Browser Market Share Statistics 2026
By Axis Intelligence Research
Co-author: Alex Rivera (Consumer Technology Analyst) | Last updated: August 12, 2026 | License: CC BY 4.0
Chrome held 68.22% of worldwide browser sessions in July 2026 — 39.10 percentage points more than its five named rivals combined. Axis Intelligence Research calculates that the worldwide browser market scores 4,973 on our Browser Concentration Index, 2.76 times the level U.S. antitrust agencies treat as a highly concentrated market.
Quick Answer
According to Statcounter Global Stats, Chrome held 68.22% of worldwide browser sessions across all device types in July 2026, ahead of Safari at 16.47% and Edge at 5.37%. Axis Intelligence Research finds the worldwide Browser Concentration Index (BCI-B) reads 4,973 for that month, and that Blink-based browsers account for 77.53% of all named sessions. Regional variation is wider than the global figure suggests: Chrome ranges from 54.01% in North America to 77.43% in Africa.
Key Findings
- According to Statcounter Global Stats, Chrome held 68.22% of worldwide browser sessions in July 2026, with Safari second at 16.47%.
- Axis Intelligence Research finds Chrome’s worldwide share exceeds the combined share of its five named rivals by 39.10 percentage points (68.22% versus 29.12%).
- Axis Intelligence Research calculates a worldwide Browser Concentration Index (BCI-B) of 4,973 for July 2026, against the 1,800-point threshold the 2023 U.S. Merger Guidelines use for a highly concentrated market.
- Axis Intelligence Research finds Africa is the most concentrated regional browser market measured, at BCI-B 6,110, and North America the least, at 3,845 — a spread of 2,265 index points.
- Axis Intelligence Research finds Safari carries a Platform Dependence Score of 0.830 against Chrome’s 0.315, meaning Safari’s share swings across device classes nearly three times as sharply.
What Is the Browser Market Share in 2026?
The July 2026 worldwide table is short, and its shortness is the finding. Six browsers account for 97.34% of measured sessions. Everything else on earth — Brave, Vivaldi, DuckDuckGo, Yandex, Coc Coc, Comet, Atlas, the entire long tail of independent and AI-first browsers — divides the remaining 2.66%.
| Browser | Worldwide share | Engine | Source |
|---|---|---|---|
| Chrome | 68.22% | Blink | Statcounter, July 2026 |
| Safari | 16.47% | WebKit | Statcounter, July 2026 |
| Edge | 5.37% | Blink | Statcounter, July 2026 |
| Firefox | 3.34% | Gecko | Statcounter, July 2026 |
| Samsung Internet | 2.06% | Blink | Statcounter, July 2026 |
| Opera | 1.88% | Blink | Statcounter, July 2026 |
| Other / unreported | 2.66% | mixed | Axis Intelligence Research, from Statcounter |
The 2.66% residual matters more than it looks. Statcounter publishes six rows per panel; anything below sixth place is not zero, it is unreported. Any analysis that treats the residual as empty will overstate every share in the table, and any analysis that assigns it to a named browser is inventing data. We carry it as its own row in the dataset and exclude it from every index, which makes our concentration figures a floor rather than an estimate.
Alex Rivera: The interesting number in a consumer-tech market is usually the attach rate — how often a second product rides in on the first. Browsers invert it. Nobody attaches a browser; a browser attaches you, at setup, in the thirty seconds before you have an opinion. Chrome’s 68.22% is not a preference share, it is a default-placement share with a preference share hiding somewhere inside it, and no panel on the open web can separate the two.
How Concentrated Is the Browser Market? The Axis Browser Concentration Index (BCI)
Everyone publishes browser share. Nobody publishes what that share does to market structure. The Browser Concentration Index (BCI) is an Axis Intelligence Research metric that scores a browser market on the same 0–10,000 scale antitrust economists use, so a browser figure can be set beside a regulatory benchmark instead of floating free.
BCI methodology and weights
BCI takes the Herfindahl-Hirschman form: square each competitor’s percentage share, then sum. It ships in two readings.
