Popular Domain Extensions in Mid-2026: What the Numbers Actually Say

“Most popular domain extension” sounds like a single ranking. It is not. Registration volume, everyday recognition, and aftermarket prestige pull in different directions—and a lot of 2025–2026 listicles blurred those lines with numbers that never appeared in registry reports.

This piece is a mid-year argument for people choosing (or evaluating) an extension in 2026—not a suffix-by-suffix tour. Figures below come from primary or near-primary sources: Verisign’s Domain Name Industry Brief (DNIB), registry announcements, DENIC, Public Interest Registry, ntldstats.com zone tracking, NameBio sale records, and ICANN’s New gTLD Program materials. Where a number is an industry estimate rather than a registry disclosure, it is labeled as such.

Popularity is three different questions

Registration base answers: How many names sit in the zone today? Verisign’s DNIB is the best public aggregate for this. At the end of Q2 2026 it counted 401.6 million domains across all TLDs—up 8.1% year over year and 2.3% from Q1 2026 (Verisign / DNIB.com, 23 Jul 2026).

Cultural default answers: What do customers type when they guess your URL? For most global consumer brands, that is still .com. For a DACH audience, .de often wins locally. For an AI product pitch deck, .ai can carry meaning that .com does not.

Aftermarket prestige answers: What do brokers and investors pay for scarce strings? That market is thin, opaque, and dominated by outliers. A seven-figure sale proves demand for that string, not a reliable average resale price for the whole TLD.

Treating those three as the same metric is how you get impossible claims—such as .com somehow holding more than 300 million names when Verisign’s own base for .com alone was 166.6 million at mid-2026.

Question one: who actually owns the registration pie

DNIB’s Q2 2026 breakdown is the cleanest public cut:

| Bucket | Registrations (30 Jun 2026) | Share of total | | --- | ---: | ---: | | .com + .net | 179.1 million | ~44.6% | | Country-code TLDs (ccTLDs) | 148.6 million | ~37.0% | | Everything else (legacy gTLDs beyond .com/.net + new gTLDs) | ~73.9 million | ~18.4% |

Within that first row, Verisign reported 166.6 million .com and 12.5 million .net. Combined .com/.net grew 5.1% year over year—solid, not explosive. New .com and .net registrations in the quarter were 12.7 million, versus 10.4 million in Q2 2025. At 166.6 million, .com alone is about 41.5% of all domains Verisign counted—not half the internet, and nowhere near the 300-million-plus figures that circulated in recycled “2026 popularity” posts.

Wholesale pricing for .com is public and boring in a useful way. Under Verisign’s agreements with ICANN and the U.S. Department of Commerce, the .com wholesale fee was $10.26 after the September 2024 increase. Registrars set retail; end-user prices usually land in the mid-teens USD per year. Further wholesale increases of up to 7% in eligible years of the current pricing cycle remain contractually possible—budget for gradual upward pressure, not a sudden collapse of affordability. What .com does not guarantee is a short, brandable string. That scarcity—not a mythical 18% annual base explosion—is why premium .com aftermarket prices stay high.

.net and .org sit in the same order of magnitude as each other (~12–12.5 million), not as rivals to .com. .net is no longer the automatic 1990s “tech alternative”; it is a familiar general-purpose fallback when the matching .com is taken. .org, operated by Public Interest Registry (PIR), crossed 12 million registrations in mid-2026 per PIR’s milestone announcement—up from about 2.6 million when PIR took over more than two decades earlier. Registration is open to anyone; the cultural signal (mission, community, open-source) is still real for audiences who read it that way.

ccTLDs grew 3.6% year over year. DNIB’s top ten by size at mid-2026: .cn, .de, .uk, .ru, .nl, .br, .fr, .au, .in, .eu. DENIC reported that .de passed 18 million registrations in June 2026—one of the world’s largest namespaces. Separately, ntldstats.com—which tracks the 2012-era new gTLD program—showed roughly 48.5 million domains across those extensions as of late July 2026. Volume leaders on that board (.xyz, .top, .shop, .online, and similar) are often shaped by first-year promotions; renewal rates matter more than launch-month headlines. In June 2026, XYZ Registry announced that its portfolio of 35 extensions had surpassed 10 million registrations, led by .xyz—registry-side confirmation that the namespace is large even when trackers disagree on exact zone size day to day.

High new-gTLD volume is not a prestige ranking. It often means aggressive first-year pricing and registrar campaigns. Semantic and restricted endings such as Google’s .app and .dev (HTTPS required) sit far below the bulk leaders on raw count yet carry clearer product meaning—another reminder that “popular on the volume chart” and “useful for your brand” are different axes.

