Should You Challenge a Patent at the PTAB?

This guide assumes you already know roughly what the Board does. For the underlying doctrine, the statutory grounds, the IPR-versus-PGR line, and the case archive, see our concept page on inter partes review at the PTAB. What follows is the decision itself: how to run the trade-off when a patent is pointed at you.

What is the first clock you need to stop worrying about missing?

Everything else is negotiable. This is not.

Under 35 U.S.C. § 315(b), an IPR “may not be instituted if the petition … is filed more than 1 year after the date on which the petitioner, real party in interest, or privy of the petitioner is served with a complaint alleging infringement of the patent.” One year from service, not from filing, not from when your litigation counsel got up to speed.

Three practical consequences:

1. The real deadline is much earlier than one year. A petition that has a chance needs an independent prior-art search, an expert declaration, claim charts for every ground, and a 14,000-word argument that survives a hostile reading. Petitioners who start at month ten file bad petitions. Work usually begins within two or three months of service.
2. The bar runs against people who are not you. If a supplier, a parent, an indemnitor, or a customer was served earlier and counts as your real party in interest or privy, their clock is your clock. Map that relationship before you assume you have time.
3. There is no appellate rescue. In Thryv, Inc. v. Click-To-Call Technologies, LP (2020), the Supreme Court held that a Board determination applying the § 315(b) time bar is part of the nonappealable institution decision. Get the date wrong and the Federal Circuit will not fix it.

Put a calendared date on this in week one of the case, alongside the answer deadline.

What are you actually choosing between?

“IPR or district court” is a false binary. The real menu, roughly in order of cost:

• Do nothing yet. Test non-infringement first. If the accused product plainly falls outside the claims, validity is a fight you may never need. Invalidity is the expensive defense; non-infringement is usually the cheap one.
• Design around. Sometimes the cheapest invalidity strategy is not needing one.
• License or settle. If the demand is smaller than a petition, the math is over before it starts. See what patent trolls are and how to respond.
• Ex parte reexamination. Far cheaper than IPR, also limited to patents and printed publications, and it carries no § 315(e) estoppel of its own. The trade-off is that you hand the art to the examiner and then sit down: the requester has essentially no participation rights once the proceeding starts, and no ability to rebut the patent owner’s amendments. It is a blunt instrument that has become noticeably more popular as IPR institution has gotten harder.
• A Rule 12 § 101 motion. For software and business-method patents, an eligibility challenge under Alice can end a case in months for a fraction of an IPR, and it is unavailable at the PTAB in an IPR.
• IPR. The subject of this guide.
• PGR, if and only if the patent issued within the last nine months. Broader grounds, tiny window, and correspondingly broader estoppel.
• Full district-court invalidity defense. The most expensive path, and the only one that can hear every theory at once. Budget accordingly: see patent litigation cost and timeline.

Most sophisticated defendants pick two or three of these and sequence them.

Does your best invalidity argument even fit inside an IPR?

Run this test before anything else, because it disqualifies more cases than cost does.

An IPR can hear only novelty under § 102 and obviousness under § 103, and only on the basis of patents and printed publications. So:

If your case lives in the bottom half of that table, an IPR is not your forum, and the good news is that estoppel will not touch those theories either. If it lives in the top half, keep reading.

A subtler version of the same question: how good is the art, really? The Board institutes only on a reasonable likelihood that the petitioner prevails on at least one claim, and institution is all or nothing. SAS Institute Inc. v. Iancu (2018) held that if the Board institutes it must decide every claim the petition challenged, and USPTO guidance issued two days later extended the same rule to every ground raised. So the Board takes all of it or none of it. That cuts both ways. You cannot get a clean institution on your two best grounds and quietly shed the weak ones; a petition padded with filler grounds drags your strong ones through the same trial and hands the patent owner easy wins to point at.

Will the Board even take the case in 2026?

This is where the calculus has changed most, and where advice written before 2025 is actively misleading.

The USPTO rescinded the Vidal-era interim discretionary-denial memorandum in February 2025 and restored pre-2022 practice under Fintiv and Sotera on March 24, 2025. Two days later, a March 26, 2025 memorandum on PTAB workload management bifurcated institution: the Director decides discretionary questions first, and only if the case survives does a panel reach the merits. Under the current process, the patent owner files a discretionary-denial brief within two months of the Notice of Filing Date Accorded and the petitioner opposes within three months, each limited to 20 pages since September 1, 2025. No reply without good cause.