- BCI-B (browser level): the sum of squares of every browser Statcounter reports for a panel.
BCI-B = Σ(browser share²) - BCI-E (engine level): the same operation after collapsing browsers to rendering engines.
BCI-E = Blink² + WebKit² + Gecko²
Weights are equal and explicit — each competitor is weighted by its own share, which is what squaring does. The unreported residual is excluded rather than distributed, so both readings understate true concentration. Every input is a row in the accompanying dataset, and any competent outsider can recompute the readings from those rows.
For reference, the 2023 Merger Guidelines issued by the U.S. Department of Justice and Federal Trade Commission treat a post-merger HHI above 1,800 as a highly concentrated market — a threshold lowered from the 2,500 that had applied since 2010.
BCI readings by panel, July 2026
| Panel | BCI-B | BCI-E | Blink | WebKit | Gecko | Source |
|---|---|---|---|---|---|---|
| Africa (all devices) | 6,110 | 7,796 | 87.85% | 8.53% | 2.29% | Axis Intelligence Research, from Statcounter |
| Worldwide desktop | 5,187 | 7,354 | 85.32% | 5.66% | 6.51% | Axis Intelligence Research, from Statcounter |
| Worldwide mobile | 5,120 | 5,878 | 72.32% | 25.44% | 0.67% | Axis Intelligence Research, from Statcounter |
| Worldwide (all devices) | 4,973 | 6,293 | 77.53% | 16.47% | 3.34% | Axis Intelligence Research, from Statcounter |
| Europe (all devices) | 4,128 | 5,803 | 73.69% | 18.67% | 4.94% | Axis Intelligence Research, from Statcounter |
| North America (all devices) | 3,845 | 5,097 | 65.05% | 29.20% | 3.58% | Axis Intelligence Research, from Statcounter |
| Worldwide tablet | 3,651 | 5,436 | 65.75% | 33.36% | not reported | Axis Intelligence Research, from Statcounter |
Not one panel comes in under 3,600. The least concentrated browser market we can measure anywhere on earth — tablets, where Safari’s iPad position is strongest — still scores twice the highly-concentrated threshold. The most concentrated, Africa, scores 3.39 times it.
Read the engine column and the picture hardens. BCI-E is higher than BCI-B in every single panel, because collapsing Chrome, Edge, Opera and Samsung Internet into Blink removes the appearance of choice without removing a single user. Worldwide desktop reaches 7,354 at the engine level: on the device class where browser switching is easiest and the software is free, roughly six sessions in seven render through code Google maintains.
Alex Rivera: The refresh-cycle logic that governs phones and laptops does not apply here, and that is why these numbers stay put. A handset gets replaced every three or four years, and each replacement is a chance for share to move. A browser gets replaced when someone decides to replace it, which is approximately never. Concentration in a zero-cost, zero-friction, infinitely-substitutable product is the least intuitive result in consumer tech, and it has now held for a decade.
Browser Market Share by Device: Desktop, Mobile, and Tablet
Device class moves the table more than any other variable, and the three panels tell three different stories about who the second-place browser actually is.
| Browser | Desktop | Mobile | Tablet | Source |
|---|---|---|---|---|
| Chrome | 70.58% | 66.79% | 48.32% | Statcounter, July 2026 |
| Safari | 5.66% | 25.44% | 33.36% | Statcounter, July 2026 |
| Edge | 11.15% | not reported | 0.91% | Statcounter, July 2026 |
| Firefox | 6.51% | 0.67% | not reported | Statcounter, July 2026 |
| Samsung Internet | 1.59% | 2.48% | 1.39% | Statcounter, July 2026 |
| Opera | 2.00% | 1.80% | 1.02% | Statcounter, July 2026 |
| Android Browser | not reported | not reported | 14.11% | Statcounter, July 2026 |
Second place is Edge on desktop, Safari on mobile, Safari on tablet. Firefox outranks Safari on desktop and trails it by nearly 25 points on mobile. Anyone quoting a single global ranking is averaging three incompatible markets into a number that describes none of them.
