Three DePIN pitches in 72 hours. One telco-side founder out of Tel Aviv. Then a compute-side founder out of London. And a sensor-network founder out of Singapore on the way home. Different sub-sectors, different stages, different decks. Hardware bills of materials weren't even in the same league. I want to write down what they had in common because I almost missed it on the first call and it changed how I evaluated the second two.
The Telco Pitch
The first one was a 5G off-load play, operator-grade radios, real telco partnerships, real spectrum work. The founder had spent a decade at one of the big European carriers before walking away to build it. The deck led with the tokenomics. That's where my brain almost shut off.
DePIN deck convention in 2026 is to lead with tokenomics, the supply curve, the emissions schedule, the staking design, the burn mechanism. I get it. It's the part founders feel most uncertain about and most need outside validation on. But for the kinds of DePIN projects that actually matter, the ones building infrastructure that has to compete with billion-dollar incumbents on unit economics, the tokenomics is the least interesting page.
I asked him what his node payback period looked like in dollar terms without the token. He paused. Then he answered honestly: 14 months. I asked what the corresponding number was for an equivalent slot of carrier-grade small-cell capex on the incumbent stack. He said about 28 months. So that conversation, not the tokenomics page, was the underwriting moment.
The Compute Pitch
The second one was a decentralized GPU-inference play, in a category that's now genuinely crowded. io.net, Aethir, Render, Akash, the four-horse race in compute DePIN has all the dynamics of an L2 war from 2023. Lots of capital, lots of teams, real demand starting to show up on the customer side.
This founder's pitch was structurally different from the first. The deck didn't lead with tokenomics. It led with a customer logo: a real, named AI lab, not one of the frontier-three, but a credible US-based applied-AI team, running a non-trivial portion of their inference burn on his network at unit economics that I could check independently. The cost per inference-hour was 38% below what they'd been paying their hyperscaler. I asked what was different about his network versus the other three. The answer wasn't speed. It wasn't even price. It was scheduling, the way his orchestration layer batched inference jobs against available GPU capacity. Boring. Plumbing. The thing that wins.
The Sensor Pitch
The third was a sensor network — environmental monitoring nodes, with a real customer ask coming from carbon-credit verifiers. Singapore-based team, Dubai-funded angel round, planning to operate primarily in Southeast Asia and the Middle East. (This is the cohort I keep finding myself spending more time with, a story for another Bridge Note.)
The pitch was almost the inverse of the first two. The unit economics weren't there yet. The customer wasn't a Fortune 500. The node hardware bill of materials was higher than the comparable Hivemapper reference for a different sensor class. What was different was the regulatory wedge — the founder had real, signed memoranda with two SEA national environmental agencies that mandate-required this kind of independently-verified sensor data starting in late 2026. The token economics, the network growth model, the hardware unit economics — all of those could be reverse-engineered from the regulatory tailwind, which was the actual moat.
What They Had in Common
Three completely different DePIN sub-sectors. What unified them, looking back at my notes, was that none of the three winning pitch moments came from the tokenomics page. The telco win came from a competitive payback-period delta. The compute win came from a customer-disclosed unit-economics gap and an orchestration insight. The sensor win came from a regulatory-mandate wedge that pulled the unit economics forward.
The token was downstream of the actual product economics in every case. That sounds obvious when I write it. It is not how DePIN is conventionally pitched in 2026. It's not how I conventionally evaluated DePIN in 2024. I'd been weighting the tokenomics page more than it deserved — and underweighting the dollar-terms-without-the-token competitive comparison, which is the one that actually matters for whether the network gets built.
What I'm Doing Differently Next Week
Three calls already booked. Two more I'd been planning to take. For all five, the first question I'm now asking, before the deck, before the demo, before the tokenomics, is: what is your node-economics or customer-economics story in dollar terms, with the token taken out of the picture? If the founder can answer crisply, the rest of the pitch tends to compose. If the founder can't, the tokenomics page is doing work it shouldn't have to do.
I'd genuinely been making this mistake. I'll be making it less.
The next Field Notes will go shorter and faster, a conversation from this week that reframed how I think about cross-chain bridges. If you're a DePIN founder reading this and want to push back on the framing, ping me on LinkedIn or Telegram. The framings I get pushed on are the ones that hold up.