GM. This is Milk Road PRO, where we’re still digging for gems even after everyone has already sold their shovels to bottom blast bitcoin.
Today's edition features our latest PRO report on Venice (VVV), one of the rare crypto projects with a real product, paying users, and a token wired straight into the business.
Here's what we've got for you today:
- 💸 The revenue math: ~$50M ARR today, adding about $2M of new ARR per week, on pace for $150M to $200M in twelve months.
- 🔥 How the buy and burn loop actually works, and why VVV is still around 3.2% inflationary until the July 1 emissions cut.
- ⚠️ The real thesis killer: OpenAI, Anthropic, or Google shipping consumer privacy as a simple checkbox.
Nexo is back in the U.S. - and new clients get 30 days of Wealth Club Premier perks! Higher yields, lower borrowing rates, and crypto cashback - start here.
Prices as of 2:00 p.m. ET. Powered by CoinGecko.
Before we continue: We want to give a huge shoutout to Yan Liberman, Co-Founder at Delphi Digital, for his excellent deep dive on Venice published on May 19, 2026.
Yan’s work was a major source of inspiration for this report, and several of the figures, assumptions, and analytical frameworks below were originally surfaced in his research. We’ve updated this report to more clearly credit those contributions.
Is VVV overpriced at $813M market cap?
Venice (VVV) is up 960% in 2026.
It is one of the only green performers in crypto, and many feel it has plenty of room to keep going. But is that really true?
Venice runs at roughly $50M in annualized revenue, has 3M signups, and burns tokens onchain every week against subscription cash. Erik Voorhees self-funded the whole company. There is no VC stack waiting to dump. The product ships, the users pay, and the token has a working buyback loop.
That is the bull case, and most people stop there.
And yet…VVV at a $813M market cap is already priced for the growth that hasn't arrived yet, and the biggest risk (OpenAI, Anthropic, or Google shipping consumer privacy) is a product decision, not an engineering one. They could announce it tomorrow.
So I am not buying VVV today. But I’m watching four things that would change that view, and three that would break the thesis entirely. This report lays out both, plus the revenue math, the buyback mechanics, and where the moat actually holds up.
And if you DO want to see what I’m buying, I actually just added two new assets to my Watchlist in Milk Road PRO, and sold quite a few things earlier this week. Remember: it’s only one buck to test out the platform for seven days.
WHAT VENICE DOES
Venice launched in May 2024 as a single interface for text, image, video, audio, and code generation. Users can switch between Claude Opus 4.6, GPT-5.4, FLUX, Grok Imagine, Sora 2, and over 230 other models in real time. No prompt logging. No content filters layered on top. The API uses the same standard format as OpenAI's, which has become the default across the AI tooling ecosystem. That means developers can point existing agents, apps, and tools (Cursor, LangChain, anything built on the OpenAI SDK) at Venice by changing two lines of code.
Erik Voorhees founded the company and self-funded it from day one. He is the founder and former CEO of ShapeShift, one of the earliest non-custodial crypto exchanges, and ran Satoshi Dice before that. No VC capital. No board pressure. No outside unlock waiting to hit the market.
The VVV token launched on Base on Jan. 27, 2025, the same day China's DeepSeek made global headlines for data privacy concerns. The timing was not an accident. Coinbase listed VVV within hours. Half of the 100M genesis supply was airdropped: 25% to existing Venice users and 25% to AI agent protocols on Base including Virtuals, aixbt, Luna, and VaderAI. The remainder went to staking emissions, the team, and the treasury, all on vesting schedules.
Two things to flag now, because they shape everything that follows.
First, VVV does two specific things, and it helps to be clear about both. Staking VVV gives the holder a share of Venice's API capacity, which means developers and AI agents can run queries through the platform without paying per request. Separately, Venice uses subscription revenue to buy VVV on the open market and burn it, which means platform growth translates directly into supply reduction. So VVV is not a governance token (Voorhees runs Venice as a company, not a DAO) and not a pure narrative play. It is an access pass for the API and a structural claim on the business.
Second, Venice ships. The platform added 230+ models, automated burn execution, AI Agent Skills, Memoria (a local-only memory system), and a full timeline-based video editor in the last six months alone.
THE REVENUE MATH
This revenue framework is heavily informed by Yan Liberman’s May 19 analysis, which used Venice’s subscriber data, signup milestones, and token throughput to estimate current ARR and forward growth.