The Axis Platform Dependence Score (PDS)
The Platform Dependence Score (PDS) is a second Axis Intelligence Research metric, built to quantify how much of a browser’s position rests on one device class:
PDS = (highest device-class share − lowest device-class share) ÷ highest device-class share
A score near 0 means a browser performs consistently wherever people browse. A score near 1 means its share is a hostage of one platform. We compute it only for browsers Statcounter reports on all three device panels — Edge and Firefox drop below the reporting floor on at least one, and computing a score from a censored zero would manufacture a result rather than measure one. That restriction is the metric working correctly.
| Browser | Desktop | Mobile | Tablet | PDS | Source |
|---|---|---|---|---|---|
| Safari | 5.66% | 25.44% | 33.36% | 0.830 | Axis Intelligence Research, from Statcounter |
| Opera | 2.00% | 1.80% | 1.02% | 0.490 | Axis Intelligence Research, from Statcounter |
| Samsung Internet | 1.59% | 2.48% | 1.39% | 0.440 | Axis Intelligence Research, from Statcounter |
| Chrome | 70.58% | 66.79% | 48.32% | 0.315 | Axis Intelligence Research, from Statcounter |
Safari scores 0.830 — the highest of any fully-observed browser, and the number that explains its second place. Safari is not a browser that lost the desktop; it is a browser whose entire position is a function of how many Apple devices are in a market. Chrome’s 0.315 is the opposite profile: present everywhere, weakest where Apple hardware is strongest.
For anyone building or testing web products, PDS is the practical reading in this dataset. A team that tests on desktop Chrome and desktop Safari has covered 76.24% of desktop sessions and a rounding error of the tablet market. Safari’s tablet share is nearly six times its desktop share.
Browser Market Share by Region: Why the Global Number Misleads
Regional variation is the largest single source of error in browser planning, and it is the part of this market almost nobody publishes with the concentration math attached.
| Region | Chrome | Safari | Edge | Firefox | BCI-B | Snapshot | Source |
|---|---|---|---|---|---|---|---|
| Africa | 77.43% | 8.53% | 3.03% | 2.29% | 6,110 | July 2026 | Statcounter / Axis Intelligence Research |
| Worldwide | 68.22% | 16.47% | 5.37% | 3.34% | 4,973 | July 2026 | Statcounter / Axis Intelligence Research |
| Europe | 60.71% | 18.67% | 7.24% | 4.94% | 4,128 | July 2026 | Statcounter / Axis Intelligence Research |
| North America | 54.01% | 29.20% | 7.46% | 3.58% | 3,845 | May 2026 | Statcounter / Axis Intelligence Research |
Chrome’s regional range is 23.42 percentage points from Africa to North America. Safari’s runs the other way and nearly as far: 29.20% in North America against 8.53% in Africa, a ratio of 3.42 to 1. Those two lines are the same line. Browser share in a region is very largely a restatement of the local mix of device prices, and Safari’s regional share is the clearest published proxy for Apple hardware penetration that exists in open data.
Two regional details worth keeping. Brave appears in North America’s reported six at 2.10% and nowhere else, which makes North America the only large market where an independent Chromium fork clears the reporting floor. And Opera at 3.89% in Africa outranks Edge at 3.03% — a reversal of the worldwide order, driven by data-compression browsing on mid-tier Android hardware rather than by anything happening on Windows.
One honest caveat on this table: Statcounter’s North America panel was serving a May 2026 snapshot when we retrieved it on August 12, 2026, while the other three panels served July 2026. The regional spreads above therefore compare a May reading to July readings. We have labelled the snapshot date in every row of the dataset rather than silently aligning them.
Alex Rivera: Sell-through explains this better than preference does. In markets where the average selling price of a phone sits under two hundred dollars, Android is the only realistic option, Chrome is preinstalled on it, and the browser conversation is over before it starts. Africa’s 6,110 BCI-B is not evidence that Africans like Chrome more than Americans do. It is evidence that Africa buys cheaper phones. The browser market is a hardware market wearing software’s clothes.