Director Squires has layered on more. An October 17, 2025 memorandum centralized institution determinations in the Director, with routine outcomes issued as summary notices rather than reasoned decisions. A November 17, 2025 memorandum made voluntary disclosure of the petitioner’s prior-art search methodology a factor weighing in favor of institution. And a March 11, 2026 memorandum directs parties to brief U.S. manufacturing footprint and small-business status: whether the accused products are made in the United States, whether the patent owner’s competing products are, and whether the petitioner is a small business that was sued.

The effect on the numbers is not subtle. Institution rates that historically ran near 70% fell to roughly a third of petitions decided in late 2025, driven almost entirely by discretionary denials rather than merits losses.

What this means for your decision:

• Budget for a denial. A meaningful share of the money now goes to a proceeding that never gets instituted. Model the expected value, not the best case.
• “Settled expectations” is a real risk for older patents. The Director has denied petitions on the ground that a patent long in force carries settled expectations for its owner, using the six-year damages window of 35 U.S.C. § 286 as a reference point. Age is not dispositive, and lack of notice, non-overlapping markets, or clear examiner error can rebut it, but if you are challenging a patent that issued eight years ago you need an answer ready.
• A Sotera stipulation helps, but does not buy institution. Stipulating that you will not pursue in district court any ground you raised or reasonably could have raised weighs against a Fintiv denial. It no longer functions as the safe harbor it was in 2022, and it costs you: you are volunteering estoppel before you know whether you will get a trial.
• Check § 325(d) too. If your references were already before the examiner during prosecution, the Board can turn the petition away on that basis alone. Art that is genuinely new to the Office is worth far more than art that is merely good.
• Discretion is unreviewable. Cuozzo Speed Technologies, LLC v. Lee (2016) put the institution decision outside Federal Circuit review. There is no appeal from a denial, only the fees you already spent.

Because these memoranda have been reissued and rescinded repeatedly since 2022, treat any specific statement about discretionary-denial policy, including this one, as perishable. Confirm current USPTO guidance before you rely on it.

What does estoppel actually cost you?

This is the trade the whole decision turns on. Under § 315(e), once the Board issues a final written decision, you (and your real parties in interest and privies) may not assert in the USPTO, a district court, or the ITC any ground you “raised or reasonably could have raised” in the IPR.

“Reasonably could have raised” is the dangerous phrase. It reaches prior art a skilled searcher would have found, not just what you actually cited. So a losing IPR does not just fail; it retroactively deletes your paper-art defense from the lawsuit you are still in.

Two things soften it, and you should understand both precisely:

Estoppel is about grounds, not references. In Ingenico Inc. v. IOENGINE, LLC, 136 F.4th 1354 (Fed. Cir. 2025), the Federal Circuit construed “ground” for the first time and held it means the invalidity theory available under §§ 102 and 103, not the evidence supporting it. A theory that could not have been raised in an IPR is not estopped, even if the documents proving it overlap with your IPR art. So prior public use, on-sale activity, and system or device art survive, and patents and printed publications can still be used in court as evidence of what a prior system was or did. This resolved a long district-court split in petitioners’ favor and materially lowers the cost of a lost IPR for defendants with product prior art.

Estoppel attaches only on a final written decision. A denied institution triggers no § 315(e) estoppel. A settlement before final written decision generally triggers none either. The bar lands when the Board rules.

Practical rule: front-load your best references. You cannot hold a killer reference back for the jury. If your prior art portfolio is deep, the petition should take the deepest of it, not the cleanest-looking of it.

Does the math work?

Two buckets.

Government fees. The USPTO charges an IPR request fee plus a post-institution fee, together landing in the tens of thousands of dollars for a standard petition, with per-claim surcharges above twenty claims. These have been adjusted repeatedly, so price the current USPTO fee schedule rather than a number in an article. Post-institution fees are generally refundable if the Board declines to institute.

Legal and expert fees. This is the real number. All-in cost through a final written decision commonly runs from a few hundred thousand dollars to around $1 million, depending on claim count, technology, and how hard the patent owner fights. AIPLA’s periodic economic surveys are the standard public reference point and their reported medians have historically sat in the mid-hundreds of thousands.

