I checked the exact-match .coms in this week's hottest AI niches. Here's what was left.

[taunon]

Disclosure: I'm the founder of NameStation; the topic lists below come from our weekly trending-topics page (namestation.com/explore/trending-topics).

I went through the last two weeks (18 topics) and checked the obvious exact-match .coms in each, then looked for what was still hand-reggable.

Niches with real buyers vs. likely fads
Real money: AI compute siting (data centers, grid, permitting), agent security auditing, AI code review, on-premise AI, medical image triage. Budgets are large and recurring, and several have regulation pushing them.
Probably fads or thin buyer pools: coding agent skills, codebase knowledge mapping (features, not companies), biofabricated textiles, book discovery communities, AI interface design.

Exact-match .coms: all taken
AgentAudit, AgentGuard, AgentFleet, OnPremAI, PrivateLLM, TriageAI, AIROI, LLMRank, MergeSafe, ReviewBot, PowerSiting.

The ones that are listed (Sedo asks, soft comps only)
GridSite.com: $11,895
PixelAgent.com: $11,195
UIcraft.com: $9,695
CarPrivacy.com: $5,000
SelfHostAI.com: $4,299
Asks, not sales. Still a useful ceiling for what a keyword in a hot niche gets priced at.

Coined .coms still unregistered as of today
Compute siting / energy: Kilovale.com, Ampervale.com, Wattalis.com, Resergrid.com, Gridjacent.com
AI code review: Diffsure.com, Lucidiff.com, PRSentry.com
Agent fleets / security: Hivecert.com, Queuence.com, Swarmend.com
AI search / mapping: Semantrek.com, Mnemap.com
AI ROI: Valumentum.com
Fleet / vehicle data: FleetPrivacy.com, Routicare.com
I'm not registering any of these, so take them if you like them. Expect some to go fast once posted.

What I take from it
1. In a hot AI niche the exact-match .com is gone before the trend reaches the news. Leave keywords to the aftermarket; you're buying, not hand-regging.
2. The hand-reg window is in short coined names built from the niche's vocabulary (watt, grid, diff, hive, queue). Reg them early, while the trend is still forming.
3. Energy/compute names have the widest exit: utilities, developers, and real estate buy them too, not only AI startups. "Agent" names depend on one buyer pool.
4. Renewal discipline matters more than the reg. Give each name a 2-3 year window, and drop it if there are no inquiries and the trend has cooled. Trend portfolios die from renewals, not bad regs.

Question for the room: when you hand-reg into a trend, do you go coined brandable or two-word keyword in an alt extension? And how many renewals do you give a trend name before you drop it?

 

[ElhagMousab]

I've done it both ways, and the answer changed for me the first time I watched an auction for a two-word keyword in a country code close below its registration cost.

Two-word keyword in an alt extension is the easier one to justify on paper, because you can point at search volume and say this means something. The problem is that the buyer for it is the same buyer who already looked at the .com, found it taken, and either bought it on the aftermarket or moved the brand. A cheap extension does not create a new buyer, it only gives a discount to a buyer who is already shopping. In tiny markets the effect is starker: if a country code carries around forty to sixty thousand active registrations and the meaningful keyword strings in it have been registered since the early days, you are competing for a category that mostly already has its owner. The names that trade are the short ones and the ones with local commercial meaning, not the generic English pairs.

Coined wins for me, and not because it sounds cleverer. It wins because it changes the buyer pool. A coined name built from a hot niche's vocabulary can be sold to a company that does not exist yet, which means nobody is holding a price ceiling over you. That is where the only real upside in trend registering lives. The trade is that you have to accept most of them will go nowhere, and you cannot defend a coined name with data the way you defend a keyword name.

On renewals: I give a trend name two renewals past the first year, and the second one only if something happened during the first, an inquiry, an offer, a type-in, a broker email. Two years with silence and I drop it. Trend names do not die from the price of the registration, they die from a portfolio where forty of them are quietly eating renewals while you tell yourself the trend is still early.

One thing I would add to the original post's list. When I check a niche now I look at what the niche's own vocabulary does at the short end rather than at the exact match. The exact match being taken tells you the trend is real. It tells you nothing about whether there is still a name worth owning.

I keep the numbers on what names actually cleared in a small country code, alongside the live offers, because a claimed asking price and a completed sale are two very different facts and the gap between them is where most trend portfolios are quietly losing money.

 

[accurate]

I'm not sure what you are talking about here.