Crypto is in a quiet stretch. Prices are sideways, attention has thinned, and most onchain projects are waiting for the next catalyst to arrive. Meanwhile artificial intelligence is moving faster than at any point in its history. New models, new agents, and new infrastructure ship every single week. The gap between those two realities is wide, and it is widening.
That split is the most interesting thing about AI4 right now. While the broader market consolidates, the AI namespace is being claimed in real time. As of June 2026, seven hundred sixty four names have been minted on the protocol. That number is worth stating, but on its own it tells you very little. A count is just a count. The real information is in the composition. When you take the full registry and sort it by what is actually being registered, a clear and surprisingly coherent picture emerges, and that picture tracks almost exactly where the energy in artificial intelligence is going.
This piece is a look at that composition. Not the price, not the hype, not the count for its own sake. What people are choosing to name, and what those choices tell us about the kind of namespace AI4 is becoming.
The substance layer: foundational vocabulary
The largest single block in the registry is general vocabulary. Foundational keywords like health, finance, energy, security, materials, manufacturing, industry, and infrastructure. Roughly half of all the names minted fall into this category. These are not lottery tickets. They are the kind of names that get claimed when participants treat a namespace as territory worth holding rather than a string to flip for a quick markup.
This matters more than it might appear at first. In the early life of any naming system, the ratio of substance names to speculative names is one of the clearest tells about where the thing is headed. A namespace dominated by random character strings and meme registrations is a casino. A namespace dominated by foundational vocabulary is being treated as infrastructure. AI4 is firmly in the second category, and it is not close.
The signal layer: the vocabulary of modern AI
The second story is the one that deserves the most attention, because it is the one that connects AI4 directly to the technology cycle driving everything right now.
A substantial cluster of names maps directly onto the working vocabulary of modern artificial intelligence. These are not generic tech words. They are the specific terms that anyone building in the field today would recognize immediately. Inference. Agentic. Multiagent. Swarms. Subnet and subnets. Protocol level registrations like mcp and webmcp, pointing at the model context standards that have become central to how agents actually operate. Even x402, which maps onto the emerging standard for agent to agent payments, a concept that did not meaningfully exist eighteen months ago.
Go deeper and the pattern holds. Alignment. Guardrail. Aioversight. Superintelligence and asi. Sentient. Worldmodel. Quantization. Reasoningagent, deepagent, and coagent. Selfevolving. Recursive. Physicalai and digitaltwin. These are the words that fill the actual research papers, the actual product launches, and the actual arguments happening across the field right now. Somebody claimed each one of them on AI4.
Read together, this is not a random scatter of registrations. It is a map of where artificial intelligence is heading, drawn by the people who are paying close enough attention to stake out the territory early. That is exactly the kind of forward looking conviction you want to see in a namespace this young.
The frontier layer: the players defining the field
Alongside the vocabulary sits a third cluster: the names of the companies and systems defining artificial intelligence today. The registry includes names matching model labs and AI products such as mistral, midjourney, elevenlabs, synthesia, qwen, nemotron, cursor, glean, mercor, lovable, twelvelabs, and openrouter. It includes the hardware and chip layer underneath all of it, with names matching tsmc, asml, arm, furiosa, rapidus, and loongson. And it includes the broader frontier, from anduril and helsing on the defense side to extropic and dwavequantum on the experimental compute side.
There is a robotics and embodied AI cluster worth calling out on its own, because it is one of the fastest moving corners of the field. The registry holds humanoid in three language variants, along with names matching figure, apptronik, unitree, doosanrobotics, hanwharobotics, engineeredarts, and the broader cobotics and android vocabulary. Embodied AI is where a great deal of the next phase of attention is going, and the namespace already reflects that.
The honest framing here is that a registration is a claim on a name, not proof of who made it. What the cluster shows is not corporate endorsement. It shows that the people minting on AI4 are thinking about the real players and the real frontier of the field, and choosing to stake out that ground. That is a signal about the seriousness of the participants, which is the thing that actually matters at this stage.
The structural layer: short, premium, and numeric
Every mature namespace develops a class of names that hold value purely because of their form rather than their meaning. AI4 already has one.
There are sixteen two character names in the registry, including io, id, mr, an, re, si, and the structurally interesting 0x and 1x. There are thirty nine three character names, among them mcp, llm, rwa, asi, sol, web, dns, key, and ai4 itself. In naming systems, short names are the equivalent of prime corner lots. They are finite, they are memorable, and they tend to appreciate as the namespace fills in around them.
There is also a numeric set, running from clean repeating sequences like 8888, 7777, and 1111 to ascending runs like 123456789 and 1234567890. Numeric names carry cultural and collectible value in many naming contexts, and their presence is a normal and healthy feature of a maturing registry rather than a sign of froth.
The long tail: personal, geographic, and speculative
The rest of the registry rounds out the picture in the way any real namespace does. There is a band of personal and vanity registrations, the names people claim for themselves or their projects. There is a geographic set that is notably international in character, with strong representation from India in particular through hyderabad, bengaluru, vizag, vijayawada, amaravati, and bharat, alongside deutschland, liechtenstein, emirates, and others. That international spread is itself a small but real signal that the namespace is not confined to one local community.
And yes, there is a speculative and meme tail. There always is. The important point is its proportion. It is a minority of the registry, not the core of it. The weight of AI4 sits in substance, signal, and structure, with the speculative layer riding on top rather than holding up the whole thing.
Why the timing matters
A naming protocol is only ever as meaningful as what people choose to name. Anyone can mint a string of characters. What tells you whether a namespace is becoming real is whether the names being claimed reflect genuine conviction about the future, or just noise chasing momentum.
On AI4, the names being claimed are the names that matter in artificial intelligence right now. That is the entire thesis in one sentence. The protocol is not riding a price cycle. It is tracking a technology cycle, and that technology cycle happens to be the strongest one in the entire market today.
The crypto lull, far from being a problem, is actually a clarifying force. A flat market strips out the participants who only show up for momentum and leaves the ones who are present because they believe the underlying idea is worth building on. A permanent, onchain naming layer for artificial intelligence agents is precisely the kind of idea that does not require a bull market to make sense. Agents need persistent identity. That requirement does not pause when prices do. If anything, the need grows more urgent every week, regardless of what any chart is doing.
The forward read
Seven hundred sixty four is an early number, and that is the entire point. The composition of the registry at this stage is far more informative than the raw count, because it reveals what kind of namespace this is becoming before any hype cycle has had the chance to distort it. What it reveals is a namespace being built around the genuine vocabulary and the genuine players of artificial intelligence, with a strong foundation of substance names underneath and a healthy structural layer of short and numeric names on top.
The work ahead is to deepen exactly that. More builder oriented registrations. More protocol level names. More of the people actually shipping agents choosing to claim their identity on AI4 rather than somewhere temporary. The market will turn eventually, as it always does. When it turns, the namespaces that mean something will be the ones that quietly built real substance during the quiet stretch, while everyone else was waiting for permission from a price chart.
That is the work in front of us. And in this particular moment, with crypto resting and artificial intelligence accelerating, it is exactly the right work to be doing.
What names would you want to see claimed before the next cycle begins?
