, and the next best option is usually engineering rather than a payment.

Designing around is not a loophole either. It is how the system is built. Under the all elements rule an accused product must include every element of the relevant independent claims to infringe, so infringement can be avoided by avoiding the language of the claims, and the courts treat that negative incentive as a benefit of the patent system rather than a defect in it. If a competitor can drop one element of your independent claim without hurting its product, your ceiling is roughly the cost of that engineering change. Not the size of the market you pictured protecting.

And where nobody is infringing, there is no alternative to price against at all. An uninfringed patent has little or no value until the industry catches up with the invention.

Notice what never enters the calculation: what the patent cost you. Not the R&D, not the attorney fees, not the years. A buyer is not reimbursing your investment. It is buying a lever against somebody else.

The four inputs that move you inside the distribution

Evidence that someone is already practising the claims

This is the input that separates a 20,000 dollar outcome from a 300,000 dollar one, and it is the one sellers most often arrive without. Not a belief that the claims are being practised. A mapping: this company, this product, this claim element, this evidence. Everything else on this list adjusts the number. This one decides whether there is a number.

Claim breadth

Broad independent claims covering many implementations are worth multiples of claims tied to one specific configuration, for the design-around reason above. The test is not how clever the invention is, it is how many ways there are around the words.

Remaining life

Patents run 20 years from the earliest effective filing date. Ten years left gives a buyer a decade of leverage. Three years left is a rounding error. It is also the only input on this list that moves in one direction while you think about it.

Family size and jurisdictions

Single-country coverage is a discount. Patents with international counterparts are worth 10 to 25 percent more than those covering only one country, because the buyers worth having sell in more than one market and want the same claim enforceable where they compete.

Pending versus granted, priced

Sellers ask constantly whether they can sell before grant. You can, and the market has a number for what it costs. Buyers typically pay 40 to 60 percent less for a pending non-provisional application than for a granted patent, since there is no guarantee the claims survive examination, while applications already in allowance command close to full patent prices.

Half your price, in other words, is sitting inside a decision the examiner has not made yet. That is worth knowing before you accept an early offer. It is also worth saying out loud from the other side of the table: our own first listed patent family is a filed and pending application, not a granted patent, and we state that every time it comes up.

The gross price is not your number

Nearly every page answering this question stops at the headline figure. Sellers get paid the net.

Broker commissions typically run 20 to 40 percent of the net sale price, scaling down for larger deals, and asking prices for issued US patents with demonstrable commercial use cluster at 200,000 to 350,000. Run the arithmetic on a 300,000 dollar sale at the middle of that commission band and roughly 210,000 reaches you, before attorney time on the assignment and before the annuities you kept paying while the process ran.

The cheaper channel charges less because it does less. Marketplace commissions run 4 to 10 percent, and a marketplace listing typically resolves in 30 to 90 days against 6 to 18 months selling direct to a large company. The gap in fees is not generosity, it is the difference between a venue and a process: a listing does not create a buyer. Which one fits depends less on the percentage than on whether you already know who should own the asset, which is really a question about the route you pick to reach buyers.

Incentives deserve the same daylight. Our brokerage is success-based: commission is paid by the seller on completion, the minimum fee is EUR 5,000 per completed transaction, and no sale means no fee. A retainer model gets paid for the process. A completion model gets paid for the outcome. Neither is dishonest, but they behave differently the month an asset turns out to be unsellable, and you should know which one you have hired.

Why we price detectability before anything else

Everything above collapses into one practical problem: proving use. Our validation gate runs before an asset goes anywhere near a buyer, using physics-grade claim reconstruction, detectability analysis and validity probability, benchmarked at 0.76 percent mean absolute percentage error against real-world outcomes.

Detectability is the unglamorous half and it moves prices hardest. A claim that is genuinely being infringed, but only inside a process nobody outside the factory can inspect, prices like a claim nobody is infringing at all. Buyers are not paying for your certainty. They are paying for what they could put in front of a court.

Two limits, stated plainly: EX IX is not a firm of patent attorneys, and a novelty search does not guarantee grant or patentability. Filing work runs through an independent licensed attorney of record.

How to put a realistic number on your own patent

1. Read your independent claims and list every element. If a competitor can remove one without hurting its product, you are pricing a design-around, not a market.
2. Name a company selling something that reads on all of those elements today. If you cannot name one, you are not pricing a sale yet.
3. Ask whether that use is provable from outside the company, from a teardown, a datasheet, a public filing.
4. Check the earliest effective filing date and count the years left. Under five changes every other answer.
5. Count the family. How many jurisdictions, and is it still open?
6. Anchor on the median and adjust upward only for inputs you can evidence to a stranger.
7. Net it down. Commission, attorney time on the assignment, and the annuities you will pay while the process runs.

If step two comes back empty, the live question was never price. It is the renew, abandon or monetise decision. And if you are earlier than a granted asset, the arithmetic starts further back, with what it costs to file in the first place.

FAQ

What is the average patent sale price, and why is it so far above the median? One analysis of more than 43,000 telecom and IT transactions found an average near 484,000 USD per document, against a median near 77,000. Averages in this market are dragged upward by a small number of very large portfolio deals. A single seller should plan against the median and let the average stay a statistic.

Can you sell a provisional patent application? Yes, a provisional can be assigned, but the buyer has to file a non-provisional inside the 12-month window or lose the priority date. That narrows the field to buyers who want the technology itself and intend to run prosecution in house, which is a very different conversation from selling a granted asset.

Is licensing more profitable than selling outright? They are different shapes rather than different sizes. Licensing keeps title and pays over time if a licensee performs and keeps reporting honestly. A sale is one payment that also ends your maintenance liability and your enforcement problem. The choice follows your cash needs and your appetite for policing the asset for a decade, not a higher expected total.

Does a patent with no product behind it have any sale value? Sometimes, though rarely for the reason inventors expect. A buyer is not paying for the idea, it is paying for the right to exclude someone who is already selling something. With nobody to exclude, the price sits near zero however good the invention is, until an industry moves toward it.

How long after agreeing a price before the sale is actually final? The assignment has to be recorded with the USPTO to be enforceable against third parties, and recordation typically takes three to eight weeks after signature. That sits on top of the months spent finding the buyer, so a deal agreed in principle is not money in the account for another one to two months.