The 2026 gTLD Round Is Already Separating Winners From Losers

Last year I compared the domain industry to Barcelona’s kiosks. The old ones piling sangria bottles and fridge magnets for whoever walked by. The new ones building the entire shelf around one specific customer.

The new gTLDs 2026 Round just closed. More than 1,600 primary applications, over 1,100 of them filed with a backup string in case of contention. And the data already coming out tells us who built a shelf, and who’s about to find out the hard way that they didn’t.

Here are 5 signals worth paying attention to before Reveal Day in October

1. Most applicants filed the same way they did in 2012, at the last minute.

ICANN confirmed that the majority of the 1,600+ applications landed in the final days of the 15-week window, the same rush pattern seen in 2012. Filing late isn’t automatically a problem, plenty of good strategy gets locked in right up to a deadline. I just hope it isn’t a signal of how the rest of the planning, the part that actually determines whether the string succeeds, got the same last-minute treatment.

2. Losing now costs real money. There’s no soft landing left.

This is the structural change nobody’s talking about enough. In 2012, when two applicants wanted the same string, 234 contention sets formed and only 16 ever reached an ICANN auction. The rest were settled privately, with winners paying losers to step aside, so even “losing” often meant getting paid. The 2026 Guidebook bans that entirely. Private auctions, side deals, and joint arrangements are now explicitly prohibited, with penalties up to disqualification from future rounds. An applicant who loses a contention set this time walks away roughly $147,550 poorer out of the $227,000 fee. No payout, no consolation prize. That single rule change makes reckless, defensive applications far more expensive to have been wrong about.

3. Technical approval is not the same as commercial readiness.

The 2026 Round introduced a Registry Service Provider Evaluation Programme, where backend providers get pre-approved once and can support multiple applicants. It’s a genuinely good efficiency fix. But it also means passing ICANN’s bar just got easier operationally, while the harder question, whether anyone has a plan to sell the string once it’s live, still isn’t part of that evaluation at all.

4. The renewal math from 2012 still hasn’t been solved.

New gTLDs from the last round have hovered around a 32% renewal rate, against roughly 75% for legacy gTLDs and 72 to 82% for major ccTLDs. Com Laude‘s own team has pointed to the same root cause: most 2012 applicants filed without a real use case or internal alignment behind the string. A bigger applicant pool in 2026 doesn’t fix that math. It just means more Registries are about to test it in real time.

5. Some players are already building for distribution, not just ownership. That’s the split.

Namecheap set up Starlight Registry and applied for 40 TLDs. Porkbun and Dynadot’s CEOs reportedly teamed up under a new entity, Oinkadot, to apply for 25 strings together. Whatever you think of the specific strings, the instinct is right: these are operators who already control distribution, not just applicants hoping a Registrar will give them shelf space later. Mi amigo Michele Neylon said it: Registrars won’t onboard a new TLD unless the value is obvious and the backend is ready to go. The applicants who understood that before Reveal Day are the early winners here. The ones still hoping a generic launch campaign will do the work are the early losers, and they just made that mistake $147,550 more expensive.

That’s the real split this round is revealing. Not who applied for the most strings, but who’s treating GTM as part of the application, not an afterthought once the string gets delegated.

That’s also exactly what I’m impatient to see from this batch of Registries. Some, like Oinkadot, already sit close to distribution through their founders’ own registrars. Others don’t have that built in yet, and for them a huge part of the outcome won’t be decided by ICANN, it’ll be decided by which Registrars pick these strings up and what kind of GTM they’re willing to build around them.

One name worth watching for the opposite reason is Link Freedom Group with Vaughn Liley as recently appointed CEO with its 300+ string portfolio, but he built his track record at Nova Registry growing .link. Nova partnered directly with WordPress-owned Gravatar and Bitly to build .link into “link in bio” and digital business card products, giving people an actual reason to use the extension instead of just registering it. Five years of that work is exactly why Vaughn could react calmly when Telegram applied for .gram with a billion-user vanity domain pitch: he’d already been proving the model works at a smaller scale. That’s the instinct the rest of this round needs more of: Portfolios with a distribution plan already attached.

By the way, this is exactly the gap Jaumatic exists to close. We build upsell agents that read public web and DNS signals, exposed tech stacks, dedicated listing sites, the organizations behind a domain, missing security records, to find which of a Registry’s or Registrar’s existing customers actually fit a new TLD. Then we use a commercial framework based on how people actually judge relevance, not discounts and promos, so outreach only reaches the customers it makes sense for, instead of going out generically to a Registrar’s whole base. That’s the difference between a renewal curve that holds and one that falls off a cliff at year one. And it’s the one part of this round that Reveal Day won’t decide.