articleCrypto & Digital Assets

Why Crypto VCs, Not AI Funds, Bet $65M on Venice AI's Uncensored Platform

Venice AI, a privacy-focused AI platform co-founded by ShapeShift's Erik Voorhees, raised $65 million at a $1 billion valuation in a Series A that closed the first week of July 2026. The lead...

ByJeff Barnes, MBA

July 4, 2026·11 min read

Reviewed by Jeff Barnes — CEO of Angel Investors Network · MBA · $1B+ in Capital Formation

Why Crypto VCs, Not AI Funds, Bet $65M on Venice AI's Uncensored Platform

TL;DR: Venice AI, a privacy-focused AI platform co-founded by ShapeShift's Erik Voorhees, raised $65 million at a $1 billion valuation in a Series A that closed the first week of July 2026. The lead investor was not Sequoia, not a16z, not any name-brand AI fund. It was Dragonfly, an eight-year-old crypto-native venture firm, joined by Coinbase Ventures, Archetype, Morgan Creek, North Island Ventures, Liquid2 Ventures, and Seattle's Founders' Co-op. That investor roster is the story, and it tells you something specific about where crypto capital is rotating in 2026.

The round, first reported by FinSMEs on July 3 and detailed further by GeekWire, marks Venice's first outside capital since its 2024 launch. Founder and CEO Erik Voorhees put up the seed money himself two years ago. President and co-founder Jesse Proudman, a Seattle tech veteran who met Voorhees as an undergraduate at the University of Puget Sound, runs product and engineering. Venice describes itself as a "private, permissionless, and uncensored" alternative to ChatGPT, Claude, and Gemini: it says it doesn't log user prompts, strips most content filters, and stores conversation history on the user's own device rather than on Venice's servers. The company says it crossed 3 million users, turned profitable in the first quarter of 2026, and now employs about 45 people, six of them in Seattle, with no physical office.

For accredited investors watching crypto-fund allocations, this deal is less about one AI startup and more about a pattern: crypto VCs are now the primary capital source for a category of AI infrastructure that traditional AI-focused funds are visibly avoiding. That avoidance is not an oversight. It is a decision, and it is the part of this story the press releases don't examine.

What Venice actually sells, and who paid for it

Venice is not a foundation-model company. It doesn't train its own frontier LLM. It operates as a privacy-hardened router that gives users access to a mix of open-source and commercial models (the company lists Llama, Mistral, DeepSeek, Qwen, and others among its catalog) while adding four tiers of privacy protection: anonymized access, zero-retention private mode, execution inside hardware-secured Trusted Execution Environments (TEE), and end-to-end encryption where even Venice cannot read the prompt. That architecture is the product. As Proudman put it to GeekWire, Venice's answer to government subpoenas and data breaches is to "create no central trove to breach or subpoena in the first place."

The company also runs its own token, VVV, an ERC-20 asset on Coinbase's Base network. Users and developers can stake VVV to mint a secondary credit called Diem, which functions as a claim on Venice's daily inference capacity instead of a per-request bill. According to CoinMarketCap data reviewed for this piece, VVV traded around a $647 million market cap with roughly 47 million tokens in circulation, and the project has run periodic token burns funded by subscription revenue since a "Genesis Burn" destroyed 33.68 million tokens in March 2025.

Investor	Type	Role in round
Dragonfly	Crypto-native VC ($650M Fund IV, closed Feb. 2026)	Lead
Coinbase Ventures	Corporate VC (Coinbase Global)	Participant
North Island Ventures	Crypto/fintech-focused VC	Participant
Archetype	Crypto-native VC	Participant
Morgan Creek	Digital-asset investment firm	Participant
Liquid2 Ventures	Early-stage generalist VC	Participant
Founders' Co-op	Seattle-based early-stage VC	Participant

Every check-writer here except Liquid2 and Founders' Co-op has crypto in its name or its core thesis. Capital raised will go toward three things, per both source reports: expanding the core engineering team, buying GPU compute so Venice can build owned data-center infrastructure instead of renting capacity, and accelerating enterprise-grade zero-knowledge data-security features aimed at business customers who want the privacy architecture without the consumer-facing "uncensored" branding.

Why crypto funds are the ones writing this check

Here's the detail that press coverage of the round undersells: Dragonfly did not stumble into AI. It just closed a $650 million fourth fund in February 2026, one of the largest crypto VC raises of the current cycle, and managing partner Haseeb Qureshi has been explicit that the firm's future focus areas include "AI + crypto convergence" alongside stablecoins, tokenization, and payments infrastructure, according to Dragonfly's published investment profile. Qureshi has also said publicly that "non-financial crypto has failed," a blunt admission that most Web3 consumer applications never found product-market fit. AI is one of the few adjacent categories where crypto's actual comparative advantages, permissionless access, cryptographic verification, token-based coordination of distributed hardware, translate into something a paying customer wants.

