#opg $OPG First time using OpenGradient Chat, I deliberately slipped in a real DeFi strategy draft—this wasn’t a test. I was betting it would actually hold up under scrutiny. The moment I hit return, I wasn’t watching how fast the answer came. I was watching whether this prompt had been seen by any intermediaries.
The result didn’t disappoint me. The TEE locks the conversation flow, and the Nova network finishes with on-chain verification—cloud providers, node operators, even the team itself can’t peek at the physical layer. But after my tests, I became more worried about something else: Intel.
The HACA architecture fully separates execution from verification. In traditional blockchains trying to do AI, there’s a deadlock: 100 verification nodes each run a 70B-parameter LLM once—costs multiply by 100, and the results still might not match. HACA’s solution is this: inference nodes run the model inside the TEE and generate hardware attestations; verification nodes only validate the attestations’ validity, without needing to know what the prompt and outputs actually are from start to finish. Verification only needs to take place at millisecond level. After more than two months on mainnet—hosting over 2,000 models and handling more than 2 million inferences—a16z Crypto and Coinbase Ventures led a $9.5 million investment.
But TEE security is completely tied to Intel/AMD microcode. In TEE.Fail side-channel attacks disclosed in October 2025, the cost is under $1,000 to steal keys from SGX, TDX, and SEV-SNP. Nova guarantees that the “execution process hasn’t been tampered with,” but it can’t guarantee that the “execution environment was clean from the start.”
Now look at the token: total supply is 1 billion, and the current circulating supply is only 190 million. In the ecosystem, 40% linearly unlocks over 60 months. As of June 16, OPG is $0.1618, with a market cap around $30.74 million, and it’s down 47% over the last 24 hours. That volatility suggests the circulating float is too thin and the tokens are highly concentrated.
My strategy is simple: take an extremely low position as an observation stake. Watch the TEE verification success rate and the node block-production stability. Only consider adding more after both have run stably for more than three months. Before that, anyone who yells “All in” gets an immediate argument from me. There are too many people telling stories in this market, and far too few who can turn stories into reality @OpenGradient
The result didn’t disappoint me. The TEE locks the conversation flow, and the Nova network finishes with on-chain verification—cloud providers, node operators, even the team itself can’t peek at the physical layer. But after my tests, I became more worried about something else: Intel.
The HACA architecture fully separates execution from verification. In traditional blockchains trying to do AI, there’s a deadlock: 100 verification nodes each run a 70B-parameter LLM once—costs multiply by 100, and the results still might not match. HACA’s solution is this: inference nodes run the model inside the TEE and generate hardware attestations; verification nodes only validate the attestations’ validity, without needing to know what the prompt and outputs actually are from start to finish. Verification only needs to take place at millisecond level. After more than two months on mainnet—hosting over 2,000 models and handling more than 2 million inferences—a16z Crypto and Coinbase Ventures led a $9.5 million investment.
But TEE security is completely tied to Intel/AMD microcode. In TEE.Fail side-channel attacks disclosed in October 2025, the cost is under $1,000 to steal keys from SGX, TDX, and SEV-SNP. Nova guarantees that the “execution process hasn’t been tampered with,” but it can’t guarantee that the “execution environment was clean from the start.”
Now look at the token: total supply is 1 billion, and the current circulating supply is only 190 million. In the ecosystem, 40% linearly unlocks over 60 months. As of June 16, OPG is $0.1618, with a market cap around $30.74 million, and it’s down 47% over the last 24 hours. That volatility suggests the circulating float is too thin and the tokens are highly concentrated.
My strategy is simple: take an extremely low position as an observation stake. Watch the TEE verification success rate and the node block-production stability. Only consider adding more after both have run stably for more than three months. Before that, anyone who yells “All in” gets an immediate argument from me. There are too many people telling stories in this market, and far too few who can turn stories into reality @OpenGradient