A blog post about the most important intellectual property decision most startup founders will never think about until it's too late, which is to say: a blog post about time, and money, and the peculiar bureaucratic machinery that converts the former into the latter, and also sometimes the reverse.
Here is a thing that is true: a startup with patents raises 47% more venture capital than a startup without them. Here is another thing that is true: most founders, upon receiving the congratulatory letter from the United States Patent and Trademark Office informing them that their patent has been granted, will do something that feels entirely natural and is in fact catastrophic. They will celebrate. They will update their LinkedIn. They will move on.
And the window—the one nobody told them was open, the one through which a single patent could have become five patents, or ten, or a thicket so dense that competitors would need machetes and a team of litigation attorneys to get through—that window will close.
Permanently.
Without so much as a sound.
I. What is a Continuation? Both Simpler and More Bizarre Than You'd Expect
The U.S. patent system, in its infinite and occasionally Kafkaesque wisdom, permits something called a "continuing application." The concept is governed by 35 U.S.C. §§ 120 and 121, and if you've never read those statutes—and you almost certainly haven't, because life is short and patent law is long—here is the essential idea: you can file a new patent application that claims the benefit of your old patent application's filing date, provided you do so while the old application is still pending.[^1]
[^1]: The word "pending" is doing a tremendous amount of work in that sentence, and we will return to it later with the kind of obsessive focus it deserves.
There are three flavors of continuing application, and they matter:
A continuation uses the exact same technical specification as the parent—same drawings, same description, word for word—but presents entirely different claims. You cannot add new technical disclosure. What you can do is ask for protection over different aspects of what you already described, which turns out to be enormously powerful if your original specification was written with any foresight whatsoever.
A continuation-in-part (CIP) lets you add new technical material to the specification—say, a new algorithm you developed eighteen months after the original filing—but here's the catch, and there's always a catch: only the claims supported by the original disclosure get the parent's priority date. Claims covering the new stuff get the CIP's later date, which means they're vulnerable to any prior art that published in between. CIPs are, in this sense, a bit like those restaurant dishes that combine lobster and hot dogs. You can do it. People do. But you should understand what you're getting.
A divisional is filed when the USPTO examiner decides your application contains more than one distinct invention and issues a "restriction requirement," which is the examiner's way of saying pick one. The non-elected inventions can be pursued via divisional applications, and these enjoy a special statutory "safe harbor" under § 121 that protects them from double patenting attacks based on the parent. This protection is worth remembering.
Now, here is the part where many founders' eyes glaze over—which is exactly the part that determines whether their IP is a picket fence or a fortress. Every continuation, regardless of type, expires 20 years from the filing date of the earliest nonprovisional application in the chain. Not 20 years from its own filing date. This is non-negotiable. A continuation filed 8 years after the parent has roughly 12 years of enforceable life remaining. The clock, as they say, is always running.
And listen: there is no statutory limit on how many continuations you can file. The USPTO tried to cap it at two back in 2007, but those rules were enjoined in Tafas v. Doll and never took effect, which is one of those rare instances where the federal judiciary did something that was unambiguously good for small inventors, and which we should probably appreciate more than we do.
One more distinction, because people confuse this constantly, and the confusion has consequences: a Request for Continued Examination (RCE) is not a continuation. An RCE reopens prosecution in the same application, with the same serial number, before the same examiner who already rejected your claims and who will, in many cases, reject them again in substantially similar terms. A continuation creates an entirely new application—new serial number, potentially new examiner, clean prosecution history. This difference is not academic.
II. The Case For
Or, Why the Smartest People in the Room File Continuations
I want to tell you a story about a company called Sonos.
According to Harrity & Harrity's Patent 300 data, Sonos derived 88% of its patents in 2023 from continuation applications. Eighty-eight percent. Palantir, Dolby, and eBay maintain similarly astronomical continuation rates. These are not companies run by people who are confused about how to allocate resources. These are companies that have figured out something that most startups haven't: a patent specification is not a product. It is a seed bank.