Time. Roughly six months from petition to an institution decision, then a statutory one year from institution to final written decision, extendable six months for good cause. Call it 18 months to a merits ruling, plus 12 to 18 months if the loser appeals to the Federal Circuit.

Compare that against a district-court validity fight that runs two to three years and frequently costs several million per side, and IPR looks cheap. But compare it against the actual demand on the table, which for many assertions is well under $500,000, and it often looks absurd. The correct comparison is never “IPR versus full trial.” It is “IPR versus the price of making this go away.”

Will you get a stay, and how much is that worth?

A stay of the district-court case is frequently the real prize, because it stops the meter on discovery, which is where litigation money actually goes. Courts generally weigh whether discovery is complete and a trial date set, whether a stay simplifies the issues, and whether it unduly prejudices the patent owner.

Two asymmetries worth planning around:

• File early and the stay case is much stronger. A petition filed at month three of the lawsuit, before a Markman hearing and before fact discovery, is a far better stay motion than one filed at month eleven. The same filing date that protects you from the § 315(b) bar also buys the stay.
• Filing early also weakens the patent owner’s Fintiv argument, since there is no imminent trial date to point at. The incentives all run in one direction: decide fast.

If the accused product is imported, note the mirror image: the ITC does not stay for IPRs, so a Section 337 case will run to conclusion on its own schedule regardless.

Which traps catch petitioners who did everything else right?

• Naming the real parties in interest. Get this wrong and the petition can be denied or terminated, sometimes after the § 315(b) bar has run, leaving you with nothing. The Federal Circuit’s Applications in Internet Time, LLC v. RPX Corp. (2018) applies a broad, fact-driven test that looks at who funds, directs, and benefits from the petition. If a trade association, an indemnitor, or a defense group is paying, disclose it and disclose it accurately.
• Joinder as a rescue, and its limits. Section 315(c) lets the Director join a properly-filed petitioner to an instituted IPR, and § 315(b)‘s time bar does not apply to a joinder request. That is a genuine escape hatch for a late-served defendant. But Facebook, Inc. v. Windy City Innovations, LLC (Fed. Cir. 2020) held that a petitioner cannot join itself to its own instituted IPR to add new claims or issues. Joinder gets you into someone else’s case on their grounds, not a second bite at your own.
• Group petitions. Joint petitions spread cost but bind every named petitioner to § 315(e) estoppel. Make sure everyone in the group understands they are trading away their own paper-art defense.
• Copying the district-court invalidity contentions. Contentions are written to preserve options. A petition must pick and win. The two documents have opposite drafting logic, and a petition drafted like contentions gets denied.
• Assuming a stay is automatic. It is not, and a denied stay plus a pending IPR means paying for both forums at once.

How does an IPR change settlement?

Filing a petition is a pricing event, not just a legal one.

• Institution moves the number more than filing does. A filed petition is an argument. An instituted petition is the Office saying you are reasonably likely to win on at least one claim, and patent owners re-rate their case accordingly. (Note the change in who says it: since the October 2025 memorandum the Director makes that call personally, and routine institutions issue as summary notices rather than a reasoned panel decision.) The window between institution and final written decision is where most IPR-driven settlements happen.
• Cancellation is in rem and permanent. If the Board cancels the claims, they are gone against everyone, not just you. A patent owner asserting one patent against thirty defendants is not risking one case; it is risking the asset. That is why a single credible petition can resolve a whole campaign.
• Settlement before final written decision usually avoids estoppel, which makes an early settlement genuinely cheaper for you than a late one. Parties can file a joint motion to terminate under § 317, though the Board may proceed to a decision anyway if it has already decided the merits.
• The threat has a shelf life. Once the § 315(b) year runs, your leverage evaporates and the patent owner knows the exact date. Expect settlement posture to harden after it passes.

The reverse view, from the patent owner’s side of the same table, is in what to do when someone is infringing your patent.

When should you not file?

