The MV2 Purge in Numbers: 44,158 Extensions, 92 Million Users, and What Edge Faces Next

On October 8, 2025 the Chrome Web Store listed 176,593 extensions. Two days later it listed 131,703. Nothing about that drop was announced, and almost nothing has been written about its size.

That was eleven months ago. This is not breaking news, it is the accounting nobody published at the time. We crawl the Chrome Web Store, Microsoft Edge Add-ons and Firefox Add-ons every day, so we can say exactly what left, who owned it, how long it had been dead, and what came back. The reason it still matters in September 2026 is sitting one browser over. Microsoft Edge has 3,339 Manifest V2 extensions live right now, and it started its own consumer phase-out last month.

Key takeaways

Chrome removed 44,158 Manifest V2 listings over 48 hours. They held roughly 92.3 million users and made up 25.4% of the catalog.

Most of what went was already abandoned. 92.5% had not shipped an update since 2021 or earlier, and ten listings accounted for 38 million of the lost users.

Google’s published timeline still dates this event to August 31, 2026. Our crawl recorded zero MV2 deletions that month, because there was nothing left to delete.

Methodology note. Proprietary figures come from exstats.com, with Chrome and Edge queried on September 5, 2026. Deletion dates are our observation dates, not Google’s action dates. Full limits are stated in the Methodology section at the end.

What Chrome removed on October 9–10, 2025

Manifest V2 is the older extension platform version, which allowed blocking `webRequest` interception and remotely hosted code. Manifest V3 is its replacement, which restricts both, and is the version Chrome now requires.

Across October 9 and 10, 2025, 44,158 listings with a manifest_version of 2 acquired a deletion date. Between them they carried about 92.3 million users. The removal was lopsided across the two days: 9,497 listings on the 9th, 34,661 on the 10th.

Daily numbers make the catalog effect unmistakable. Chrome had been adding roughly 150 net listings a day through early October. Then it lost 44,890 in 48 hours.

It was permanent. Across the whole store, October 2025 recorded 47,288 delistings against 460 restorations, a ratio of about 103 to 1. Ordinary Chrome churn in the surrounding weeks ran 20 to 100 delistings a day with a handful restored. This was not a crawl artifact or a temporary visibility change. The listings went and stayed gone.

It was also the end of the line for MV2 on Chrome. We have recorded no Manifest V2 deletion since October 2025 and no live MV2 listing in the store at any point since, out of 254,621 live listings today.

Google’s own timeline still dates this to August 31, 2026

This is the part that keeps tripping people up. Google’s Manifest V2 support timeline lists a milestone reading "Aug 31st 2026: All remaining Manifest V2 extensions removed from the Chrome Web Store" Chrome for Developers. That date is still the one most coverage cites.

It did not happen in August 2026. We recorded 4,432 Chrome deletions that month, and every single one was a Manifest V3 listing from an unrelated enforcement sweep. There were no MV2 extensions left to remove. The store had been cleared ten and a half months earlier.

The same timeline explains what came first. MV2 extensions were disabled for every user on every channel on July 24, 2025 with Chrome 138, and the enterprise ExtensionManifestV2Availability escape hatch was removed in Chrome 139. That leaves an eleven-week window worth noting. Between late July and October 9, 2025, tens of thousands of MV2 extensions were still listed and still browsable in the store, and completely non-functional in the browser for everyone who already had them. The store caught up with the browser in October. The calendar caught up with both a year later.

Most of what died was already dead

An abandoned listing is one that is still published in the store but has not shipped an update in over a year; the staleness figures below use the last recorded update date on each listing.

The instinct on seeing a quarter of a catalog vanish is to picture a functioning ecosystem being demolished. The staleness profile says otherwise. Of the 44,158 removed listings, 40,838 (92.5%) had not shipped an update since 2021 or earlier. Only 160 had been updated at any point during 2025. Every removed listing had a recorded update date, so that denominator is the full set, not a sample.

The size distribution tells the same story from another angle. 6,808 removed listings had zero users. Another 31,254, or 70.8% of everything removed, had fewer than 100.

Sorted by the year of each listing’s last update, the shape of the catalog Google swept becomes obvious. Most of it was abandoned between 2019 and 2021, by developers who had walked away years before Manifest V2 stopped working.