These are companies that have figured out something that most startups haven't: a patent specification is not a product. It is a seed bank.
Here is what I mean. When you file a patent application, the specification—the detailed technical description—discloses a technology. The claims define the specific legal boundaries of what you own. But the specification almost always describes more than the claims cover. A good patent attorney will pack the specification with alternative embodiments, variations, and implementations precisely because the specification is the reservoir from which all future continuation claims must be drawn. You cannot add anything later.[^2] Everything you'll ever claim in a continuation must already be lurking in the original description, like money sewn into a mattress by a grandparent who survived the Depression.
[^2]: Unless you file a CIP, which, as previously noted, comes with its own set of complications that we might charitably describe as "nontrivial."
So: your patent issues. Congratulations. You have claims covering your product as it existed on the day your patent attorney wrote them. But your market is not static. Your competitors are not static. Your own product roadmap is not static. And this is where the continuation becomes, in the words of our attorney Samar Shah, a tool for ensuring claims read on a competitor's new product.
That idea deserves to sit with you for a moment.
You can watch a competitor launch a product. You can study its technical implementation. And then you can file a continuation with claims drafted to read directly on that product—claims that nonetheless enjoy the priority date of your original filing, which predates the competitor's product entirely. This is legal. This is common. This is how the game is played.
We recommend a specific two-phase approach: file initial claims of modest scope to secure fast issuance (useful for fundraising, for putting "patented" on your website, for the general signaling value that makes investors feel warm), then pursue broader claims through continuations once prosecution reveals the prior art landscape. Get the narrow patent quickly. Get the broad patents strategically. This is not complicated. It is, however, rarely done.
We recommend a specific two-phase approach: file initial claims of modest scope to secure fast issuance (useful for fundraising, for putting "patented" on your website, for the general signaling value that makes investors feel warm), then pursue broader claims through continuations once prosecution reveals the prior art landscape. Get the narrow patent quickly. Get the broad patents strategically. This is not complicated. It is, however, rarely done.
The Money Part
Because founders care about money—and should, because money is oxygen for startups, and you cannot pivot gracefully while suffocating—let us talk about the financial implications.
Research by Hsu and Ziedonis found that doubling a startup's patent stock was associated with a 28% higher pre-money valuation in venture capital deals. An Israeli startup study found each additional patent application corresponded to roughly a 45% increase in valuation in subsequent financing rounds. The Farre-Mensa et al. "Patent Lottery" study found that a first patent increased five-year employment growth by 55% and sales growth by 80%.
These numbers are not subtle.
And continuations are, comparatively, cheap. Because they reuse the parent specification—same drawings, same description—they avoid the $10,000–$20,000+ cost of drafting an entirely new application. A continuation filing at a boutique firm typically costs $3,500–$6,500 for new claims and prosecution. You are, in effect, getting a new patent for a fraction of the original investment. This is the closest thing to arbitrage that exists in intellectual property.
For M&A specifically, the value proposition is even more compelling. An acquirer looking at a startup with a pending continuation sees something different than one looking at a startup with only issued patents. The pending application means the IP can still be shaped—sculpted to cover the buyer's product roadmap, aimed at the buyer's competitors, expanded into technical areas the buyer cares about but the startup never commercialized. Multiple sources confirm that M&A buyers value open continuation families specifically because they represent future optionality. You are not just selling a portfolio; you are selling the ability to build a portfolio.
The Litigation Part (Which Is Grimmer But No Less Important)
A patent family with multiple continuation patents featuring distinct claim sets is substantially harder to kill than a single patent. If a challenger takes down one patent through an inter partes review—a proceeding that has, since 2012, become the weapon of choice for defendants who'd prefer not to have patent claims aimed at their revenue streams—the remaining family members with different claims may survive intact. You cannot kill a hydra by cutting off one head. This is basic mythology, and it is also good patent strategy.
