Infrastructure telemetry is the scarcest and most valuable DePIN data — the actual measurement of nodes, GPUs, and storage. Only 296 of 800 cataloged networks have it, which is exactly why it's a moat.
The DePIN data map has a hard edge. Of 800 cataloged networks, only 327 carry live data, and only 296 have first-party telemetry. Here is where data exists, where it's missing, and why naming the gaps is part of the service.
DePIN analytics is the discipline of turning raw chain data and telemetry into investment decisions. Here is the pipeline: collect from 69 sources, normalize 327 networks into canonical metrics, and anchor every number to its source.
DePIN telemetry is the raw measurement of physical infrastructure: active nodes, deployed compute, real storage. Here is how we collect first-party supply data from 296 networks across 69 sources, normalize it, and prove it hasn't been tampered with.
DePIN intelligence is more than market caps and token prices. It is verifiable telemetry from physical infrastructure — compute, storage, bandwidth, sensors — normalized into one canonical schema. Here is how we read 327 networks from 69 sources and anchor every number to its source.
Most DePIN tokens have no underlying revenue. Protocol fees — real economic activity read from chain state — separate the networks that earn from the ones that just hold a token. Here is how we track them.
DePIN infrastructure telemetry is a rich and largely untapped AI training signal. With 10,706 live series across 327 networks, here is what you can build — and why verifiable data beats scraped data for training.
One public URL, one second, four defensible node counts. Storj's us1 satellite reports 123,580 total nodes but only 32,796 active — a 3.8x gap that is entirely dead history. Which number is 'the node count' depends on which denominator you accept, and a third trap (summing satellites) overstates the fleet ~4x. Every number here is re-checkable from the same public endpoint.
437 military aircraft airborne, 122 active FAA flight restrictions, 926 map cells at HIGH GPS interference, 4,148 ships on national AIS feeds, and Kuiper at 391 tracked objects — all measured today, all landing in one table, every row carrying its source. Here's the data, the method, and the caveats.
We now poll all 129 Bittensor subnets directly from the chain: neurons, active neurons, alpha pools, TAO reserves, emission, burn cost, tempo, and more — 13 metrics per subnet. First findings: 118 of 129 subnets are at their registration cap, yet seven of them have two or fewer neurons actually working, and one subnet alone draws 13.9% of tracked TAO-in emission.
First-party DHT measurement: our Nebula-based crawler sampled 6,416 IPFS DHT server peers (53.4% dialable, 257 distinct agent versions), and our BEP-5 sampler estimates the BitTorrent Mainline DHT at ~2.16M nodes. No aggregators — we joined the networks and counted. Method, numbers, and every caveat.
On August 9 our daily snapshot showed 127 of 423 Akash GPUs leased — 30.0%. Today Akash's own API reports 212 of 440 active — 48.2%. The denominator barely moved. This is a demand ramp, and here is exactly how to check it yourself.