Say no if any of these are true:

• The art is merely good. Institution now requires surviving a discretionary gate before anyone reads your merits. Marginal art plus a hostile discretionary posture is money spent to teach the patent owner your case.
• The dispute is worth less than the petition. Obvious, routinely ignored.
• A trial is already scheduled ahead of the Board’s deadline and the court has invested heavily. Fintiv was built for exactly that fact pattern.
• Your references were before the examiner. Section 325(d) exists to turn those away.
• The patent is old and quiet. Settled expectations cut against you, and you will spend real briefing pages fighting a factor that has nothing to do with whether the patent is valid.
• Your best defense is non-infringement. Do not spend $700,000 proving a patent is invalid that you do not practice anyway.
• You cannot afford to lose the paper-art defense. If the district-court case is existential and your prior art is your only real defense, putting it in front of a preponderance standard is attractive, but understand you are wagering the defense itself, not merely the filing fee.

Say yes when the art is strong and genuinely new to the Office, the exposure is large, you are early in the lawsuit, and a stay is realistically available. That combination is what IPR was built for, and in that posture it remains the single most effective defensive move in U.S. patent practice.

The bottom line

Treat this as a portfolio decision made under a hard deadline, not a legal question with a right answer. The § 315(b) one-year bar from service forecloses the option entirely and is unreviewable, so the analysis has to happen in the first weeks of a case, not the last. Say yes when your invalidity case is built from patents and printed publications, the art is strong and was never before the examiner, the exposure justifies roughly $500,000 to $1 million over 18 months, and an early filing buys you a stay. Say no when your real defenses are § 101, § 112, prior use, or system art, because an IPR cannot hear them and, after Ingenico, estoppel will not touch them either. And price the downside honestly: with institution rates near a third and discretion centralized in the Director, a great deal of IPR spending now buys a denial, while a loss on the merits costs you the very defense you were counting on.

Frequently asked questions

Should I file an IPR or fight validity in district court?

It depends on what your best invalidity argument is made of. If your case rests on prior-art patents and printed publications, IPR gives you a lower burden of proof (preponderance instead of clear and convincing evidence), technically trained judges instead of a jury, a statutory decision deadline of about eighteen months, and a decent chance at a stay of the lawsuit. If your best arguments are § 101 eligibility, § 112 defects, a prior public use, or a prior sale, an IPR cannot hear them at all and district court is where they live. Many defendants do both, filing an IPR on the paper art while preserving the rest for court.

How long do I have to decide whether to file an IPR after being sued?

One year from service of the complaint, under 35 U.S.C. § 315(b), and the clock runs against your real parties in interest and privies too, not just the named defendant. Because a competent petition needs a prior-art search, an expert declaration, and claim-by-claim mapping, the practical deadline for making the decision is far earlier: most petitioners start work within a few months of service. The Supreme Court held in Thryv v. Click-To-Call Technologies (2020) that the Board’s time-bar rulings are not appealable, so there is no second look if you get the date wrong.

What does IPR estoppel actually stop me from arguing later?

Under 35 U.S.C. § 315(e), a final written decision bars you, your real parties in interest, and your privies from asserting in the USPTO, a district court, or the ITC any ground you raised or reasonably could have raised in the IPR. In Ingenico Inc. v. IOENGINE, LLC, 136 F.4th 1354 (Fed. Cir. 2025), the Federal Circuit read “ground” as the invalidity theory rather than the individual reference, so estoppel does not reach theories an IPR could never have heard: prior public use, on-sale activity, system or device art, § 101, and § 112. What it does forfeit is your paper-art case, which for most defendants is the strongest one they have.

When is filing an IPR a bad idea?

When the prior art is mediocre and estoppel would cost more than a denial saves; when the patent has been in force long enough that the Director may find settled expectations favor the owner; when a parallel trial is already scheduled ahead of the Board’s deadline and Fintiv denial looks likely; when your references were already considered during prosecution, inviting a § 325(d) denial; when the amount actually in dispute is smaller than a petition’s all-in cost; and when you would rather buy a cheap license than spend a year proving a point. A denial leaves you with legal fees, no stay, and a patent owner who now knows your best art.

This guide is general education, not legal advice, and does not create an attorney-client relationship. PTAB strategy, estoppel, discretionary-denial policy, and filing deadlines turn on the specific facts and posture of your dispute, and USPTO guidance in this area has changed repeatedly. Consult a patent litigator licensed in your jurisdiction before acting.