The ten extensions that carried the damage

Read that last table right-to-left and a second event appears inside the first. The 160 listings updated during 2025 held 21.4 million users between them, more than any other cohort in the table. That is not evidence that fresh extensions were the biggest. It is one extension wearing a cohort’s clothes: uBlock Origin alone accounts for 19 million of the 21.4 million.

That is the real shape of the purge. Ten listings carried 38 million of the 92.3 million lost users. 41% of the total user impact sat in 0.02% of the removed listings. Widen it to the 121 listings with 100,000 users or more and you capture 62.6 million users, or 68%.

uBlock Origin deserves its own paragraph, because it breaks the abandonment narrative cleanly. It was no zombie. Its last update landed on September 16, 2025, three weeks before it was removed, and it held a 4.7 rating across 35,407 ratings. It was removed for running on Manifest V2 and using the blocking webRequest API that MV3 replaced, not for being neglected.

Where the losses landed by category

By listing count, the damage was concentrated where the catalog itself is concentrated: productivity tooling. The workflow subcategory alone lost 17,235 listings, 39% of everything removed.

By users, one category behaves completely differently. Chrome’s privacy category lost 20 listings and 22.1 million users, 24% of the entire user impact. Two extensions produce almost all of it: uBlock Origin at 19 million and Touch VPN at 3 million. If you want the content-blocker story compressed into a single figure, that is it. Eleven months on, the categories that lost the most listings are not the ones with the most room left, which we mapped separately in the category opportunity map.

The permission data connects the categories back to the API change that caused all of this. Of the 44,158 removed listings, 2,459 declared the webRequest permission and 1,871 declared webRequestBlocking, the blocking network interception that Manifest V3 replaced with declarativeNetRequest. That is only 4.2% of what was removed. Most of these extensions were not blocked because of what they could do to network traffic. They were removed because nobody had touched them in five years and the manifest they shipped with had expired.

The developers who never came back

The most useful question about a mass removal is whether the people behind it stayed. We can answer it, with one honest limit.

26,315 of the removed listings expose a developer email address, and those map to 21,257 distinct addresses. Joining them against every live Chrome listing today: 17,951 of those 21,257 addresses (84.4%) have no live Chrome listing under the same email. The other 3,306 do.

The join is reliable in one direction. Only 239 of 254,621 live Chrome listings hide the developer email, which is 0.1%, so a developer who republished under the same address would almost certainly show up. What the method cannot see is someone who came back under a different email, a new company account, or a different address entirely. So read the figure precisely: 84.4% of affected developers are not visible in the store today under the identity they published with. That is not the same as proving they never returned, and we are not going to write it as though it were.

The remaining 17,843 removed listings expose no developer email at all and sit outside this count entirely.

The catalog recovered in five months

Chrome regained its pre-purge listing count on March 31, 2026, at 178,299 against the 176,593 it had held on October 8, 2025. It has not slowed since. The store now lists 254,652 extensions, 44% above its pre-purge peak.

The user side is the interesting half, and it needs a caveat before a conclusion. Chrome publishes user counts in rounded buckets, so the store-wide total is a sum of approximations and moves by tens of millions on its own. Across the eleven months since the purge, the monthly average has oscillated between roughly 1.97 and 2.16 billion, the same band it occupied in the months before. Removing 92.3 million users’ worth of listings, about 4.5% of the aggregate, is not distinguishable from that noise. On October 8 the store total read 2.03 billion. On October 10, with 44,158 listings gone, it read 2.07 billion.

So the honest summary is narrower than the headline invites. The catalog grew 44%, and the store’s aggregate user count did not visibly respond to either the purge or the recovery. A catalog that is 44% larger is not reaching 44% more people. We found the same pattern in the Q2 2026 cross-store report, where the median Chrome extension had 14 users. Our Q1 2026 catalog data, taken mid-recovery, caught the store at 178,299 listings, the exact point at which it drew level with where it had been before the purge.

Edge has not done this yet

Microsoft Edge Add-ons currently lists 32,478 extensions. 3,339 of them are Manifest V2, 10.3% of the catalog, and they hold 85,557,455 active installs.