Moreover, and this is crucial: a startup with a pending continuation can draft new claims in response to a competitor's design-around or validity challenge, creating what patent strategists call a moving target. The claims don't exist until you write them. The specification already supports them. The priority date already predates the competitor's product. It is, if you'll forgive the martial analogy, the intellectual property equivalent of holding a reserve force in perpetuity.
III. The Case Against
Or: The Part Where I Tell You About the Costs and Risks, Because Honesty Demands It and Because There Are Enough Hucksters in IP Strategy Already
I would like to be straight with you.
Continuations cost money. Each one incurs the same USPTO filing, search, and examination fees as a brand-new application—approximately $2,000 for large entities, $1,000 for small entities, and $500 for micro entities. Attorney fees for drafting claims and prosecuting the application add $17,000–$24,000 on average. Lifetime maintenance fees run another $5,800 for small entities. BlueIron IP estimates the average U.S. patent costs roughly $60,000 over its full lifecycle. Multiple continuations create what I'll call a "cost treadmill"—a phrase that doesn't appear in any patent law textbook but which accurately describes the experience of a seed-stage startup paying its third continuation filing fee while also trying to make payroll.
And then. And then. As of January 19, 2025, the USPTO introduced a brand-new surcharge specifically targeting late continuations. File a continuation more than six years after the earliest benefit date: $2,700 surcharge (large entity) or $1,080 (small entity). More than nine years? $4,000/$1,600. These fees, codified at 37 CFR § 1.17(w), represent the USPTO's explicit policy preference that you should stop filing continuations after a while, please. We'll discuss timing implications below, but the short version is: the government has made procrastination expensive.
Prosecution History Estoppel
Or: The Thing That Can Ruin Your Day in Court Years Later
This is where it gets technical, and I apologize, but the technicality matters.
Under the Supreme Court's decision in Festo Corp. v. Shoketsu Kinzoku Kogyo Kabushiki Co. (2002), any narrowing amendment or argument made during prosecution to overcome a prior art rejection creates a presumption that the applicant surrendered coverage over equivalents. In plain English: if you change your claims to get around prior art, you generally can't later argue that your patent covers things equivalent to what you gave up.
Here is the nasty part, the part that practitioners sometimes discuss with a kind of grim relish, like oncologists describing a particularly aggressive tumor: this estoppel extends across the entire patent family. The Federal Circuit held in Fate Therapeutics that estoppel from a narrowing amendment in a later-issued continuation applied retroactively to limit the scope of an earlier-issued related patent. Let me restate that to make sure we're feeling the appropriate level of discomfort: a careless argument your attorney makes in continuation #3 can narrow the enforceable scope of patent #1, which issued years earlier and was never itself amended.
Each continuation filing creates additional prosecution history. More prosecution history means more material for defendants to mine for damaging admissions. More continuations mean more exposure. There is a real and underappreciated tension between the strategic value of filing many continuations and the litigation risk of creating many prosecution records, and resolving this tension requires the kind of careful, experienced patent prosecution that costs real money, which brings us back to the cost treadmill, which never stops.
Double Patenting: The Bureaucratic Paradox
The doctrine of obviousness-type double patenting (ODP) exists to prevent a patentee from obtaining multiple patents on essentially the same invention, thereby extending the effective monopoly period beyond 20 years. When your continuation claims overlap with your parent claims—which they will, almost inevitably, because you're claiming different aspects of the same disclosure—the examiner will reject them on ODP grounds and require you to file a terminal disclaimer.
A terminal disclaimer is an irrevocable agreement that your continuation patent expires no later than the parent patent, and—this is the important part—that all terminally disclaimed patents remain commonly owned. You cannot sell them separately. You cannot license them to different parties. If you're a startup hoping to monetize individual patent assets, this is a meaningful constraint.
