Who Buys Patents? The 5 Buyer Types and What They Pay For

Who buys patents is a shorter list than most sellers expect. Five types of entity buy them: operating companies, non-practicing entities (NPEs), defensive aggregators, universities and research institutes, and startups. Two of those five do almost all of the buying, and one of them is the type most inventors have been told to avoid. The pages ranking for this question give you the list and stop. We want to give you the numbers behind it: how much each buyer type buys, what it selects for, what it pays, and the honest filter at the end, which is who it buys from. That last part decides whether your patent ever gets an offer. ## The short answer: five buyer types, and two of them do most of the buying Start with the size of the pond. The brokered secondary patent market, the part of the market where listed packages change hands and someone tracks it, was about USD 158 million in 2024, in line with 2022 and 2023, and brokers were involved in 87 percent of sold packages. That is a small market for an asset class with millions of live patents in it. It is also not an exchange. The market is quasi-public: roughly 600 to 1,000 packages a year, 50 to 100 new ones a month, sourced by contacting each broker individually, with no listing service and no published prices. A buyer in this market is not browsing. It is receiving teasers from brokers it already knows, screening them against criteria it already wrote down, and passing on nearly all of them. The five buyer types, in the order we meet them in deal flow: 1. Operating companies, which make products and buy patents to protect them. 2. Non-practicing entities, which hold patents to license or assert them. 3. Defensive aggregators, which buy on behalf of member companies so nobody else can. 4. Universities and research institutes, who are almost entirely on the selling side. 5. Startups, who buy selectively, usually by licence, for freedom to operate. ## Operating companies: the deepest pockets and the slowest closings ### Why they buy The clearest practitioner description still comes from a broker who had sold more than 2,000 patents: operating companies look at third-party assets to complement their own portfolio or to avoid litigation where they may be infringing, they rarely buy to monetise, which gives them more price flexibility and deeper pockets than most NPEs, they pay in cash, they care about licence grant-backs, and they take longer to close because buying patents is not their core business. Over the last five years, operating companies bought 25 to 45 percent of the packages on the brokered market, buying defensively to preempt NPEs, offensively for licensing revenue, to block competitors, or for cross-licensing leverage in crowded fields like smartphones and semiconductors. Their share of assets is far higher than their share of packages, because when a corporation buys, it tends to buy a whole family or a whole portfolio. ### What they select for Corporate buyers screen before they read. One of the market's long-time advisers to corporate buying programmes says that defining the business need first eliminates over 75 percent of diligence cost, that an asking price above five times market is a reason to walk away and above three times signals an unsophisticated seller, that first-time buyers should buy in their own technology field, and that the corporate development team's passed-over acquisition targets are a private source of patents nobody else sees. The lesson for a seller is that a corporation buys what fits a gap it has already named, not what is clever. ### Who they are in practice The names are familiar. In the second quarter of 2025, Samsung was the most frequent buyer with six deals and 199 assets, IBM sold patents to Abnormal AI, Anthropic and Dexcom, and Acer sold 17 electronics assets to a licensing company that runs the BlackBerry portfolio. Operating companies are also the biggest sellers, pruning what they no longer need. If you are considering selling a patent to a company directly, the sibling article covers the door you have to knock on and why most cold pitches never reach it. ## Non-practicing entities: the buyers who price on enforcement ### The three flavours An NPE's business is holding patents, not practising them. The practitioner taxonomy splits them into large aggregators building licensing programmes, defensive funds (covered below as their own type), assertion entities that monetise through litigation, and a few sovereign IP funds buying with public money to protect a national industry. In the market's formative years, sales to these financial buyers were more than 75 percent of transactions and more than 60 percent of market value, and their arithmetic was set by the roughly USD 5 million it takes to fund offensive patent litigation through trial. The largest of them has changed shape. Intellectual Ventures acquired approximately 75,000 assets and has sued on only 148 unique US patents, and across the brokered market only 10.2 percent of sold packages ended in litigation at all. Most NPE-held patents are licensed or sit in a portfolio, which is a different thing from the troll caricature. ### What an NPE pays for The buyer study of Ocean Tomo's auctions is the best evidence on selection. The probability that a traded patent goes to an NPE rather than a practising firm rises with triadic family members and backward citations, and falls