Webinar | Women’s Nutrition: Nourishing Womanhood From Pregnancy To Menopause
11 September | 12 PM ET: Register Now
Webinar | Women’s Nutrition: Nourishing Womanhood From Pregnancy To Menopause
11 September | 12 PM ET: Register Now
Ten institution decisions came out of the Board in August, and it denied six of them. Five of those six never got near the prior art, which means the petitions died on discretion and procedure rather than on anything the references did or didn’t say. Then the merits cohort went the other way. Petitioners took a clean or partial win in sixteen of the eighteen final written decisions that landed this month. Sixteen of eighteen.
That gap is the thing to plan around. Your references still decide the merits, but they only ever get tested if you clear institution first, and institution now runs on a screen that has almost nothing to do with how good those references are. Below is what each stage looked like in August, and where the month sits against the national numbers.
Sixty-four proceedings carry an August date in our tracker. That covers a petition filed, an institution decision issued, or a final written decision entered, and we’ve grouped every figure below by the August event rather than by when the case first started.
31 new petitions reached the Board in August, 29 of them inter partes reviews and two post-grant reviews. The filer list is short. 3 companies account for 20 of the 31 filings, which is 64.5% of the month
| Petitioner | Petitions | Share of month |
|---|---|---|
| Google LLC | 8 | 26% |
| Meta Platforms, Inc. | 6 | 19% |
| Apple Inc. | 6 | 19% |
| Zoom Communications | 3 | 10% |
| Luxottica of America | 2 | 6% |
| Seven others (one each) | 6 | 20% |
The petitions all look the same
Where the parallel cases sit
Texas runs through most of this month's data. Twenty-one of the thirty-one August filings name a Texas district-court venue, and all seven IPR institution decisions recorded one, with six of those listing a Texas district as the sole venue in the Western and Eastern Districts. These are the fast-docket courts that generate the early trial dates the Board weighs when it applies its discretionary framework. So the trial date next door is doing real work in your institution decision.
August filers won on the merits, but they had to clear the discretionary gate before any of that counted. If your parallel case sits in a fast-trial venue, your petition timing and your Fintiv record matter as much as your best reference.
Twenty-seven of thirty-one filers ran multi-petition campaigns. The USPTO has said parallel petitions on the same claims should be rare, so a stack you haven't justified can draw a denial on structure alone, before anyone looks at the art.
One note on the data. We have complete technology classification for only nine of the thirty-one new filings, so there's no full sector split this month and we're reporting only the subject matter we can verify. Of that nine, the work clusters in networking and computer architecture, Tech Centers 2400 and 2100. Meta is challenging solid-state storage and sensor patents, while Google sits in Tech Center 2400 network technology.
| Denial basis | Count | Cases |
|---|---|---|
| Discretionary (Fintiv, parallel case) | 3 | Microsoft v. ToutVirtual x3 |
| Discretionary (general, 314(a)) | 2 | Biocon v. Regeneron, Viant v. Intent IQ |
| Procedural (RPI defect) | 1 | Viance v. Koppers |
| Instituted, merits proceed | 4 | Altria, Cisco, Samsung x2 |
Five of the six denials never touched the strength of the challenge at all. They turned on discretion or procedure, which matches the wider Board picture. The USPTO’s March 2025 interim process split institution decisions into two parts, one discretionary and one on the merits, and August is a clean look at what the discretionary half is doing.
The Fintiv trio, Microsoft v. ToutVirtual
The nine-month trap, Viance v. Koppers
The Board found a real-party-in-interest defect in the Viance petition. Most of the time you amend and move on. Here the petitioner couldn’t, because adding the party would have reset the filing date and pushed the petition outside the nine-month window that governs post-grant review. A defect that looked curable turned fatal on the statutory clock. Get the RPI list right the first time or lose the forum.
| Outcome | Count | Share |
|---|---|---|
| Petitioner, all claims down | 10 | 56% |
| Mixed, some claims down | 6 | 33% |
| Patent owner, all claims survive | 2 | 11% |
Sixteen of the eighteen August FWDs invalidated at least some claims. Two left every challenged claim standing.
The two patents that survived
The median August final written decision arrived 579 days after the petition was filed, a little over nineteen months. Three of them are outliers. Samsung’s matters against Manufacturing Resources International had been sitting since 2022, and taking them out moves the median to 573 days across the remaining fifteen, inside a band of 320 to 735 days. The fastest was Amazon Web Services against Ziklag IP, at 320 days. Those Samsung cases show what stays, appeals and remands can add to a timeline, with each one running past 1,350 days from filing to decision.
This is where the multi-petition pattern lands. Ipsen Biopharm defended the same patent, number 11,344,552, against CSPC Pharmaceutical and Apotex separately, and both proceedings ended mixed. So a patent owner in that position fights the same validity case twice, and the second challenger gets to file with the first one’s record already in hand. That’s an advantage worth watching for.
A patent that reaches a final written decision at the Board has a better than even chance of losing every challenged claim, and a nine-in-ten chance of losing at least some of them. That belongs in your licensing model, your acquisition screen, and any damages position built on an asserted patent.
The offset is the gate in front of it. Six of ten institution decisions this month never got past discretion, so fewer challenges reach the stage where that risk applies at all.
Line the three cohorts up and August reads like two separate contests with two different winners.
| Stage | August result | Who it favors |
|---|---|---|
| Institution | 60% denied, mostly on discretion | Patent owner |
| Merits (FWD) | 89% lose claims, 56% lose all | Petitioner |
| Post-grant channel | H1 2026 (national) | Share of post-grant filings |
|---|---|---|
| Ex parte reexam | Now the dominant channel | 74.7% (594 requests) |
| Inter partes review | Q2 filings hit an all-time low of 57 | 21.9% (174 petitions) |
| Post-grant review | Smallest channel by volume | 3.4% (27 petitions) |
National figures from public post-grant data (Unified Patents), first half of 2026. Not from the August tracker.
1. Don't treat a Texas parallel case as an automatic bar. Three of the four petitions instituted this month ran alongside a Texas district court case, which is the same profile that produced the Fintiv denials. It isn't fatal. What decides it is how you argue the discretionary factors and how carefully you build the trial-date and investment record, and defendants who write the petition off early are leaving instituted petitions on the table.
2. Age protects a patent at the gate and nowhere else. The settled-expectations factor keeps challenges to older patents out at institution, and then does nothing at all once a patent is inside. Look at Ziklag's U.S. Patent No. 6,182,128. It reached FWD in two separate proceedings in August and lost every challenged claim in both.
3. Check who filed before you. Two patents drew challenges from different petitioners this month. Ziklag's patent faced Apple and Amazon Web Services, and Ipsen's 11,344,552 faced CSPC and Apotex, with the later challenger in each case filing with the earlier record already in hand and sharpening the attack on it. If you're a co-defendant, that's an argument for coordinating early and for watching your estoppel dates.
4. A discretionary denial isn't a clean bill of health. Intent IQ won its discretionary denial and the patent survived, but the Board never weighed the prior art. Nothing was decided about validity. If you're valuing that patent or licensing it, the question is still wide open.
5. Justify the stack before you file it. Twenty-seven of the thirty-one August filings were multi-petition, every petition carried expert support, and a small group of firms ran most of the docket. The USPTO has said parallel petitions on the same claims should be rare. File three at once without explaining why each one is needed and you invite a denial on structure alone, before anyone reads the art.