Question two: what people type when they guess

Volume does not decide cultural default. Strangers guessing a global consumer URL still default to .com. Local businesses in Germany, the UK, France, Brazil, India, and similar markets should treat the matching ccTLD as first-class—sometimes ahead of an awkward .com—because “popular” there means trusted at home, not loud on social media.

A smaller set of ccTLDs hitchhiked into global use because the letters read as English words or tech shorthand. .ai (Anguilla) crossed roughly one million registrations around the turn of 2025/2026—up from on the order of ~100,000 in 2020 and a few hundred thousand in 2023—after registry operations moved onto Identity Digital’s platform in early 2025 under a revenue-sharing arrangement with the government. Retail pricing is typically in the low-to-mid hundreds of USD for a two-year term (2026 registry fee changes nudged that higher), so renewal cost belongs in the decision, not just the first invoice. .io remains widely used in developer and SaaS branding; exact live counts are less consistently published than Verisign’s .com figures, with industry reporting in recent years placing it in the roughly one-million-plus range. Buyers should also track geopolitical and policy coverage around the British Indian Ocean Territory code—reputation and continuity risk sit outside pure branding taste.

Neither hitchhiker “beats” .com on trust for a mainstream consumer audience. Both can beat a compromised .com hyphenation when the audience already speaks startup shorthand. The same logic applies to meaning-bearing new gTLDs (.tech, .cloud, .studio, .app, .dev): they help when the word matches the business and the second-level string is strong. They fail as a consolation prize for a weak name. Use meaning only when it is earned; do not rent ambience from a suffix and call it strategy.

Question three: aftermarket headlines are not a TLD ranking

Publicly reported sales are useful as ceiling anecdotes, not as TLD-wide averages.

NameBio’s all-time leaderboard lists ai.com at $70 million (April 2025, brokered via GetYourDomain.com)—widely covered as the largest disclosed domain sale, later confirmed in reporting around Crypto.com CEO Kris Marszalek’s purchase. Earlier record markers such as voice.com ($30 million, 2019) and high-teen-million sales like chat.com show that category-defining .com strings still set the price frontier.

What those deals do not support is invented “average resale” figures for entire extensions (for example, claiming .com averages five-figure resales or that .ai averages six-figure resales across the board). Most registered names never trade. Most that do trade sell for far less than the headlines. If you need a market check, look up comparable sold names on NameBio or major aftermarket venues for your string pattern—not a single TLD-wide average pasted from a blog roundup.

Choose the extension without crowning a winner TLD

1. Start with the audience’s default. Global B2C and many B2B buyers still assume .com. Local markets with strong ccTLD norms should treat those codes as peers, not afterthoughts.
2. Use meaning only when it is earned. .ai for an AI product, .org for a mission brand, .dev for a developer tool—these help when the rest of the name is clean. They do not rescue a hard-to-spell brand.
3. Price the renewal, not the promo. Cheap first-year new gTLDs and premium ccTLDs can diverge wildly at year two. Model three-year cost before you fall in love with a string.
4. Prefer one primary domain. Defensive registrations on alternate TLDs are fine; splitting your public identity across three endings usually confuses more than it protects.
5. Check operational friction. Some new gTLDs face higher spam filtering friction or weaker recognition among non-technical users. Test how the name sounds on a phone call.

If you are still generating candidates, availability across several endings matters more than crowning a single “winner” TLD. Tools that search multiple extensions at once—including NewName.ai—are useful for that inventory step; the strategic choice remains yours.

What ICANN’s 2026 round changes—and the mid-year map that does not

ICANN opened the New gTLD Program: 2026 Round application window on 30 April 2026, closing 12 August 2026 (ICANN announcement; program hub at newgtldprogram.icann.org). This is the first full application round since 2012. It matters for companies and communities that want to operate a TLD (.brand, city, or industry strings). It does not instantly create better retail domains for startups shopping next week—delegation and launch take years after evaluation.

For everyday registrants, the near-term landscape is still the one DNIB describes: a huge .com base (~166.6M), a large and locally loyal ccTLD layer (~148.6M, with national giants like .de at 18M+ and global hitchhikers like .ai at ~1M+), and a long tail of newer endings where bulk volume (.xyz and peers) must not be mistaken for prestige. Highest disclosed sale energy still concentrates in ultra-premium .com strings; .ai commands strong category interest, not uniform six-figure averages. Choose the extension that matches how your customers already navigate the web. Let registry data correct the hype—not the other way around.