The Federal Circuit's 2023 decision in In re Cellect made things worse by holding that patent term adjustment—extra time the USPTO adds to your patent when it takes too long to examine your application—does not insulate patents from ODP. The August 2024 Allergan v. MSN Laboratories decision provided partial relief: a first-filed, first-issued, later-expiring parent patent cannot be invalidated by its own child. And now, in early 2026, the PTAB's Ex Parte Baurin Appeals Review Panel is considering these issues with amicus briefs due in March—which means the rules may shift again.
This is, to put it mildly, a lot to keep track of. Vonnegut would say: And so on.
Diminishing Returns Are Real
The USPTO itself acknowledged, in a 2006 Federal Register notice, that the exchange between examiners and applicants suffers from diminishing returns with successive continuations. After two or three well-targeted continuations covering commercial embodiments and likely design-arounds, additional filings tend to yield claims that are either too narrow for commercial value or too similar to existing claims—triggering more ODP headaches without proportional benefit. Sophisticated investors evaluate patent quality, not raw count. Multiple weak continuations can be actively worse than fewer strong ones, signaling to due diligence teams that the startup's patent counsel is running the meter rather than building defensible IP.
IV. When to File
I said we'd return to the word "pending," and here we are.
The rule is absolute: a continuation must be filed before the parent application issues as a patent, is abandoned, or has proceedings terminated. Once the parent is granted and no other family member remains pending, the continuation opportunity is gone. Not deferred. Not paused. Gone. Like a door that locks from the outside and has no handle.
The best practice, endorsed by essentially every patent strategist and IP attorney who has thought about this for more than fifteen minutes, is to file the continuation before paying the issue fee after receiving a Notice of Allowance. This maintains copendency while giving the startup maximum time to evaluate whether a continuation is warranted.
The new CAF surcharges introduce two additional timestamps that every startup should calendar immediately, in red, with reminders:
Six years after the earliest benefit date: $2,700 surcharge (large entity) kicks in. Nine years after: $4,000 surcharge.
This means: if you filed your original nonprovisional application in 2020, the six-year deadline arrives in 2026. Any continuation filed before that date avoids the surcharge entirely. This creates a strong incentive to file what we might call placeholder continuations—applications with broad initial claims that can be refined during prosecution—before the deadline passes.
The never-let-the-chain-die principle: Major technology companies maintain at least one pending application in every important patent family at all times, filing a new continuation before paying the issue fee on the last pending family member. This is described as "tried and true" by practitioners who know what they're talking about, and "something I never heard of" by founders who will later wish they had.
V. Startup Stage as Destiny
Or: How Much You Should Spend Depends on How Much You Have, Which Is Obvious, But the Specific Allocation Is Not
Pre-seed and seed stage ($5K–$20K annual IP budget): Your priority is a comprehensive provisional application that describes as many embodiments as possible. This creates the specification reservoir. Convert to a nonprovisional with modest, easily-allowed claims. Budget for at least one continuation to keep the family alive. At this stage, the specification breadth matters more than claim breadth—every dollar spent enriching the specification pays dividends for a decade of future continuations.
Series A and growth stage ($30K–$75K): This is the optimal window. File continuations targeting competitor products, broader system-level claims, and method-of-use claims. The empirical data shows patents filed at this stage correlate most strongly with higher subsequent valuations. The "Patent Lottery" study found a first patent increased five-year employment growth by 55% and sales growth by 80%.
Series B and pre-exit ($100K+): Maintain rolling continuations. File surgically targeted claims. If an M&A exit approaches, ensure all valuable patent families have pending continuations. Acquirers value the optionality. A pending continuation can be shaped to cover the buyer's product roadmap, making your IP more valuable to the specific entity most likely to buy it.
The industry differences are worth noting. In biotech, where patents essentially are the product, continuation families directly determine drug exclusivity horizons worth billions—the 2025 biopharma M&A wave saw $86 billion in top-10 deals driven by layered IP protection. In software, the landscape is different: only about 19% of software startups file patents, partly because the Alice/Mayo § 101 eligibility framework has made software patents feel like a gamble. But the USPTO's 2025 Ex Parte Desjardins decision—designated precedential—held that improvements to machine learning models constitute patentable practical applications, which has made the terrain considerably more hospitable.