Microsoft announced the phase-out on August 7, 2026: "Beginning in August 2026, Microsoft Edge will start the consumer transition away from Manifest Version 2 (MV2) extensions and toward MV3. Our goal is to complete the consumer transition by the end of 2026," with enterprise deprecation "following in early 2027" Microsoft Edge Blog. The rollout starts in Canary, Dev and Beta and broadens to Stable. Managed devices are excluded during the consumer phase.

The Edge population has the same barbell shape Chrome’s did. Only 24 of the 3,339 listings were updated at any point in 2026, but those 24 hold 41.9 million installs, 49% of the exposure. At the other end, 2,432 listings have fewer than 1,000 installs each and 1,252 have fewer than 100.

Thirteen Edge extensions with a million or more installs carry 63.4 million between them, 74% of the total exposure. Two of the three largest are content blockers that were updated within the last two weeks.

Microsoft frames the scale of this very differently. Its announcement says "there are only 58 MV2-based extensions in the Edge Add-ons website with any meaningful usage," of which "only three don’t have a publicly available MV3-based extension already available."

Both numbers are true, and the gap between them is the most useful thing in this dataset. We count 52 live MV2 listings with 100,000 or more installs, and they hold 77.6 million of the 85.6 million exposed installs, or 90.7%. That is close enough to Microsoft’s 58 to suggest their "meaningful usage" threshold sits somewhere around the 100,000 mark, and they are right that the head of the distribution is almost entirely ready to migrate.

The other 3,287 listings hold 7.9 million installs between them. Microsoft’s number describes migration readiness. Ours describes user impact. Nobody is going to port an extension last touched in 2021 with 400 installs, so 7.9 million installations are going to stop working with no MV3 successor and no announcement naming them. Chrome’s October 2025 numbers are the closest thing to a forecast Edge publishers have. There, the equivalent tail was 31,254 listings under 100 users, and it went quietly.

What this means now

Four things the numbers actually support.

• The purge was mostly cleanup. 92.5% of what Chrome removed had been abandoned since 2021 or earlier, and 70.8% had fewer than 100 users. "Google deleted a quarter of the Chrome Web Store" is true and misleading in the same breath.
• The user impact was real but concentrated. Roughly 120 large extensions carried two-thirds of the 92.3 million affected users, and one of them, uBlock Origin, was actively maintained right up to the day it went.
• Store size is not store health. Chrome is 44% larger than it was before the purge and its aggregate user count has not moved outside its normal band. Growth in listings is not growth in reach.
• Edge is heading into the same event with a similar distribution and a public deadline. The head is ready. The tail, 3,287 listings and 7.9 million installs, is not.

If you publish on Edge and you have not checked your own manifest version, the end-2026 consumer deadline is now inside a single quarter. If you compete with someone who publishes on Edge, their manifest version and update cadence are both public, and worth watching between now and December.

Methodology

All extension-level and store-level figures come from our own daily crawl of the Chrome Web Store, Microsoft Edge Add-ons and Firefox Add-ons via exstats.com. Chrome and Edge figures were queried on September 5, 2026. Manifest version is read from each listing’s published manifest JSON, not inferred.

Four limits worth stating plainly. First, our deletion field records the date we observed a listing gone, not the date Google acted. The 9,497 / 34,661 split across October 9 and 10 may reflect our crawl cadence rather than two separate Google actions, so treat the two-day total as solid and the daily split as approximate. Second, our Chrome crawl begins on January 1, 2025, so any MV2 listing removed before that date is invisible to us. 44,158 is the October 9–11 window. 50,505 is every MV2 deletion we have observed since January 2025, and that is a floor rather than a total. Third, Chrome publishes user counts in rounded buckets, so "19 million" means a bucket, not a measurement, and store-wide user sums carry that rounding at scale. Fourth, Edge reports active installs while Chrome reports cumulative users. The two are not the same metric and are never added together here.

One genuine gap: Firefox. The manifest field is null for all 100,353 live Firefox listings in our data, so we cannot produce an MV2/MV3 split for Firefox and have left it out of this analysis rather than estimate it. Mozilla continues to support both manifest versions, so there is no equivalent removal event to count, but we cannot yet size the population.

External dates and quotations are cited inline to their primary sources: the Chrome for Developers Manifest V2 support timeline and the Microsoft Edge developer blog post of August 7, 2026.