Venice AI raises $65M at $1B, tops $70M ARR on privacy pitch

The interesting part of Venice AI's $65 million Series A isn't the unicorn label, it's that a small consumer AI company built explicitly around 'don't ask, don't store' has quietly grown into a profitable business. According to TechCrunch, the startup is running over $70 million in annualized revenue, serves 3 million active users, and handles about 1.7 million API calls a day. That is not a rounding error on the way to a bigger AI player. It is a category the mainstream vendors are choosing not to serve.

The pitch is architectural, not just brand. User data is encrypted client-side, routed through external proxies, and never stored on Venice's servers, with end-to-end encryption sitting behind a paid tier. Venice hosts 'uncensored' open-source models on its own infrastructure and passes queries to closed-source options from OpenAI and Anthropic when users want them. The founder and CEO is Erik Voorhees, who previously built ShapeShift and Satoshi Dice, and his framing is squarely libertarian: he says the company is 'optimizing for freedom and actually respecting users as adults, which is, I think, rare these days.'

Why this matters if you are building or buying AI tools: the trust surface of hosted AI has become a real product axis. If a small team can reach $70M ARR by positioning as 'we can't hand over what we never had,' that is a working template for users who have decided the mainstream assistants are too cautious or too watchful. It also means the 'uncensored' category has meaningful revenue attached to it now, which changes the political dynamics for anyone arguing safety-first as the only viable posture.

The honest caveats are load-bearing. The revenue and user figures come from the company via a single article and are not audited in what is reported, so take the specifics as reported rather than settled. Only about 8% of users pay in crypto, so the two token programs, VVV and DIEM, are more marketing than economic engine today. And routing higher-end queries to OpenAI and Anthropic means Venice's differentiation lives one policy change away from the very providers it positions against. What the reporting doesn't give you is churn, unit economics on the 1.7 million daily API calls, or the securities posture of the token program.

The upside worth watching is whether Voorhees actually converts this round into owned GPUs and proprietary data centers, which the company says it intends to. If Venice can move off leased infra, the margins on an already profitable business get more interesting, and privacy-first AI stops being a slogan and becomes a defensible cost structure.