.si Domain Investing Mistakes

The most dangerous number in the .si market isn't an asking price. It's $12.

At $11–12 a year, registration is cheap enough to bypass every quality filter an investor normally has. The result: 46,066 September registrations, most of them bought by people who will sell to each other, not to end users. The mistakes below aren't about picking bad names — that's a separate evaluation problem, covered in our guide to what makes a strong .si domain. These are about behaving badly in a cheap market. Each one is documented. Each one has a fix.

Mistake 1: Registering in bulk because it's cheap

What it looks like: 200-name portfolios of two-word keyword combos, registered in an afternoon. September produced 46,066 of them.

Why it hurts: the registration fee is a one-time illusion; the renewal is a recurring liability. A $2,400 registration spree becomes a $2,400-a-year carrying cost, due again in twelve months — against names that, per witscode, "rarely resell above the registration fee." Meanwhile 52% of the first wave was immediately relisted, so you're competing against your own supply the moment you try to exit.

As NotRenewing's operator Sully put it: "Nobody ever talks themselves into two hundred bad names at $85 apiece. At twelve bucks you don't even slow down enough to ask the question."

What to do instead: budget by renewal, not registration. Before you register, ask whether you'd pay the renewal fee a year from now with no sale in sight. If not, don't pay the registration fee today.

Mistake 2: Assuming .si = .ai economics

What it looks like: "AI did it, SI will do it" — underwriting .si on .ai's price history.

Why it hurts: the analogy breaks three ways. First, .ai's rise took more than a decade plus genuine startup adoption — 140,000 domains in 2022 to over 1 million by early 2026. .si is trying to compress that into weeks on a political headline. Second, the China tonal problem: as investor Ray Hackney noted via NamePros, ài (love) helped .ai in China while sǐ (death) doesn't do .si any favors — and China is a major domain-buying market. Third, the structural handicap: Google lists .ai among the ccTLDs it treats as generic; .si is not on that list, so Google geotargets it to Slovenia — a ranking handicap for exactly the US startups investors hope will buy.

What to do instead: underwrite .si on its own adoption evidence — confirmed sales, live end users, renewal rates — not on another extension's history.

Mistake 3: Buying to flip to other flippers

What it looks like: purchasing at investor asking prices on the assumption that a greater fool is one listing away.

Why it hurts: the buyer pool is shockingly thin. Domain Name Wire found 33 of 35 first-half 2026 sales were investor-to-investor. Netcraft measured 52% of first-wave registrations instantly relisted for resale. And on one investor platform, 44 .si sales went to 21 buyers — with three buyers accounting for 21 of those sales. As Sully summarized: "Nearly half the demand is three people. That isn't a market yet. That's a few guys with a thesis and a credit card."

Meanwhile, the end-user side is nearly empty: Leanne Mac's ProductHunt TLD tracker, cited by Domain Name Wire, recorded zero startup launches on .si in 30 days — versus fifty on .ai in the same period.

What to do instead: assume your buyer is an end user who doesn't exist yet, and price your entry accordingly. If a name only makes sense as a flip to another speculator, you're buying the last chair in musical chairs.

Mistake 4: Using asking prices as comps

What it looks like: "Build.si is listed at over $14 million, so my two-word .si must be worth five figures."

Why it hurts: asking prices are wishes, not transactions — and in this market the gap is grotesque. Build.si last sold for $17,435 (NameBio) and is now asking over $14 million. Eighty-seven brand-plus-"superintelligence" .com domains were registered September 29 and listed at $20,000 each — zero ownership changes detected. Twenty-one two-word .si names hand-registered September 22 for $13.76 a year are asking $1,995 each.

Comps are completed sales. The largest confirmed post-wave .si sale is $18,888 (otherwise.si, Spaceship, September 24). Everything in the millions is an asking price.

What to do instead: build comps only from reported completed sales — see our confirmed .si sales ledger — and treat any "comp" without a venue and date as fiction.

Mistake 5: Believing .si "sales" posts on X

What it looks like: screenshots of spectacular .si "sales" circulating on X, each one nudging you to buy before it's too late.

Why it hurts: fabricated .si "sales" are actively circulating to pump the narrative — a "beware fake domain sales" thread is running on NamePros for exactly this reason. The highest-profile example: a widely shared X thread alleged insider profiteering around the executive order, including the Build.si $14M figure — but The Register reported the author could not produce categorical evidence linking any investments to the alleged insiders, and BeInCrypto found no public evidence tying buyers to the claims.

What to do instead: verify every claimed sale against NameBio before letting it move your pricing by a dollar. If it isn't in a sales database with a venue and date, it didn't happen.

Mistake 6: Ignoring registry friction

What it looks like: treating .si like .com with a cheaper price tag.

Why it hurts: Slovenia's registry (ARNES) has rules that surprise newcomers. There is no WHOIS privacy — your contact email is permanently exposed in the public database, a fact generating real complaints from new .si holders on NamePros (spam, scrapers, phishing exposure). The registry publishes no zone file, so ownership is opaque. It operates a domain dispute procedure — and its September 30 advisory urging trademark holders to defensively register .si signals disputes are coming.

What to do instead: register with a dedicated contact email you can afford to expose, keep holder data accurate, and read the registry terms before you hold 200 names there — not after the first dispute notice.

Mistake 7: Reading registration counts as demand

What it looks like: "46,066 registrations in September — the demand is undeniable."

Why it hurts: registration counts measure registrar throughput, not buyer demand. Forty-six thousand registrations can mean 46,000 buyers — or 460 buyers averaging 100 names each, most of them immediately listed for resale. The data points to the latter: 52% instant relisting, three buyers driving half of one platform's volume, zero startup launches in 30 days.

The real demand signals are confirmed sales, live end-user deployments, and — the verdict that matters most — renewal numbers. As Sully put it: "I'll be watching next year's renewal numbers." September 2027 will tell you whether this was a market or a moment.

What to do instead: track the three numbers separately — registrations (booming), confirmed sales (thin), end-user adoption (near zero) — and make decisions on the middle one. Our pillar article, .si Domains: Super Intelligence Opportunity or Speculative Bubble?, keeps all three separated with sources.

Conclusion

None of these mistakes is about intelligence — they're about discipline. Cheap registration fees manufacture the illusion of a liquid market; asking prices manufacture the illusion of value; registration counters manufacture the illusion of demand. The investors who survive the .si wave will be the ones who bought few names, paid renewal-math prices, verified every comp, and waited for end users instead of flippers. In a $12 market, restraint is the entire edge.