When an Intangible Asset Becomes Collateral
For venture lending, a patent matters only when it can answer a lender’s recovery question. This chapter follows the path from a startup’s ownership of an intangible asset to a recorded security interest, then asks whether the right could retain value outside the company if repayment fails. The distinction helps a reader judge patent-backed debt without assuming that a patent is either worthless because it is intangible or valuable simply because it exists.
A patent is not collateral until the claim is made legible
A patent gives its owner legally defined rights over an invention. It does not, by itself, give a lender a payment stream. The startup may own a patent and still be unable to sell it, license it, or use it successfully. Its commercial value may depend on the original engineering team, a product that has not reached customers, or a market that never develops. A lender considering venture debt needs a separate right: a security interest that connects the loan to an asset and gives the lender a recognized claim if the borrower defaults.
In the U.S. lending market studied by Yael Hochberg, Carlos Serrano, and Rosemarie Ziedonis, venture lenders commonly take a blanket lien over a startup’s assets. “Blanket” matters because the lender is not necessarily betting on one carefully valued patent. The lien can cover the company’s assets broadly, including patents, so that the lender has a claim over what remains valuable if the business cannot repay. When patents are included, the security interest is recorded with the U.S. Patent and Trademark Office. That record helps establish the lender’s secured status and communicates the lien to other potential lenders.
The record changes the financing relationship in two ways. First, it makes the claim enforceable and visible within the relevant legal system. A lender that has only an informal promise may have to compete with other claimants or prove what was pledged. A recorded security interest provides a clearer basis for asserting rights. Second, it makes the company’s financing state legible to outsiders. Another lender can see that assets may already be encumbered rather than treating the patent portfolio as unclaimed collateral.
If a borrower defaults and enters bankruptcy, a secured creditor generally has priority over unsecured claimants in the distribution of whatever value remains. That priority is important, but it is not the same as a guarantee. Priority determines who is paid first from recoverable value; it does not create recoverable value. If a patent cannot be sold or licensed for enough money, being first in line may still leave the lender with a loss. The legal claim solves the question “Who has the right to pursue the asset?” It does not solve “How much will the asset produce?”
The security arrangement also affects the startup before default. The threat of losing control over pledged assets can give the borrower an incentive to avoid actions that make repayment less likely. This is one reason collateral can reduce a lender’s risk even when the lender hopes never to seize it. Yet the same protection can constrain a company’s choices. A blanket lien may make it harder to offer the same assets to another financier, and a default can threaten inventions that the company needs to keep operating. The borrower receives financing capacity in exchange for giving the lender a stronger claim over future assets.
The parties also see different versions of the same patent. For the startup, it may be a strategic building block whose value depends on keeping the team together and continuing development. For the lender, it is a contingent recovery right that matters only if the company cannot meet its obligations. For another lender, it is an asset that may already be encumbered. For a potential buyer, it is a technology right whose usefulness must be assessed independently. The public record connects these perspectives, but it does not make their valuations identical.
That separation is important in a default. Enforcement does not instantly convert a patent into cash. The lender may need to establish the scope of the claim, protect its priority, locate a buyer, and wait for a transaction. During that period, the asset may lose value as technology changes or as the failed company’s related know-how disappears. A patent with strong legal protection can still have weak economic recovery prospects. Conversely, an asset with several plausible users may be more valuable to a lender even if its contribution to the original startup’s product was difficult to measure.
The blanket lien reflects this uncertainty. Rather than pretending that one patent can be priced precisely at the time of the loan, the lender may take a claim over the company’s asset base and rely on the combined value of what survives. That arrangement can improve expected recovery, but it also means that the lender’s protection may reach beyond the particular invention discussed during underwriting. The startup should therefore understand collateral as a continuing constraint on financing and operations, not as a one-time description attached to the loan documents.
The U.S. study’s data trail illustrates both the importance and the opacity of this interface. The researchers found many more patent security interests than the standard venture-finance database reported. That gap is not evidence that the recorded liens were fraudulent or that every missing deal had the same structure. It shows that the legal record and the financing database answer different questions. A patent-office record can reveal that a security interest exists even when the size, price, covenants, and use of proceeds are not publicly visible.
A patent pledge therefore has four separate layers:
The startup owns the patent. The lender receives a security interest rather than the patent itself. The public record makes the lien visible. A default can trigger enforcement, while a buyer represents the separate possibility that the patent can move to another user. Collapsing these layers into the sentence “the loan is backed by patents” hides the economic problem that determines the lender’s expected loss.
Recovery depends on a market, not just a document
The missing layer is salvage value: the amount a lender expects to recover if the startup cannot repay and the pledged asset is seized. Salvage value is not the same as the patent’s value to the original company. A patent can be strategically important to its owner yet difficult for anyone else to use. Conversely, a right that did little for the failed company may be attractive to a competitor, a technology buyer, or an organization assembling a complementary portfolio.
