Crypto & DePIN

“Having to own a piece of… [an] actual model and getting royalties when people use it, I think, is going to be a big shared ownership model the world has never seen before.” — Greg Osuri, 00:40:34

“A person of his nature and stature helping Akash design economics is just the tip of the iceberg… his proposal essentially: let Akash borrow security from Bitcoin — that way we don’t need to allocate that many tokens for security, but rather actually borrow from a better security, from a proof-of-work model.” — Greg Osuri, 00:11:00

“There is a chance the studio will go so big that they determine the protocol completely… I think there’s a heavy risk now that it may lose its decentralization nature… and if that model is going to be exported for the rest of the crypto ecosystem — which most likely it looks like it is — then we’re in big trouble.” — Greg Osuri, 00:41:24

“We’re doing DeFi rails… we can decouple the ownership of the asset from the earnings… sell the ownership of the asset to someone that wants depreciation… crypto has got incredible primitives — the DePIN primitive and DeFi primitive, when you bring them together you can create amazing financial assets… you can purchase an asset that will give you like 15% return.” — Greg Osuri, 00:26:40

“The only way we will be able to get chips is through a peer-to-peer marketplace, and that’s really what Akash is building.” — Greg Osuri, 00:24:24

“Starbonds are designed to reward early participants up to 6.4 times return on investment in high demand scenarios, only 25% principal at risk in low demand cases. The internal rate of return increases with early tier discounts, from 36.7% to 44.8%… This isn’t speculative crypto. These are productive digital assets tied to real-world AI infrastructure powering actual AI workloads.” — Greg Osuri, 04:37:07

“Akash is permissionless, decentralized. It runs fully on-chain — every transaction, every lease happens on-chain — whereas Render is closed source and off-chain. [The] only thing that happens on-chain is payments and nothing else… one is permissionless, another one is not. And we emphasize quite heav[ily] on self-custody.” — Greg Osuri, 00:14:54

“We introduce the Starbonds, a digital asset that represents fractional ownership of compute… You reserve a Starbond. A node keeper provisions and operates the hardware. You receive a share of the revenue generated by the node. Transparent, on-chain, and aligned from day one. This isn’t just capital formation. It’s capital democratization.” — Greg Osuri, 00:10:34

“How do we fund and govern a cloud with no single entity controls? With Star Bonds, a new model of fractional on-chain ownership for compute infrastructure… Anyone can reserve a star bond, a node keeper provisions and operates the star node, AI workloads run, revenue is earned, payouts flow back to holders. It’s transparent, permissionless, aligned from day one.” — Unknown, 00:00:01

“Starbonds are designed to reward early participants up to 6.4 times return on investment, or ROI, in high demand scenarios; only 25% principal at risk in low demand cases… And this isn’t speculative crypto. These are productive digital assets tied to real-world AI infrastructure powering actual AI workloads.” — Greg Osuri, 00:13:02

“Each Starbond is an NFT that represents ownership in a specific node. When you mint one, you’re backing the deployment of a real-world rack by a node keeper in a real facility serving real AI workloads. In return, you earn a pro-rata share of the node’s revenue automatically on chain without intermediaries. It’s like staking, but instead of securing a blockchain, you’re powering intelligence.” — Greg Osuri, 00:12:49

“Star Bonds are designed to reward early participants — up to 6.4 times return on investment in high demand scenarios, only 25% principal at risk in low demand cases. The internal rate of return increases with early tier discounts, from 36.7% to 44.8%… First issuance Star Cluster Alpha will release about 75,000 bonds, each with a par value of $1,000.” — Unknown, 00:00:45

“The business model for distributed training — almost all of the models I’ve looked at — the common thing they have is you contribute compute, you get some representation of that contribution in form of a token, and that token will guarantee you future revenues from the model, from inference.” — Greg Osuri, 01:06:10

“I think there’s enormous money to be made in the next one to two years in the decentralized space… there’s zero hype on these protocols… [Gensyn] alone is right now training about 10,000 models concurrently. 10,000 models. Name a single AI lab that can do that.” — Greg Osuri, 01:08:18

“DePIN shines when it mirrors its own design — like countless small parts, like trees in a forest or neurons in a brain… fragmented assets like GPUs or rooftop solar or Wi-Fi routers or city sensors… the line is simple: decentralize what is abundant and idle, and centralize what is scarce and indivisible.” — Greg Osuri, 00:14:42

“Tokens are rocket fuel: they help you launch and bootstrap a network, but once you’re in orbit you have to run without the fuel, or like very less fuel… somehow we got to orbit using hot air balloons instead of rocket fuel.” — Greg Osuri, 00:31:23

“If you remove all the bells and whistles of decentralization, from a technology standpoint it’s just a really bad database — that’s what it is, end of the day. But decentralization on top of it makes it extremely powerful. It’s the only database where you do not need permission to write… In SQL you need permission to write, you need a password, username; in blockchains you don’t need permission to write. That’s the most attractive part of a blockchain, but that comes with a lot of trade-offs.” — Greg Osuri, 00:05:00

“Only way to offset this risk of unknown cost is have massive amounts of transactions being written. Well, it’s very unlikely to have that many amount of transactions for a non-financial product… you need a lot of volume to be profitable. So economics at a certain stage becomes unsustainable without the volume… To summarize: crypto-economic value was not justifiable in terms of security, and second, validator profitability cannot be sustained with the current economics… and the market conditions don’t seem to be that favorable considering what’s happening geopolitically.” — Greg Osuri, 00:07:06

“We’re looking at potentially giving direct exposure for stakers into the compute itself, instead of just security right now — because the stakers are staking for security but they’re not directly engaging with the AI compute, which is the bigger sort of opportunity… I’m particularly excited for staking on Akash evolving to a new paradigm.” — Greg Osuri, 00:23:41

“I think the Cosmos model in general has failed — this is coming from someone who’s been in the space for a long time, Akash being one of the first Cosmos chains. The model where it failed is sovereignty before product market fit. It should be flipped… You don’t go build your own data center when you’re shipping an app.” — Greg Osuri, 00:29:37