Do Different Measurement Panels Agree on Browser Share?
They do not, and the size of the disagreement is a finding in its own right.
Statcounter weights page views from tagged sites and reports 68.22% for Chrome in July 2026. Cloudflare’s 2025 Radar Year in Review measures HTTP requests across its own network and states that roughly two-thirds of 2025 request traffic came from Chrome — a figure Cloudflare publishes qualitatively rather than to two decimals. The U.S. General Services Administration operates a third independent panel at analytics.usa.gov, measuring real visits to thousands of federal websites.
We are deliberately not merging these into a single blended number. Page views, HTTP requests, and government-site visits are three different units of measurement with three different bot-filtering regimes and three different site populations. Averaging them would produce a figure that is more precise than any of its inputs and less true than all of them. Publishing the divergence, and saying why we will not collapse it, is the more useful analysis.
Two structural reasons the panels diverge:
Bot classification. Automated traffic overwhelmingly identifies as Chrome. Any panel that filters bots differently will report a different Chrome share, and the direction of the error depends entirely on where the filter sits. Cloudflare reports that global internet traffic grew 19% in 2025 and that iOS generated 35% of worldwide mobile device traffic, up two points year over year — figures drawn from a request-level view no page-view panel can reproduce.
Engine masquerade on iOS. Every browser on an iPhone renders through WebKit, including Chrome. The user agent says Chrome and the rendering says Safari. Our WebKit row is therefore a floor, and our Blink row is a ceiling — a distinction that matters enormously to anyone using engine share to plan compatibility testing.
What Regulation Has Actually Changed in Browser Share
Two jurisdictions have now intervened directly in browser distribution, and the measurable results are modest.
In the United States, the D.C. district court’s September 2025 remedies decision in United States v. Google declined to order divestiture of Chrome, imposing behavioral remedies instead. The Department of Justice characterized the outcome as significant relief. The Congressional Research Service records that the plaintiffs had argued for divestiture on the basis that Chrome accounts for 20% of all searches in the United States and defaults to Google Search — a distribution argument, not a browsing-quality one.
In the European Union, Chrome and Safari are both designated browser core platform services under the Digital Markets Act, among 23 core platform services currently designated across the gatekeepers. Edge is not designated, which is why Windows users see no choice screen while Android and iOS users in the EU do.
Mozilla reports that since choice screens began rolling out in March 2024, Firefox daily active users in Germany rose 99%, and Firefox daily active users on iOS in France rose 111%. Those are Mozilla’s own figures, published in March 2025, and they are percentage growth on a small base rather than share points — Firefox’s Europe-wide share still reads 4.94% in July 2026, and Europe’s BCI-B of 4,128 remains 2.29 times the highly-concentrated threshold. Choice screens moved a real number. They did not move the structure.
Alex Rivera: A doubling that leaves you under five points is the most honest possible summary of what regulation has achieved here, and Mozilla deserves credit for publishing it in that form rather than as a share claim. The lesson for the next jurisdiction writing a choice-screen rule is not that choice screens fail. It is that a one-time prompt competes against a default that is re-presented every single time the device boots.
Where Do AI Browsers Rank in 2026?
Nowhere yet, and that is the citable fact.
ChatGPT Atlas launched worldwide on macOS on October 21, 2025, joining Perplexity’s Comet and a growing tier of agentic browsers. Nine months later, no AI-first browser appears in any of the seven Statcounter panels examined here — worldwide, desktop, mobile, tablet, Europe, North America, or Africa. Every one of them sits inside a residual that tops out at 2.70% and is shared with Brave, Vivaldi, DuckDuckGo, Yandex, Coc Coc and everything else outside the top six.
Axis Intelligence Research finds that as of July 2026, the ceiling on the entire combined AI-browser category worldwide is 2.66% — the full unreported residual, which they do not have to themselves. That is a bound, not an estimate, and it is the most precise thing that can be honestly said about AI browser share from open data today. Anyone publishing a specific AI-browser percentage is either working from vendor-reported installs or making it up.