Finding that next user is a market process. Buyers and sellers must locate each other, understand what rights are being transferred, assess legal and technical quality, and decide whether the asset fits another business. In a thin market, there may be few buyers and high search costs. A lender expecting a quick sale after default must discount the asset for the possibility that the right sits idle or sells only after a costly delay. In a thicker market, more potential matches can make the next-best use easier to find. That can raise the lender’s expected recovery and make lending more attractive even before any default occurs.
This is the logic behind the patent-market liquidity measure in the U.S. study. The researchers did not observe each lender’s private estimate of recovery. Instead, they estimated how likely patents in a relevant sector and recent age window were to be traded in a given year, then adjusted that measure to the startup’s portfolio. Their proxy treats more frequent trading as evidence of a market with more opportunities for reallocation. It does not say that every patent in that sector will sell, or that a recorded sale price equals the amount a lender would recover in bankruptcy.
Market thickness matters most when an asset is redeployable. A patent is more redeployable when another owner can use, license, defend, or combine it without needing the failed company’s entire team and commercial strategy. It is more firm-specific when its value depends heavily on the original company’s specialized knowledge, product architecture, customer relationships, or unfinished development work. The distinction is continuous, not a binary legal category. A patent can be partly transferable while still losing much of its value when the original organization disappears.
The study approximates firm specificity through citations. It measures the share of follow-on patent citations that come from the focal startup itself within three years of the patent being granted. A high self-citation share is treated as a sign that the invention’s subsequent use is concentrated inside the original firm, and therefore may be less redeployable. This is a useful research proxy because direct observations of post-failure patent value are scarce. It is not an appraisal of a particular patent. Self-citation can reflect many features of a technology and does not prove that other companies cannot use it.
The empirical prediction is consequently conditional. If more active trading improves expected salvage value, a rise in market liquidity should increase lending more for redeployable patents than for firm-specific patents. The preferred regression in the study follows that pattern. The interaction between firm specificity and patent-market liquidity is negative: lending responds less to a thicker market when the startup’s patents are more tied to the original firm. At illustrative points in the sample, a one-percentage-point increase in liquidity is associated with a 1.35-percentage-point increase in the annual debt rate for highly redeployable patents, about 1.16 points at average specificity, and 0.4 points for highly firm-specific patents.
These are estimated changes in the probability of receiving patent-backed lending, not a promised return on a patent. They also do not mean that liquidity causes every individual lender to approve a loan. The market measure is built at the sector and year level, and the specificity measure is indirect. The study includes startup fixed effects and other controls, which make the pattern more informative than a simple comparison, but lender expectations and realized liquidation proceeds remain unobserved. The evidence supports a salvage-value mechanism; it does not show a universal patent-pricing formula.
The scale of the observed lending market helps put the mechanism in perspective. The study follows 1,519 U.S. venture-capital-backed startups in software, semiconductor devices, and medical devices that received at least one patent. By 2008 or exit, 36% had secured a loan using patents. More than 25% of the 14,514 patents awarded to the sample startups had been involved in at least one security-interest agreement. Among startups with loans, an average of 92% of the patent portfolio was under lien by the final reported transaction year. The high portfolio share is consistent with the common use of blanket liens rather than a lender selecting one isolated invention.
The data also show why this market is difficult to see from standard financing databases. Silicon Valley Bank was the most common lien holder in the sample. The researchers identified 239 annual debt deals between the bank and sample startups, but only eight—3%—appeared in the VentureOne database. A study based only on reported equity and debt rounds would therefore miss much of the patent-backed activity. The patent-office record is not merely a legal detail; it is also an unusual measurement window into an otherwise opaque market.
Still, prevalence is not proof of recovery or benefit. The 36% figure applies to patent-producing, VC-backed companies in three innovation-intensive sectors and to debt visible through patent security interests. It excludes startups without patents and may miss loans secured by other assets or recorded through other channels. It does not tell us how often borrowers defaulted, how much lenders recovered, or whether taking debt improved the companies’ eventual outcomes. The fact that a patent was encumbered shows that a lender accepted the collateral arrangement. It does not show that the asset would have paid the loan in a failure.
That boundary is the practical lesson. When evaluating an intangible asset as collateral, ask first whether the lender’s legal claim is clear. Then ask whether a secondary market contains plausible buyers, whether the asset can be separated from the original team, and how much time and cost a sale would require. Finally, keep recovery in its proper place. It can reduce the lender’s expected loss and support a loan that would otherwise be difficult, but it cannot replace a credible operating or refinancing path. The next question is how the venture investor’s own commitment adds another form of protection when the startup’s assets alone are not enough.