You're about to experience me review the top 50 DePIN coins in CoinGecko's listings right now, today. And I'm going to do it — I promised to wrap it up within 25 minutes. I let the AI do this research to minimize some of my bias in here, and this is what it came up with. Although I had to help it, because for some reason CoinGecko doesn't list ICP under decentralized physical infrastructure. Absolutely sketchy — and ridiculous that it lists Bittensor TAO under decentralized physical infrastructure. Absolutely insane what the technical facts are about TAO versus ICP. But nevertheless, we're going to go through the top 50 on here. And for some reason ChatGPT hallucinated putting Shiba Inu on the thumbnail. I don't know why it did that, but I thought it was funny, so we're just going to leave it. People will be like, what the hell is Shiba Inu doing on a DePIN thumbnail? I don't know.
We're going to go through all 50 of these. And if you have some balls, post your crypto portfolio in the Jerry Banfield Family and let us roast it.
An hour and a half of ChatGPT Pro research
I've prepared a massive research document — and by "I," I mean I prompted ChatGPT and it did it. It took an hour and a half on the pro mode to come up with all of this. You want this document? I'll drop it in the Jerry Banfield Family. We're going to stick mostly to the executive findings, because this thing is big. It's so huge I can barely take all of it. That's what she said. Oh, and it looks like I've got a tennis date set up on Monday — check out the vlogs and the dating channel for that.
This is 50-plus pages. And one of the big findings: CoinGecko's DePIN category is heavily polluted. Bittensor is an off-chain AI scoring market — this is why you should at least be using AI to help you with your research. This is how screwed up the DePIN category is. Bittensor is an off-chain AI scoring market and you're calling it DePIN. Zebec is payroll and payments. IOTA is a layer one. Chia uses disks for consensus rather than serving consumer storage. The label "DePIN" is ridiculous on here — and almost everything in crypto is ridiculous, because almost nothing has real functionality. Almost everything has been engineered with the least possible effort to make the most possible money.
Is there anything real in DePIN?
So my question is: is there anything real in DePIN? Well, ICP is real. And I disagree with some of the ratings — I even tried to get the AI to give lower ratings, but it disagreed with my prompt. It gave Filecoin a 91, which is crap, and Siacoin an 89, which is further crap, and I would penalize Render heavier. But the bottom line from this is that ICP has the realest technology, the realest DePIN infrastructure in all of crypto, by far. And the more you research it, the more you will find that. I'm going to show you that here. Now, this is not financial advice, and I hold only ICP — because of doing research like this and wanting something real, not some BS speculation.
Here's the headline: ICP is the structural outlier — replicated application execution and direct resource burn. In other words, it solves a huge real-world problem and actually puts it all on chain. Nothing else does that. Nothing else is even close to doing that. The structural outlier is what you're looking for in investing. That's all I care about investing in. That's all I care about in dating, too — I want to date an outlier, a woman who is nuts about me and for some reason loves me. You can see my exact Hinge dating profile coming out soon on Jerry Banfield Dating. Is this a crypto video, or is he trying to promote his dating channel? Well, a little of both.
The "genuine outliers" that are still too narrow
Here's the thing you need to understand: look for what's different. The research says Filecoin, Siacoin, and Arweave are genuine non-ICP outliers. I disagree — that's way too generous. The AI is too nice in this case. Because all these do is cryptographically enforce storage obligations, and cryptographically enforcing storage obligations is not the same as solving major real-world problems like data verifiability, cybercrime, and putting everything on chain so you can actually use AI to build like crazy. Cryptographically enforcing storage obligations is junk, in my opinion — and this whole video is my opinion. That's the best of this category: protocols cryptographically enforcing storage obligations, which, when you try to access the storage, can be problematic. To me that's mostly useless. And that's the best besides ICP.
Most of the remaining projects put markets, rewards, or receipts on chain while the actual physical service stays outside of consensus. That means it's not actually decentralized. You don't really own anything at all. So this whole DePIN category is absolute crap outside of ICP — and maybe, if you want to be generous about a protocol cryptographically enforcing storage obligations, you can include some of these others.
What ICP actually does differently
Again, let's look at ICP. ICP is the only general-purpose service layer that executes application state under replicated consensus. Wow, ChatGPT, you're not trying to make this simple for the average people watching my YouTube videos. I hope nobody's feelings got hurt — but when you realize how much you have to learn, it can help you learn, instead of thinking you're smart at everything. This research shows me I have lots to learn. So what does this mean? It means ICP is the only thing that puts everything on blockchain. And that's a huge deal for cybercrime. That's a huge deal for sovereignty — governments and corporations care a lot about these things. And it enables a new infrastructure where AI can build out and go crazy. Nothing else solves these huge problems.