The category is also structurally awkward for engine-share analysis: Atlas and Comet are Chromium-based, so every session they win increases Blink’s share while decreasing Chrome’s. Fragmentation at the browser layer and consolidation at the engine layer can happen at the same time, and BCI-B and BCI-E are built to show exactly that divergence when it arrives.
Methodology
Collection. Browser share figures were retrieved directly from Statcounter Global Stats panels on August 12, 2026, one panel at a time: worldwide all-device, worldwide desktop, worldwide mobile, worldwide tablet, Europe, North America, and Africa. Statcounter states its browser panel is based on over 3 billion monthly page views. Regulatory, engine and product facts were retrieved from the primary publisher in each case — the Congressional Research Service, the U.S. Department of Justice, the European Commission’s Gatekeepers Portal, Cloudflare, Mozilla, and OpenAI. No aggregator or secondary summary was used as the source of any figure in this report.
Snapshot alignment. Six of the seven panels served a July 2026 snapshot. The North America panel served May 2026 at retrieval time. Every row in the dataset carries its own as_of_date; no dates were harmonised.
Reporting floor. Statcounter publishes six browsers per panel. Browsers below sixth place are recorded as unreported, never as zero, and the arithmetic residual is stored as its own row and excluded from all index calculations. Gecko’s tablet value is empty for this reason.
BCI formula. BCI-B = Σ(browser share²) across reported browsers. BCI-E = Blink² + WebKit² + Gecko², where Blink aggregates Chrome, Edge, Opera, Samsung Internet, Brave, UC Browser and Android Browser. Both readings exclude the unreported residual and are therefore conservative floors. Benchmark threshold of 1,800 index points is the 2023 U.S. Merger Guidelines standard for a highly concentrated market.
PDS formula. PDS = (max device-class share − min device-class share) ÷ max device-class share, computed across the worldwide desktop, mobile and tablet panels, and only for browsers reported on all three.
Scope of the readings. Page-view-weighted panels count sessions, not people: a heavy desktop user contributes more measured page views than a light mobile user. Engine attribution follows the browser’s engine on its home platform, so WebKit is understated and Blink overstated for iOS traffic, where all browsers render through WebKit. Regional panels aggregate countries with very different device mixes. All arithmetic in this report was independently recomputed before publication, and every figure in the prose has a matching row in the dataset.
Verification. Each number in this article was matched against its dataset row and its source URL was confirmed to resolve and to contain the claim before publication.
About This Dataset
Contents. 119 rows covering browser session share for six to seven browsers across seven Statcounter panels (worldwide all-device, desktop, mobile, tablet, Europe, North America, Africa), Axis-calculated engine shares, Browser Concentration Index readings at browser and engine level, Platform Dependence Scores, regional spreads, and dated regulatory and product context from U.S., EU, and vendor primary sources.
Temporal coverage. May 2026 – August 2026, with regulatory context from December 2023 onward.
Spatial coverage. Worldwide, with regional cuts for Europe, North America, and Africa.
Creator and publisher. Axis Intelligence Research.
License. CC BY 4.0 — free to use, share and adapt with attribution.
Cite as:
- APA: Axis Intelligence Research, & Rivera, A. (2026, August 12). Browser market share statistics 2026: Concentration, engines, and the numbers panels disagree on. Axis Intelligence Research. https://axis-intelligence.com/browser-market-share-statistics/
- MLA: Axis Intelligence Research and Alex Rivera. “Browser Market Share Statistics 2026: Concentration, Engines, and the Numbers Panels Disagree On.” Axis Intelligence Research, 12 Aug. 2026, axis-intelligence.com/browser-market-share-statistics/.
- Chicago: Axis Intelligence Research and Alex Rivera. “Browser Market Share Statistics 2026: Concentration, Engines, and the Numbers Panels Disagree On.” Axis Intelligence Research, August 12, 2026. https://axis-intelligence.com/browser-market-share-statistics/.
Frequently Asked Questions
Why does Chrome’s share differ by several points between Statcounter and Cloudflare?
