ICANN 2026 opens April 30 · Next round after policy review — 3-7 years minimum

ICANN 2026.
Who's really applying?

Public disclosures list 31 organizations. Our model predicts 2,239 more — scored by country, sector, and probability of filing. Built from 1,930 filings in 2012, 50,721 UDRP cases, and World Bank macro signals.

By Matthieu Credou — Afnic 2011-2017, on the new gTLD program team. Lived through the 2012 round.

2,487
strings with candidate predictions
2,239
modeled applicant entities
30
countries forecast (NB GLM v3)
1,777
central forecast · IC80% 1,513–2,079

Why 2026 matters

Once every 14 years. Minimum.

The 2012 round had 1,930 applications. It took ICANN 14 years to open the 2026 round. After this one closes, ICANN will conduct a policy review before launching another — 3 to 7 years, in the most optimistic scenarios.

Every string awarded in 2026 is locked until the 2030s. Every string you miss filing on is a competitor's leverage for the same period.

Key finding — teaser

Of 14 L'Oréal gTLDs filed in 2012,
zero are operational today.

All eight brand TLDs (.lancome, .garnier, .maybelline…) were delegated and then returned to ICANN. All five generic TLDs (.salon, .makeup, .skin, .hair, .beauty) were lost in auction to portfolio operators. The Chinese IDN was captured by a mainland registry.

This is the cautionary tale at the heart of our forecast — and it's why predicting 2026 requires more than a 2012 growth factor. 200+ named-entity predictions, each with a calibrated P(applies) and revealed-preference class, inside the Global Master.

See the full forecast →

Methodology

Not a market-research report. A formal Bayesian forecast, validated against IPO-retry benchmarks and pre-2010 retrodiction.

Log-normal prior on total applications

Centered on the 2012 anchor (1,930 apps), with an 80% credible interval of ≈ [1,315; 2,833]. Every country-level deviation requires structured covariate evidence, not arbitrary optimism.

Hierarchical Poisson model

Country-level application counts modeled with partial pooling: log λ_c = α + u_c + X_c β. Shrinks noisy low-base outliers (UAE, Saudi Arabia) toward a global mean conditional on GDP, UDRP intensity, active-brands count, and digital-transformation program signals.

Four revealed-preference classes

Entity-level P(apply in 2026) calibrated by 2012 outcome: still-operating dotBrand · self-terminated · lost-in-auction · never-applied. Priors anchored to external benchmarks (IPO re-filing 9–13%, SEO reissue 25%, failed M&A 25–40%) rather than picked from air.

Baseline + named-entity decomposition

N_c = N_baseline + N_named. Post-2012 entrants (OpenAI, Anthropic, Neom, ByteDance, BioNTech) are modeled explicitly and subtracted from baseline covariates to avoid double-counting.

Retrodiction validation

The same framework, constrained to pre-2010 covariates, is used to retrodict the 2012 landscape. Appendix documents where it lands and where it misses.

Peer-reviewed by an external forecaster

The framework was reviewed and refined in consultation with an expert forecaster specializing in one-shot corporate decision prediction. Credits + methodology appendix in the Global Master report.

Founding CustomerFirst 10 Master buyers — limited slots

Pricing

Lock in the lowest price forever + extras

String Contention Report

Unlock one string's full candidate list.

499

One-time purchase · VAT extra · Full report + candidate detail

  • Full named-candidate list for one string
  • Per-candidate P(file) probability
  • Country + sub-sector tagging
  • Rationale excerpts from our research
  • Poisson-binomial auction probability

Country Report

One country, deep-dive.

7991 299

One-time purchase · VAT extra · 30-40 pages

  • 2012 → 2026 forecast for one country
  • Top 30 named brand predictions with P(applies)
  • International competition targeting this market
  • Contention-risk strings
  • Defensive vs offensive scenarios

Founding Customer extras

  • +1h analyst call
  • +Grandfathered price for 2027 edition
Most complete

Global Master

The full 2026 picture — database + PDF + Excel.

4 9998 999

One-time purchase · VAT extra · 100-120 pages + full CSV (4,132 rows)

  • Total 2026 forecast + 80% confidence intervals
  • 30-country breakdown (Bayesian NB GLM v3)
  • 4,132 named-candidate master database (CSV + Excel)
  • 16-sector classification + per-candidate P(applies)
  • Top-200 brand predictions + contention forecasts
  • Post-2012 entrants (OpenAI, Anthropic, Neom, ByteDance…)
  • Methodology appendix + UDRP-cited rationale

Founding Customer extras

  • +2h methodology deep-dive session
  • +1h analyst call
  • +Grandfathered price for 2027 edition
  • +Founding Customer logo on report (opt-in)

Need a sector report (Banking, Pharma, AI & Web3, Luxury…) or bespoke analysis? Let's talk →

Corporate procurement / invoice payment on Master tier? Request an invoice →