I woke up on 13 August 2026 expecting a quiet confirmation that the window had shut. Instead I got a number that made me sit down. The ICANN 2026 round closed with more than 1,600 primary gTLD applications. Full disclosure: I almost treated 1,600 as "about what 2012 did" and moved on. That's lazy. 2012 produced a little over 1,900 applications across a longer, messier first expansion. This round packed 1,600-plus into 15 weeks, and ICANN says most of them landed in the final days. Last-minute piles always hide surprises. That's the point. I wanted the paid list, not the applause line.

If you only buy and sell names — not run a registry — this still hits your week. Reveal Day is coming. Contention sets will leak into .com comps. I've already had two founders ask whether they should "wait for the new endings" before buying a brandable. My take: no. Not if they need to launch. A clean name like AiFolio.app is a product you can use this quarter. A new TLD is a years-long project you don't control. Different sports. I keep saying that because founders mix them up every week.

I verified the figures the same morning against ICANN's 13 August announcement. Not a Twitter screenshot. The primary post. As of August 2026, the count is still provisional because the evaluation fee hasn't all cleared. That caveat matters. I'll come back to it.

What the ICANN 2026 round actually filed

More than 1,600 primary applications. More than 1,100 of those also carried replacement strings — secondary applications ICANN will use if the first string dies. That's a lot of backup names. It tells me applicants expected collisions, objections, or string-similarity kills. I've tracked this across two decades of aftermarket noise, and backup strings are a tell. People don't pay extra paperwork for fun.

The submission window ran 15 weeks and closed 12 August 2026 at 23:59 UTC. ICANN is blunt about the rush: the majority arrived at the end. I've seen this in every ICANN process with a hard clock. People wait. Then they dump. The dump is where the interesting applications hide — brands that didn't want to signal early, funds that needed a last board vote, geo applicants still arguing with a city council. Not a theory. A pattern.

For industry color beyond the official count, I also read the DNJournal note that Link Freedom Group published 300-plus submissions instead of waiting for Reveal Day. That's one player. Tens of millions in fees if those applications all pay. I'm not endorsing the strategy. I'm noting the scale. One company trying to define a generation of endings will show up in contention sets you care about.

MilestoneDate / ruleWhy I care Window closed12 August 2026, 23:59 UTCNo new strings. The map is frozen. Primary applicationsMore than 1,600 (provisional)Bigger fight than a sleepy round. Replacement stringsMore than 1,100 applications included themExpect collisions and plan-B names. Evaluation fee due19 August 2026, or 7 days after invoice — whichever is laterThe real count drops after no-pays. Reveal DayNo later than ~9 weeks after close; date due mid-SeptemberThat's when aftermarket pricing twitches.

Is 1,600 the final number?

No. ICANN said the final total cannot be confirmed until the gTLD evaluation fee is received. The fee is due by 23:59 UTC on 19 August 2026, or 23:59 UTC seven days after the invoice date — whichever is later. Miss it and the application dies. I've watched people obsess over essays and forget the wire. Don't laugh. It happens more often than the applicant guide admits.

So the ICANN 2026 round headline is a ceiling until money clears. Some of those 1,600 will vanish. Some replacement strings will never see daylight. If you're building a model off "1,600 new endings," you're modeling a press release, not a registry. I want the paid list. That's the list that hits comps.

Program mechanics still live on the New gTLD Program pages. I keep that tab open when I argue with people who think Reveal Day is next Tuesday. It isn't. ICANN expects to announce the Reveal Day date in mid-September 2026, and to publish applications that pass the administrative check no later than nine weeks after the window closed — absent extraordinary circumstances. Mid-October is the honest window in my notebook. Not a promise. A planning date I can actually put on a calendar without lying to myself.

When is Reveal Day, and what should I do before it?

Reveal Day is when ICANN publishes the public-facing parts of applications that passed the administrative check — the strings, the applicants, and which names sit in contention with each other. You will not get a live TLD the next morning. You will get a map. That's enough to move money.

Here's the checklist I'd actually run between now and mid-October. Not a slide. A week's work stretched across two months.

  1. Write a watchlist of 20–40 strings that would hurt or help names you already hold if they become endings — brand words, city words, category words.
  2. Mark which of those you would buy as .com / .app now versus wait. Waiting is a decision. Write it down.
  3. Pull recent comps on NameBio for your watchlist so you have a baseline before Reveal Day adrenaline hits.
  4. Read ICANN's own next-steps list: comments, objections, auctions, and the ban on privately settling contention. Private deals are out this round. Auctions are in.
  5. Set a calendar reminder for mid-September when ICANN says it will name the Reveal Day date — then a second reminder for the nine-week outer bound.

I've said this to investors until I'm bored of my own voice. My take hasn't changed. I'd rather look early than clever. Honestly, the boring calendar wins.

Should aftermarket buyers wait for the new endings?

Usually no. A new gTLD is not a substitute for a name you can type, email from, and put on a pitch deck in 2026. Evaluation, objections, contention, and delegation stretch for years. I've watched founders freeze a rebrand waiting for an ending that still doesn't resolve. That's how you lose a year.

What I would wait for: speculative inventory that only makes sense if a specific string launches and you want to be the first to squat under it. That's a different bet. High holding cost. Ugly if the string dies in evaluation. If your job is selling brandables to operating companies, keep selling. Look at the live premium domain marketplace the same week you skim the ICANN PDF. One of those actions funds the other.

For sales context while this circus runs, I still start with the midyear 2026 sales roundup. Big .com checks didn't pause because someone filed for .something. They rarely do. And if you're a founder stuck between a taken .com and a fresh alternative, how to choose a startup domain is the decision tree I send people who are spiraling.

What the ICANN 2026 round changes in the aftermarket

Three messy effects, not a framework. First: brand owners who applied for a .brand may stop shopping defensive second-level names in every junk TLD — or they may buy more, because 1,600 endings is a lot of Sunrise periods. I don't know which way that net goes yet. I'd rather say that out loud than fake a model.

Second: contention. When two or more applicants want the same string, this round sends them to ICANN auctions. No private settlement. That means more public price discovery and more leftover applicants with cash and a bruised ego. Bruised applicants sometimes buy the matching .com they should have bought in year one. I've seen that movie after 2012. Not identical. Rhymes enough that I still watch leftover applicants for .com shopping sprees.

Third: noise. Every journalist will write "the internet is getting 1,600 new endings." Customers will ask if .com is dead. It isn't. Email still defaults there. Investors still notice. I keep repeating the ICANN 2026 round number because the number is the story people will misquote. Correct them with the fee caveat and the years-long pipeline. Then sell them a name they can use.

Escrow questions still cluster around the same sticking points — lock, auth code, release. Our acquisition FAQ covers the ones I get weekly. If you're ranging an offer before Reveal Day FOMO, the calculators on domain tools are what I actually open. Not decoration. I use them when a founder wants a number that isn't a vibe.

If you filed, good luck with the long part. If you didn't, you're not late to the only game that matters this quarter — owning a name people can type. Tell me which strings you're watching. I'm still updating my own list.