Coinbase Ventures has said the same thing in its own words. In a public post titled "Ideas we are excited for in 2026," the firm names AI plus robotics, agentic tooling, and what it calls the "Onchain AI economy" as active investment priorities, describing a future where "AI agents operating on crypto infrastructure can drive economic activity and growth." Venice's autonomous-agent features, letting AI agents stake VVV and mint their own API keys without a human in the loop, fit that thesis directly.

This is the crossover mechanism worth understanding if you allocate to digital-asset funds: crypto VCs spent a decade building expertise in a specific, narrow competency, verifying computation and payments without a trusted central party, and they are now redeploying that competency into AI infrastructure precisely where centralized providers create points of legal and reputational exposure (subpoenas, data breaches, content-moderation liability). That is a different bet than "AI is going to be big," which is the bet every traditional fund already made in 2023. It is a bet that decentralization solves a specific problem for a specific slice of AI demand, and it only makes sense to funds that already have the technical and legal muscle memory from building crypto rails.

The a16z-shaped hole in the cap table

Notice who isn't here. Andreessen Horowitz has an entire dedicated AI investing arm and has led rounds in Character.AI, Mistral, and a long list of frontier-model and AI-infrastructure companies. Sequoia, Thrive, and General Catalyst have all written nine-figure AI checks in the past eighteen months. None of them appear on Venice's cap table. That is not necessarily because they passed and were rejected. It more likely reflects that a16z-style funds have their own portfolio conflicts (many hold positions in OpenAI, Anthropic, or their competitors) and, more pointedly, brand risk from being associated with a company that explicitly markets "uncensored" outputs and minimal content filtering.

That's a real distinction, not a marketing detail. Compare Venice's position to what happened to Anthropic and xAI in the first half of 2026. Reporting from Ars Technica details how the Trump administration flagged Anthropic's Claude models as national security risks and forced weeks of government safety testing before lifting export curbs, while a Reuters-reviewed SpaceX regulatory filing covered by Channel NewsAsia disclosed that xAI faces active investigations across multiple countries tied to sexually explicit AI-generated imagery, including allegations involving minors, with SpaceX warning investors the probes "may hurt market access." Those are the two most direct comparables in the market for what happens when an AI platform relaxes content controls. Traditional AI funds watched both situations unfold in real time this year. Crypto funds, who already operate in a sector accustomed to regulatory friction, evidently priced that risk differently.

The founder's own regulatory history is part of the diligence file

Erik Voorhees is not a first-time founder testing the regulatory waters. He founded ShapeShift in 2014 and ran it as a no-KYC crypto exchange for years before compliance pressure forced changes. Two SEC actions bookend that history and belong in any serious diligence memo on Voorhees personally. In 2014, the SEC charged Voorhees directly over unregistered securities offerings tied to two of his earlier ventures, FeedZeBirds and SatoshiDICE; he settled by disgorging roughly $15,844 in profits plus a $35,000 penalty and agreed not to participate in unregistered virtual-currency securities offerings for five years. A decade later, in March 2024, the SEC charged ShapeShift itself with operating as an unregistered dealer, and the company settled for a $275,000 penalty without admitting or denying the underlying findings. Voorhees has been publicly and consistently critical of both settlements as regulatory overreach.

None of that means Venice will face the same treatment. It does mean the founder has a documented pattern of building products that push right up against, and sometimes across, existing securities and financial-services regulation, and that he treats settling with regulators as a cost of doing business rather than a signal to change course. For a company now building a product explicitly marketed on refusing centralized oversight, that history is directly relevant, not incidental biography.

The regulatory setup around "uncensored" AI is getting less permissive, not more

Venice is scaling into a regulatory environment that tightened meaningfully over the twelve months before this raise, not one that's static. California's Transparency in Frontier AI Act (SB 53) took effect January 1, 2026, requiring large frontier-model developers to publish safety frameworks, report safety incidents, and disclose risk assessments, according to a client briefing from Wilson Sonsini. New York's amended RAISE Act, revised in March 2026 to align more closely with California's framework, delays enforcement to January 2027 but applies to any large frontier developer over $500 million in annual revenue, a threshold Venice would need to watch as it scales. Separately, the White House has pushed AI companies toward voluntary pre-release review of frontier models, a process critics have called a "de facto involuntary licensing regime," per TechCrunch's reporting on OpenAI's restricted GPT-5.6 rollout.