Have some more technical language: canister update calls are deterministically executed across subnet replicas, and cycles are consumed for compute, storage, and networking. Here's the key line — this is the whole research, an hour and a half, and it comes down to this: that is a fundamentally stronger architecture than paying an off-chain GPU operator, telecom, or hotspot. I just picture all of you selling Render and Helium — dumping all the rest of these DePIN projects — because once you see what ICP can do, all the rest absolutely suck by comparison. Even the AI is having a hard time seeing how big this difference is, but it can easily see a big difference between ICP and the vast majority of the other crap on here.
Now, it says "not fully on chain does not automatically mean fake." Okay, yes — it's not totally fake. Though some of this, like Bittensor TAO being in the category, is totally fake, and how did that happen? Filecoin, Siacoin, and Arweave keep bytes on physical disks but make storage proofs part of protocol enforcement. Here's the key thing: they solve a narrower problem than ICP. If your blockchain is going to have a chance in the future, it needs to solve a big problem, because nobody's going to use it otherwise. Why would you put your stuff on Filecoin when you could put it on Amazon and have so much more access? Yes, these are more trust-minimized than ordinary cloud marketplaces — true, they're not totally garbage. But why would I want to invest in something that solves a narrower problem than ICP, when ICP is valued at around a billion dollars and solves $10 trillion problems like cybercrime? It solves problems that are 10,000 times bigger than its market cap. Why would I want something with a narrower scope than that?
Confidential compute is not the best middle ground
The research says confidential compute projects are the best middle ground. No, they're not. Look at the people at DFINITY, who are in some of the best positions in the world to know about this. Projects like iExec RLC, Acurast — that's an ugly name, Acurast, sounds vile — and Phala use trusted execution environments. Trusted execution environments are not the same and do not provide a fraction of the value of running everything directly on blockchain. They have some limited utility, but they can be expensive and they introduce a whole bunch of other problems. The research itself says they're meaningful technology, but the security root moves to chip vendors and enclave implementations. That's a no-go. Then you have to trust the chip vendors — that's not good. When you run all this, you're just moving your trust around, and you want the trust totally on the blockchain. Trusting chip vendors introduces a vector to be compromised — and we know for a fact that chip vendors have put backdoors in chips so they can be remotely accessed. How is that ever going to be decentralized?
Next one: wireless and geolocation DePIN can be real without being fully trustless. Yes — GEODNET and Helium have physical networks, but they're not really decentralized, and Helium's removal of proof of coverage materially weakens its verification.
GPU marketplaces: decentralized supply, centralized execution
Here's a big one: Render, Akash, Flux, Aethir, io.net, Golem, Nosana, Livepeer, AIOZ, and others can provide useful compute. Not really, compared to what the centralized companies can do — but if you for some reason don't want to work with a centralized company, here's the reality: these GPU and cloud marketplaces are mostly decentralized supply with centralized or unverifiable execution. In that case, most people would be better off just using centralized tech. If the token is not actually controlling any of the stuff off chain, then what? It's centralized and unverifiable. So why are we calling this DePIN? To me, Render, Akash, Flux, Aethir, io.net, Golem, Nosana, Livepeer, AIOZ, and the others have no business calling themselves decentralized physical infrastructure. They are tokens where the layer between the token and the actual network is centralized, and the execution is unverifiable. Most of the DePIN category, to me, is blatantly fraud.
And on top of this, a real product is not the same as a real token thesis. For example, Storj can sell excellent distributed storage, but the token is just a payment asset. Powerledger — and JasmyCoin is horrible and going to zero, absolute garbage. MVL, Ankr, and Zebec can build businesses, but their tokens don't have any real ownership over the business value.
What "fully on chain" honestly means
Now let's talk about fully on chain, because physical infrastructure can't literally live inside of a blockchain. But fully on chain means the application logic and the mutable state are executed by a consensus group, not by one provider — hence decentralized physical infrastructure. ICP is able to do this. You can still get screwed up by external inputs and physical node administration, but this is the highest integrity level. And the research produced this on its own — I didn't tell it to do this. This is what ChatGPT Pro's research can do, and I don't know why, if you have more than a thousand dollars in crypto, you wouldn't be doing this kind of research on your own coins instead of sitting there listening to people like me.