with the number of inventors and with how large a technological advance the invention represents. NPEs want legally robust patents in consolidated technology trajectories, with international coverage and remaining term. Originality works against you with this buyer, because a pioneering invention has nobody infringing it yet. The current picture is larger than the caricature suggests. In the second quarter of 2025, 45 NPE buyers took 45 percent of all assets sold, with litigation funders enabling acquisition at scale (the AST quarterly report already linked above). And NPEs are the buyer class most open to unrepresented sellers: they source primarily from private holders, individual inventors, startups and small companies, and from public companies only about 29 percent of the time. ### The risk you inherit by reputation The trade-off is in the Richardson Oliver data already cited: between 30 and 52 percent of packages sold to NPEs end up with at least one litigated patent. The price an NPE offers is a discount on expected enforcement value, often structured as a hybrid with a back-end share so you carry part of the risk. Whether that is acceptable is a question about who your invention's users are, because they are who gets sued. ## Defensive aggregators: buying so nobody else can ### RPX and Allied Security Trust Defensive aggregators are the market's answer to assertion entities. RPX and Allied Security Trust (AST) are the most purely defensive buyers: they pool their members' private information about which patents threaten them, buy those patents before an assertion entity can, license their members, and in AST's case run an explicit catch-and-release model that resells the patent afterwards, with the result that patents acquired by defensive entities are measurably more valuable than those acquired by assertion entities. In the auction data already cited, defensive aggregators prefer narrower scope, higher technical merit and longer residual life than the average NPE. ### The cheque sizes This is the one buyer type that publishes what it pays. AST's annual open-submission programme for 2024 had more than 30 members including Cisco, Google, IBM, Intel, Meta and Microsoft review over 1,000 valid submissions from more than 300 sellers and brokers, invested USD 4.2 million in 51 deals, saw asking prices of USD 10,000 to 250,000 per deal with an average of USD 81,829 per lot, and found that 96 percent of deals asked under USD 500,000 and 94 percent were single-family offerings. Twenty-seven of those 51 deals were sublicences rather than purchases, under a model where the owner keeps title. That explains the market-share shape in the Richardson Oliver report: defensive aggregators bought 20 to 40 percent of packages but only 2 to 22 percent of assets, because they buy small, and their share of outright purchases is expected to fall as they shift toward licensing without ownership. For a seller holding one granted family that reads on big-tech products, a defensive aggregator's submission window is the one institutional door that is genuinely open to single patents. ## Universities, research institutes and startups: mostly the other side of the table Universities appear constantly in transaction data, but as sellers. In the first quarter of 2025, universities and research institutions were sellers in 111 deals, 24 percent of all transactions, while 296 operating companies sold 2,802 assets in 305 deals and the top buyers by deal count were an NPE, Samsung, Allied Security Trust and Volkswagen. A university technology transfer office licenses out. It almost never buys in. Startups buy, but selectively and usually not by assignment. The buyer classes include startups needing cleared freedom to operate, and the same patent can command very different prices because a strategic acquirer may pay a premium for exclusivity while an aggregator or NPE prices to its own monetisation model. For a startup, an exclusive licence in its field is normally the right instrument, which is why owners should understand how to sell patent rights carved by field of use before they assume the only deal is whole title. ## Who does not buy patents, and why your inbox is quiet Now the filter. The honest version of the corporate buyer is that few companies actively seek to acquire patents from outside sources, and even where a patent fits perfectly most will pass because external acquisition is not part of their development model, specifically not from individual owners, while the aggregators that do buy are hard to reach without knowing the right person inside. The buyer universe is also concentrated. In one full-year study of the brokered market, 138 buyers purchased 261 packages, 30 repeat buyers accounted for 59 percent of them, NPEs took 42 percent of packages, defensive aggregators 19 percent and operating companies 37 percent, and Intellectual Ventures' purchases had fallen from 40 to 13 to 6 packages before it announced the end of its buying programmes. A few dozen repeat players, each receiving up to 100 packages a month, from brokers they already trust. The Richardson Oliver report says it plainly: individual inventors have a harder time selling than operating companies and NPEs do, and thousands of patents are listed every year with rarely a competing buyer. That is the structural reason behind how hard it is to sell a patent on your own. ## Matching your patent to the buyer that will actually pay Every