Venice may argue it isn't a frontier-model developer since it routes to third-party open-source models rather than training its own, and the current California and New York statutes are aimed squarely at large frontier developers, not routing platforms. But that is precisely the kind of definitional gap regulators tend to close once a category gets big enough to notice, and "notice" is exactly what a $1 billion valuation and a Dragonfly-led press cycle generates. An academic survey on LLM censorship policy published in mid-2026 and reviewed for this piece notes that the EU's AI Act, which began phased implementation in 2026, uses a risk-based framework broad enough to capture generative AI services regardless of who trained the underlying model, which matters if Venice pursues users or enterprise customers in Europe.

What this signals about crypto capital rotation

Step back from Venice specifically and look at the flow of capital. Katie Haun's Haun Ventures, founded by a former a16z partner, closed a $1 billion fund in May 2026 explicitly expanding from blockchain infrastructure into "AI agents and the intelligent economy," with a thesis centered on decentralized GPU compute as the infrastructure layer that benefits regardless of which AI application wins, according to coverage reviewed on Gate's research blog. Aethir, a decentralized GPU network, closed a $344 million compute reserve backed by a NASDAQ-listed treasury vehicle in April 2026. Grayscale launched a Bittensor Trust in December 2025 to give accredited investors traditional-fund exposure to decentralized AI compute. Coinbase's own venture team has publicly named AI infrastructure and "agentic" tooling as 2026 investment priorities.

The pattern across all of these: crypto capital isn't funding AI applications that compete head-on with OpenAI or Anthropic on model quality. It's funding the infrastructure layer, compute, payments rails for autonomous agents, and privacy/verification tooling, where token-based coordination of distributed resources is a genuine structural advantage rather than a speculative narrative bolted onto an unrelated product. Venice fits that pattern closely: its pitch isn't "our model is smarter," it's "our architecture makes surveillance and censorship structurally impossible," a claim that depends on decentralized, cryptographically verifiable infrastructure crypto funds understand better than most generalist AI investors do.

Risk section: what could go wrong, named plainly

Regulatory risk is the first and largest exposure. An "uncensored" positioning is a marketing asset today and could become a liability the moment a widely publicized misuse incident, of the kind that hit xAI's Grok with multi-country investigations, gets tied to Venice's platform. Venice says it includes some abuse safeguards, but the entire product thesis rests on minimal centralized control, which is structurally in tension with rapid incident response.

Reputational risk compounds that exposure. Every institutional LP evaluating a crypto fund's portfolio will eventually ask about a fund's "uncensored AI" position, and the honest answer involves explaining why the fund is comfortable with a product design that deliberately strips content moderation. That's a harder conversation for a pension fund or endowment allocator than explaining a stablecoin or a Layer 2 bet.

Concentration and correlation risk matters too. A meaningful share of Venice's backers, Dragonfly, Coinbase Ventures, Archetype, Morgan Creek, North Island, are crypto-native funds whose own capital bases are correlated with token prices and crypto-market sentiment. If crypto enters another prolonged downturn, several of these backers' ability or willingness to participate in Venice's Series B could compress at the same time, which is a form of correlated funding risk that doesn't show up in a single press release.

Finally, there is straightforward business risk. Venice's revenue model depends partly on VVV token economics (staking, Diem credits, buy-and-burn mechanics) alongside conventional subscription and API revenue. Token-linked revenue models introduce volatility and accounting complexity that a pure-SaaS AI company doesn't carry, and investors should not assume the $1 billion valuation was built on subscription revenue multiples alone.

The one diligence question that matters most

If you are evaluating a fund's exposure to this growing category, crypto VCs backing decentralized or privacy-focused AI infrastructure, ask the fund manager directly: what is your incident-response plan if a portfolio company's "uncensored" or "unmoderated" positioning produces a headline-generating misuse event, and does the fund have any governance mechanism (board seat, contractual safety commitments, kill-switch provisions) to influence the company's response after the fact? Funds with a real answer to that question, beyond "we trust the founding team," have thought seriously about the tail risk. Funds that treat the question as hypothetical are underwriting a scenario they haven't priced.

Venice AI's $1 billion valuation is a real data point about where sophisticated crypto capital thinks the next several years of AI infrastructure spending will go: toward decentralized compute, privacy architecture, and agent-native rails, funded by investors who already understand how to underwrite technology that operates outside centralized control. That is a legitimate, growing allocation category. It is also one where the downside case looks less like a failed startup and more like a regulatory or reputational event that a subscription-software diligence checklist was never built to catch.