Level five: replicated service execution. This is the highest level of integrity in decentralized physical infrastructure — which CoinGecko apparently thinks doesn't even need to be in the category. Why bother putting ICP in there? This is how sketchy this space is. ICP is the only example — and this is what you're looking for: outliers where there is no other example. There is no other example of this happening.
Then you go down a level: on-chain contracts plus cryptographic physical storage proofs. Again, much more narrow. Filecoin, Siacoin, and Arweave do this. They have some utility, but the utility is very limited because you're not solving big problems with it. Level three: attested or challenge-verified off-chain execution — trust is reduced, not eliminated. To me it sucks. Trusted execution environments have very limited applications, and it's not a true decentralized physical infrastructure setup. So there are only about six of these that are even decent.
Then go down to level two: an on-chain market, payment, or staking around an off-chain service. This is blatantly misleading. You have things like Akash, Render, Helium, these GPU markets, GEODNET — when they call it decentralized physical infrastructure, the reality is it's an on-chain payment coin with maybe some staking, and everything else is off chain and not actually decentralized. So ridiculous. And then it gets even worse when there's just a token wrapper: IOTA, Zebec, Chia, Radworks, and many others — anything I haven't mentioned is almost definitely in level two or level one. Just absolute trash that people have spun up to steal money from you. Crypto is basically organized crime, in case you hadn't figured that out — and you need ICP to sift through it. What's not organized crime in here? ICP. What is organized crime? Basically everything else: people conspiring to take money from you on something they know is not valuable. Sorry if you didn't know that already, but I've known it for a while.
ICP still has some limitations. Its physical nodes are standardized machines in data centers, and they have to be admitted through the NNS. And the non-replicated — I don't know, this is too technical for y'all. Basically, ICP is not perfect. But it's the best we have, by far. The methodology scored on-chain control and execution, verifiable physical service, and provider decentralization, with hard penalty rules it came up with on its own — I suggested that fully on-chain real execution is pretty important, but it came up with all of this itself. TEEs earned substantial verification credit, but never the same credit as replicated execution or protocol-native cryptographic proofs.
Reading out all 51 scores, from the bottom
So let's scroll down and look at the full list of ranked results — all 51. Technically it's 50 plus ICP, because I had to manually add it, since the people at CoinGecko either don't know what they're doing or there's something dishonest going on. Which do you think it is? That they just don't realize the highest level of decentralized physical infrastructure isn't in the category, or something dishonest? I don't know which one it is. I can guess, but I'll let you guess.
Let's look at the bottom. DexNet, F. XPIN Network. Spacecoin. Radworks, D. All these are D's and F's down here. Holo — HOT — D. Zebec Network — so many of y'all have been excited about Zebec Network; look at the reality of this. JasmyCoin — for God's sakes, if anyone ever likes Jasmy again... I will never read another Jasmy comment on YouTube, that's for sure. I'm not reading comments on YouTube anymore. Some of them are very loving and supportive, others are insane, and the insane ones bother me more than the nice ones help. If you're not paying me, I don't see why I'd pay attention to you — can we just be realistic and honest about that? I'm doing all this research and putting it out here for free. If you want me to hear what you have to say, I've got a perfect place set up. So I'm glad I'm never seeing another Jasmy comment on YouTube. Jasmy is such garbage — I've told you it's been garbage the entire time.
MVL, D. Chia, D. Janction — what kind of name is that? Beldex — more nasty names. IOTA — I sound like I'm swearing in some alien language. I'm looking at the camera right now — bro's looking good, look at my face all shiny. No wonder I'm getting dates.
Bittensor TAO gets a D
Bittensor TAO: a D. A project where the founder goes around telling people there have been three big inventions in the last 30 years: one, the internet; two, Bitcoin; three, Bittensor. You've got to be kidding me. This is a nothing burger. It is a blockchain-made ranking market for off-chain AI inputs. It is a joke for people who think it's funny to get ripped off and buy garbage coins. Was that too harsh? I'm going for humor here — but really, I can't stand Bittensor TAO, because it is crap. It shouldn't be in this category. Either CoinGecko has no idea what they're doing or somebody got bribed to put it in there. I don't know which it is. You know which I'd guess.