VI. The Terminal Disclaimer Rule That Almost Was (And Might Yet Be)
Or: A Brief History of a Bureaucratic Near-Death Experience
I need to tell you about the worst thing that almost happened to continuation strategy, because it reveals both the fragility and the resilience of the system.
In May 2024, the USPTO published a proposed rule that would have required any patent subject to a terminal disclaimer to become unenforceable if any claim in any linked patent was found invalid on prior art grounds. One falls, all fall. Since continuations routinely trigger ODP rejections requiring terminal disclaimers, this would have transformed continuation-based portfolio building from a strategic advantage into a strategic liability. Your carefully constructed patent family would become a house of cards.
The response was extraordinary. Over 300 comments flooded the USPTO, the vast majority opposed. Five former USPTO Directors and Deputy Directors wrote a joint letter urging withdrawal. AIPLA, the U.S. Chamber of Commerce, PhRMA, and the Council for Innovation Promotion all opposed it. Only the FTC supported it, voting 3–2. The USPTO withdrew the proposed rule on December 4, 2024, citing resource constraints—though the real reasons were almost certainly the combination of overwhelming opposition, the Supreme Court's Loper Bright decision overturning Chevron deference, Director Vidal's departure, and the incoming administration's more patent-friendly posture.
Under new Director John Squires, the USPTO has signaled a distinctly warmer orientation toward patent holders. Proposed PTAB rule changes published in October 2025 would significantly limit IPR proceedings, which would increase the value of continuation patent families by reducing the risk of serial validity challenges across family members.
The net environment, as of early 2026, favors continuation strategy for startups. But three caveats: the CAF surcharges demand earlier filing discipline, the ODP case law requires careful navigation of terminal disclaimers, and the withdrawn TD rule could resurface under future leadership. Congressional interest in patent thickets has not evaporated.
The system is, as always, contingent. Which is another way of saying: it's American.
VII. What All of This Means If You're a Person Starting a Company and Trying Not to Drown
Here is what I think, for whatever a synthesized voice channeling two dead literary geniuses is worth on matters of patent prosecution strategy:
One. The single highest-leverage IP investment a startup can make is a rich, multi-embodiment initial specification. Every continuation draws from this well. A thin specification—one that describes only the product as it exists today—constrains every future option. Invest disproportionately here. This is the seed bank. You cannot go back and add seeds later.
Two. The six-year CAF deadline has created a new strategic inflection point that did not exist before 2025. File placeholder continuations before the six-year anniversary of your earliest nonprovisional filing date, even if you're not sure what claims you want. Uncertainty about future claims is not a reason to let the deadline pass. It is, in fact, the best reason to file.
Three. The risks of prosecution history estoppel across patent families are underappreciated by approximately everyone except the litigation attorneys who exploit them. Each continuation creates more prosecution history that opponents can weaponize against every patent in the family. This argues not for more continuations, reflexively, but for better continuations, strategically—and for working with patent counsel who understand the litigation implications of every amendment, every argument, every response to an office action.
Four. Never let the chain die. File a continuation before paying the parent's issue fee. This is a small investment that preserves enormous future optionality. Once the window closes, it closes forever. There is no appeals process. There is no extension. There is no "but I didn't know." The universe does not care about your ignorance. It simply enforces its rules.
The startups that treat patents as fundraising checkboxes—file one, get the stamp, move on—are leaving real protection on the table. The startups that build continuation pipelines, that maintain open families, that invest in specification depth and claim strategy over time, are building something that compounds. Not quickly. Not dramatically. But relentlessly, and in the specific way that matters most in markets where the difference between success and failure is the difference between having defensible intellectual property and having a nice story about the technology you used to own.