buyer type above shares one gate. The broker quoted earlier put it in one line: across 2,000 patents sold, the factor that trumps every other is whether the patent is infringed. Operating companies will not pay for something nobody practises, and NPEs cannot monetise it. With that fixed, the mapping is short. ### A single granted patent that reads on a shipping product This is the asset the market exists for. Realistic first doors are a defensive aggregator's submission programme and NPE buyers, with a strategic operating company possible if the product is theirs or a competitor's. Price expectations should start from the Richardson Oliver medians: a 2024 asking price of USD 163,000 per asset and USD 286,000 per US-issued patent, with sold assets at a median asking price of USD 95,000, and closing prices below asking. The sibling on how much you can sell a patent for works through the tiers. ### A small family in a crowded high-tech field Wireless, semiconductors, software and media are where most packages trade. Here you compete with operating-company sellers pruning hundreds of assets at once, and your buyer is more likely to be an NPE or aggregator sorting packages by claim chart than a corporation. Triadic coverage and remaining term matter more than elegance. ### A pending application, or a patent nobody practises Say it plainly: most buyers will not pay for either. A pending application sells only when a buyer can see what the claims will cover and that someone already practises them, which is why selling a patent that is still pending is a different conversation. An unpractised patent is a licensing or build decision, not a sale, and the patent licensing companies that might take it on will price it accordingly. One warning that applies to every path: do not contact potential infringers yourself. The broker literature cited above explains why, since a solicitation can hand the recipient legal standing to challenge your patent before you have a buyer. ## How we see buyers from inside a live deal flow We sit on both sides of this map. EX IX brokers third-party patents in an M&A-style process of valuation, buyer targeting, controlled outreach and deal support, on a success commission paid by the seller on completion with a minimum fee of EUR 5,000, so no sale means no fee, and separately sells or licenses the rights it holds, carved by industry, territory or field of use. That second activity means we are also a selective buyer and licensor, and it shapes what we think a seller should know. The buyer taxonomy above has a practical consequence: corporate IP teams managing thousands of patents do not browse listings, and the repeat buyers who matter screen teasers from people they trust. So the work that earns a response is the work done before outreach. Every asset we represent passes a validation gate before it reaches any buyer: physics-grade reconstruction of the claims, a detectability analysis of whether infringement can be observed in a product, and a validity probability, benchmarked at 0.76 percent mean absolute percentage error against real-world outcomes. A teaser that explains exactly which gap a patent fills in a buyer's portfolio is what turns a cold file into a reviewed one. What a patent broker does, and when you do not need one, is covered separately. The limits belong in the same paragraph. EX IX is not a firm of patent attorneys, and a novelty search does not guarantee grant or patentability. Nobody can promise you a buyer, and anybody who does is either a finder for a single purchaser or has not read the sell-through data. What a good process can promise is that the right five or ten buyers will look, and that is the whole game in a market with no exchange. ## FAQ ### Do big companies like Google or Apple buy patents from individual inventors? Rarely directly. They buy through brokers and through consortium programmes such as Allied Security Trust's annual IP3 window, where members including Google, Microsoft, Meta and IBM review single-family submissions on a fixed timetable. A cold email to the company itself goes to an unsolicited-ideas policy, not to the IP team. ### Is Intellectual Ventures still buying patents? No. Its open-market purchases fell from 40 to 13 to 6 packages across three consecutive years before it announced the end of its buying programmes, and NPE buying rose anyway because other entities, many of them backed by litigation funders, took its place. ### How can I tell whether a buyer is an NPE? Practitioners use a simple heuristic: a buyer named as an LLC with no products is usually an NPE. Check the USPTO assignment record for what else the entity owns, and check whether those patents have been litigated and against whom. ### Do patent buyers prefer granted patents or pending applications? Granted, with remaining term and evidence of use. NPE buyers specifically select for longer residual life and international family members, and defensive aggregators for technical merit and remaining life. A pending application sells only when the buyer can see what the claims will cover and that someone already practises them. ### Do investors buy patents as an investment? Litigation funders and some institutional investors back NPE acquisition campaigns, and a few sovereign IP funds buy selectively to protect a national industry. They buy through intermediaries rather than from inventors, and they price on enforcement outcomes, not on the technology.