The C tier: real products, weak decentralization
BitTorrent, C. Dolphin — POD, that's decent branding — C. Powerledger. Ankr Network. Ocean Protocol. The Graph. Starpower. Grass. Theta Network. Impossible Cloud Network — junk. Nosana. io.net. Bless. Storj. Golem. XYO Network. Aethir. Auki. Theta Fuel — this is literally a fuel coin, a second coin for Theta; come on, man, it shouldn't even be in this category. StorX. peaq. AIOZ Network — garbage; I've ripped on AIOZ Network the whole time. And look what the research says about Theta: a broad, real edge network stack, but its chain mainly coordinates nodes and payments while content and computation remain off chain. How is that decentralized physical infrastructure?
The AI is clearly able to cut through the BS going on here. Even though the AI is not quite as smart as I am in this respect — it doesn't have as much context — the AI is getting this smart: it could pick ICP out of all of this, and that's impressive, because most human beings in crypto cannot. The AI can because it reads actual documentation. It doesn't just take some influencer's word for it and buy the coin. That's why this research is powerful — it took an hour and a half for the AI to read everything it needed.
Livepeer, C. World Mobile Token, C. And I'd put all these A's and B's probably down to C's myself, but I can't be bothered to micromanage ChatGPT and tell it to do things my way, because it's less honest that way. This is what it came up with from a five-or-ten-minute prompt.
The A and B tier — still too narrow
Then we continue up: Flux. Render. Akash Network. OriginTrail — that is good branding, but again, a verifiable data infrastructure with a real token loop that only loosely fits the physical infrastructure category. Why don't we have ICP on CoinGecko's list, but we have stuff like OriginTrail that loosely fits the category — the AI itself telling you it loosely fits? IoTeX. Helium. GEODNET. Phala. Acurast. iExec RLC. Arweave. Siacoin. And Filecoin.
So Filecoin, Siacoin, and Arweave are a little bit decent, but they're too narrow. They may last longer and have a lower bottom than some of these other coins. Siacoin is technically a coherent storage DePIN, where renters and hosts use on-chain contracts, collateral, and storage proofs, and SC pays for the service — that's better than most of the other setups. Now look at Filecoin: the research calls it the strongest non-ICP DePIN in the set — storage deals, collateral, and ongoing proofs enforced by the chain. Here's the problem, though: the bytes live on physical disks, not on chain, and that introduces problems. What happens if some of the physical disks get swapped out? What happens when you need to access the data, and the storage is mainly set up to prove it's there, not to facilitate accessing it? That's why ICP is so incredible — the whole thing is set up to maximize being able to access and use all the data, whereas you're not going to build huge things with AI or enterprise or governments on Filecoin, Siacoin, or Arweave.
And the rest of these — dang, I can't say that. It's probably against the terms and conditions. I'm like, these start to suck more than some of the videos I've seen lately. If you want to know exactly what I was going to say, ask me and the Jerry Banfield Family what that joke was going to be, and you can hear about it.
Market cap versus reality
Now look at the table based on market cap and ratings as we wrap up. The top coin in the category by market cap gets a 43 from the AI — a blockchain-paid ranking market. Shove this in the founder's face: what you've built is a blockchain-paid ranking market for off-chain AI inputs, not a verifiable physical infrastructure network. And shove this in the face of the people at CoinGecko: the one that's not even on there is the best one. Render gets a 61 — I guess that's B tier in this scenario; I remember getting B's in school, so we'll give Render a B. Beldex, another $651 million: a real privacy-node network, but its physical-service claims are secondary to a privacy currency and largely unverifiable at the service layer. That sounds pretty sketchy to me — a D. Filecoin has a decent-sized market cap — but for double that market cap, you could get something absolutely incredible called ICP.
Someday the price will reflect it
And someday, the price is going to reflect it. Someday the price is going to reflect it — either that, or we're going to zero and something else will have to solve the same problem. I doubt ICP is going down. Then look at the rest of these: BitTorrent, 45. Jasmy, 37. Grass, 49. Zebec, 37. Arweave down here does stand out a bit for the market cap and what it actually does. For God's sakes, please do some research like this yourself — the document will be in there with the deep details the AI went through and exactly how it scored everything.
And the 11th row is Bittensor TAO one more time: the hardware does the work off chain, validators grade their friends' homework, and the blockchain proves the grades were paid. This is what the AI came up with — the AI is making jokes about how bad Bittensor TAO is. And y'all keep telling me ICP and TAO are the future. The AI doesn't agree with you. You make me laugh. You can watch every one of these coin-by-coin breakdowns as they come out on my ICP Crypto playlist.
That's 25 minutes, y'all. Get into the Jerry Banfield Family — come on, man, get in there. We're having new people join every day, and we're going to blow this community up.
Now's the best time to get in while it's small. That's what she said.