Hello friends. Hope you've all been well. I want to share a thesis I've been drafting, one I've been asked about repeatedly. An asset that, in my view, remains profoundly mispriced relative to what it actually is.
This asset sits at the intersection of two sectors I'm incredibly bullish on: AI and privacy. But to truly understand it, we need to talk about where we are in the world right now.
AI is no longer a novelty. It's now infrastructure. And like all infrastructure, it's quietly being absorbed into the existing power structure. Every major AI provider is one glowie love letter away from logging your prompts. OpenAI already does this. Anthropic will if the government asks. Google doesn't even need to be asked, they're the government's favorite contractor for a reason. The EU has already formalized surveillance norms through legislation (passage of EU AI Act). China has never pretended otherwise, and it's only a matter of time until the US drafts legislation requiring AI providers to flag "dangerous" queries. The definition of "dangerous" will remain deliberately ambiguous. That ambiguity is a feature, not a bug.
AI and LLMs are obviously here to stay. But our privacy, in the conversations we have and the prompts we engineer with our models, is eroding rapidly, right before our eyes.
And since I know some jackass is gonna mention it in my comments, I'll address it now. Yes, you can host a local LLM if you're concerned about privacy. But this isn't practical for the vast majority of people. It's inconvenient. And inconvenience is the enemy of adoption. Every system that wins does so by being easier than the alternative. If you want to anonymously pay someone, you can mail cash with a fake return address, or you could just use Monero like everyone else. Local LLMs are important. They should exist. But they aren't the solution for the masses.
This brings me to the asset.
VVV, the token that powers the Venice ecosystem. Think of Venice as uncensored, private AI inference. Like Mullvad and your favorite language model had a beautiful baby together. It's private by default. There are no logs. No conversation data recorded, and chat history is preserved through local storage only.
Most crypto tokens are vaporware searching for a purpose. VVV is the opposite. It's a finite resource with a token attached. VVV represents a direct claim on something scarce and physical, the right to anonymous inference. If you stake VVV, you reserve a share of Venice's total daily inference, proportional to that of your token share. There's no subscription, billing relationship, or approval needed. All you need is to stake VVV and that entitles you to your god-given right to privately infer.
And here's where it gets interesting. By holding VVV, you also experience positive optionality through Jevons' paradox.
If inference becomes less expensive, your prompts get cheaper and as such your reserved inference allocation produces more output. On the other end, let's say for some reason global compute becomes more expensive, your finite compute allocation appreciates in dollars, and thus puts direct buying pressure on VVV. No matter which way you look at it, every directional outcome strengthens the position of whoever already holds VVV. This is Jevons' paradox essentially applied to thinking itself.
Last year Venice introduced DIEM. The idea is you stake VVV, mint DIEM, and receive a perpetual share of inference credits. It's also important to note that over 70% of the circulating supply of VVV is staked. In crypto, conviction is revealed through mobility. Who stays when they can leave. Most VVV holders aren't speculating but instead voluntarily choosing to lock their tokens to use the product.
Some tokenomics also worth noting: VVV currently inflates at around 6 million tokens per year through emissions. But since Q4 last year, Venice has been running a buy-and-burn campaign by punting protocol revenue into open market VVV purchases, and then burning it. At current levels, the burns offset about a quarter of annual emissions. The token is still net inflationary today, but the gap is closing. If usage grows and revenue scales, deflation becomes possible. That's the flywheel: more users, more revenue, bigger burns, tighter supply.
And then there's the agent angle. Autonomous agents, like OpenClaw, have recently been all the rage on twitter. And if you're in the camp that believes autonomous agents are not only here to stay but will grow exponentially, then you might understand what I'm implying here.
For those who don't, let me explain.
Autonomous agents cannot open bank accounts. They cannot pass KYC. They cannot obtain credit cards. But they can hold tokens. They can stake them. And through Venice, they can consume inference permissionlessly and autonomously, as long as you have enough staked VVV to provide that (sometimes insane) amount of demand. So if agents become persistent economic actors, and all evidence suggests they will, they will require infrastructure that doesn't hold them back with KYC and credit card constraints. Venice is one of the few protocols uniquely designed and positioned for this market need.
Switching over to risks, the primary one here is concentration. Voorhees controls this, and by staking VVV, you're trusting him, and solely him, to deliver. There is no decentralization, no DAO, no committee. You're betting on a competent founder, but at the end of the day he is still a person. And people make mistakes. But you're also no longer paying for a hypothetical. You're buying a live system with real users and sticky revenue.
I'm quite convinced that the agent-driven demand boom is not priced in, and this is one of the few assets in this space I'll probably hold with duration. Not because it's guaranteed to win. Nothing in life is. But because the payoff distribution is so asymmetric and because the upside potential compounds with every additional agent that comes online and needs somewhere to think.