Every good name is already taken. That's not an exaggeration.
In the U.S., technology trademarks fall into a single legal bucket called Class 9. Four decades ago, it held roughly 23,000 registered marks. Today it holds well over 2 million, more than four times the number of words in a full dictionary. Naming a new tech company or product in 2026 means competing against every one of those 2 million marks, plus every unregistered name already in use somewhere.
This isn't a minor inconvenience for branding teams. It's a structural scarcity problem, the same kind of scarcity that made a short, memorable .com valuable in the first place. And it's about to get worse, because the entities that need names next aren't just companies. They're AI models, AI agents acting on people's behalf, and DAOs, decentralized organizations with no single founder to file the paperwork.
Why AI agents make the naming problem bigger, not just busier
A company registers one name, maybe a handful of product names, over its lifetime. An organization running AI agents might deploy dozens or hundreds of them, each one plausibly needing its own identity: a name people or other agents can look up, verify, and trust. Multiply that by every company adopting agentic AI, and you get a naming demand that scales the way employee headcounts never did.
Trademark and naming professionals are already describing this as an inflection point, not a temporary crunch. The tools for generating candidate names have gotten dramatically better and faster. The supply of genuinely available, defensible names has not grown at all, because there are only so many short, ownable words in any language. AI can help you find a name faster. It can't invent more good names to find.
Why DAOs make this even stranger
A DAO doesn't have a founder in the traditional sense, and often doesn't have a single legal entity that can walk into a registrar's office and file paperwork. Traditional domain registration assumes exactly that: a person or company, with an address, agreeing to a contract. A DAO governed by token holders across a dozen countries doesn't fit that shape at all, which is part of why so many DAOs end up with awkward, improvised naming setups instead of a real, ownable identity.
Why the ICANN land grab of the 2010s doesn't repeat itself here
When new generic TLDs like .app, .tech, and .store rolled out through ICANN in the 2010s, it was genuinely a land grab: a wave of new namespace opened up all at once, and early movers who grabbed good names in each new TLD often did well. But that model has a hard ceiling. ICANN approves new TLDs slowly, expensively, and only ever a few at a time. There's no version of that process that can keep up with the naming demand created by millions of individual AI agents needing verifiable identities.
Handshake works differently in a way that matters here: anyone can register any TLD that isn't already taken, through an open auction, with no approval process and no waiting years for ICANN to decide a new namespace is allowed to exist. That doesn't just mean more names, it means the naming layer can actually keep pace with the rate at which new kinds of entities show up needing one.
What owning a name early actually gets you here
This isn't a pitch to speculate on TLDs the way people speculated on .com domains in 1999. It's a more specific point: an organization that's going to deploy AI agents at scale, or a DAO that needs a real identity, benefits from owning the TLD its agents and members will operate under before that naming space gets crowded the same way Class 9 trademarks did. An agent named procurement.acmecorp/ only works if acmecorp/ was claimed by the right party in the first place, and claimed on a system that can't later be revoked by a registrar deciding the name is now too valuable to keep renting out.
The scarcity that made .com valuable was never really about "the internet." It was about short, ownable, verifiable names being a genuinely limited resource. That scarcity doesn't go away because the entities needing names changed from people and companies to agents and DAOs. If anything, it gets sharper, because there are about to be a lot more of them asking for a name at once.
Written by NIHON — Handshake Infrastructure & Web3 Identity.