The two panels measure different units. Statcounter weights page views from sites carrying its tag; Cloudflare counts HTTP requests crossing its own network. They also filter automated traffic differently, and bots overwhelmingly present as Chrome, so the bot-filtering decision alone can move Chrome’s reported share by multiple points in either direction. Neither is wrong — they answer different questions, which is why Axis Intelligence Research publishes both readings side by side instead of blending them.
Should engine share or browser share drive cross-browser testing budgets?
Engine share, with one adjustment. Rendering bugs live in engines, and worldwide Blink accounts for 77.53% of named sessions against WebKit’s 16.47%. But because iOS forces every browser onto WebKit, the true WebKit rendering share is higher than Safari’s reported number, and the practical rule is that a Blink target plus an iOS WebKit target covers the overwhelming majority of real rendering paths. Gecko at 3.34% worldwide is a deliberate budget decision, not an automatic one — its 6.51% desktop share means a desktop-heavy product faces a different calculation than a mobile-first one.
What does the Browser Concentration Index measure that market share does not?
Share tells you who leads. BCI tells you how much room the structure leaves for anyone else, on a scale regulators already use. Two markets can both have a 60% leader and score very differently depending on whether the remaining 40% sits with one strong challenger or twelve weak ones. Europe at BCI-B 4,128 and North America at 3,845 have leaders 6.7 points apart, but the index gap between them is driven mostly by how much of the residual Safari holds in each.
Why is Safari second worldwide but fourth on desktop?
Safari’s Platform Dependence Score is 0.830, the highest of any browser reported on all three device panels. Its 25.44% mobile and 33.36% tablet shares carry its global position; its desktop share is 5.66%, behind Chrome, Edge and Firefox. Safari’s ranking is a function of Apple’s device mix in a given market, not of desktop browser competition.
Does the unreported residual mean Brave, Vivaldi and DuckDuckGo have zero share?
No. Statcounter publishes six rows per panel, so anything in seventh place or below is unreported rather than absent. The residual runs from 0.89% on worldwide tablet to 2.70% in Europe. Brave clears the floor in exactly one panel examined here — North America, at 2.10% — which sets a useful reference point for how large a browser has to be to become individually visible in this data.
Have EU browser choice screens changed measured share?
They have changed measured users more than measured share. Mozilla reports Firefox daily active users up 99% in Germany and up 111% on iOS in France since choice screens began in March 2024. Firefox’s Europe-wide share reads 4.94% in July 2026, and Europe’s BCI-B of 4,128 remains 2.29 times the U.S. highly-concentrated threshold. Choice screens are one-time prompts competing against defaults that re-present at every boot.
Why did Chrome’s share not fall after the U.S. antitrust ruling?
Because the ruling did not touch distribution in the way the plaintiffs asked. The September 2025 remedies decision declined to order divestiture of Chrome and imposed behavioral remedies instead, principally around exclusive contracts and data sharing. Behavioral remedies operate on contract terms; browser share moves on default placement, and default placement was left substantially intact.
Is Opera really larger than Edge anywhere?
In Africa, yes — Opera reads 3.89% against Edge’s 3.03% in July 2026, reversing the worldwide order where Edge leads Opera 5.37% to 1.88%. The driver is Android hardware and data-cost-sensitive browsing rather than anything happening on the desktop, and it is a useful reminder that a global ranking table describes no individual market accurately.
How should this dataset be used for forecasting?
Cautiously, and not on a monthly basis. Browser share moves on device replacement cycles and default-placement changes, both of which operate over years. Month-to-month movement in these panels is dominated by panel composition and sampling rather than by user behaviour, which is why the review trigger for this page is a 100-point BCI-B move rather than a calendar date.
What would actually move the Browser Concentration Index?
Three things, in descending order of plausibility: a change to default browser placement on Android or Windows at OS level; a shift in the global smartphone price mix that changes Apple’s regional penetration; or an AI-first browser clearing a reporting floor for the first time. The third is worth watching precisely because Atlas and Comet are Chromium-based — they would lower BCI-B while leaving BCI-E almost untouched, which is the specific divergence this index pair was built to catch.