Venice runs at roughly $50M in annualized revenue today, adding about $2M of new ARR per week. At that pace, the company will cross $150M ARR in twelve months.
The weekly data already shows it.
Yan’s piece also surfaced the signup cadence and paid-user assumptions used below.
Venice reported 2M signups on Feb. 4, 2026 and 3M by May 16, roughly 300K new signups per month. At 6% paid conversion, that puts paying subscribers near 150K, with 88% on the entry-level Pro plan at $18 per month.
The Pro+ ($68) and Max ($200) tiers only launched in mid-April, so the mix has not yet had time to shift. Weighted revenue per user (ARPU) sits at $18-19 per month today, implying subscription annualized revenue of approximately $33M.
The mix is a forward signal worth flagging. If even a small share of existing Pro users migrate up as they hit usage caps, ARPU rises mechanically without Venice needing to add a single new subscriber. A shift from 88% Pro to 70% Pro (with the rest splitting between Pro+ and Max) would lift ARPU by roughly 50%, all else equal.
Yan’s report makes a similar case that API revenue is becoming material, based on token throughput growing much faster than the paid subscriber base
API revenue is harder to pin down. Venice does not disclose it. But daily token throughput grew from 20B in early February to over 60B by early May, a 3x ramp while paid subscribers grew only 50%. That gap implies API workloads are scaling faster than subscriptions, likely driven by agent deployments and integrations with OpenRouter, Cursor, Brave Leo, Fleek, and VSCode extensions. Let’s assume $20M in annualized API revenue, on the conservative side of public estimates.
The question for the next twelve months is whether the addition rate sustains.
If it does, Venice scales from $50M ARR today into the $150-200M range. That is a real business with real numbers behind it.
But the business and the token are separate questions. Plenty of crypto projects have built revenue without the token capturing any of it.
CRYPTO SHOULD WORK HARDER FOR YOU
Most people hold crypto and hope.
The smart money? They're earning interest on it, borrowing against it without selling, and trading it.
Where can you do the same all in one place? Nexo.
And right now, new U.S. clients get 30 days of Wealth Club Premier (benefits normally reserved for loyalty program members):
- Enhanced interest rates on your digital assets
- Lower borrowing costs against your crypto
- Up to 0.5% cashback on trades
No need to sell to access liquidity. No juggling 5 different platforms.
*Disclaimer: Geographic restrictions and terms apply.
DOES VVV CAPTURE THE VALUE?
Venice, the company, and VVV, the token, are separate things. Voorhees self-funded the company, so there is no preferred equity stack waiting to extract value ahead of VVV holders. But VVV is still a token, not a share. How much of Venice's growth actually flows to it depends on three things: staking utility, buyback mechanics, and net new supply.
Yan’s original piece laid out this token-value framework in detail, especially around staking utility, DIEM, emissions, and revenue-linked burns.
Staking VVV does three things:
- It earns yield from emissions (currently 14.7% APR, dropping to roughly 9% after the July 1 emissions cut).
- It gives the holder a proportional share of Venice's API capacity, so developers and AI agents can run unlimited queries without paying per request.
- As Yan explains in his report, VVV can also be locked to mint DIEM, a tradeable ERC-20 token where each unit represents $1 per day of perpetual API credit. DIEM gives users predictable compute pricing without taking on VVV price exposure.
Voorhees has stated explicitly that VVV is not a governance token, so voting is not part of the utility set.
Venice also uses subscription revenue to buy VVV on the open market and burn it. Two streams:
- A monthly discretionary burn from accumulated revenue.
- And since April 2026, a programmatic burn triggered by every new Pro subscription (yellow bars).
Cumulative recurring burns since November 2025 total approximately 190K VVV, or roughly $3.0M at current prices. That is small. Burns are not the demand driver for VVV today. The company is reinvesting cash to grow the business rather than scaling buybacks aggressively. What burns do is signal that the token is economically tied to the platform, and they give observers a verifiable onchain proxy for subscription growth.
A buyback that destroys 20K VVV per month does not matter much if more tokens enter circulation than get burned. So it is worth understanding where the new supply actually comes from.
Yan’s report also detailed the emissions schedule and why the July 1 cut matters for the supply story.
The main source is staking emissions. Venice pays stakers in newly issued VVV at a fixed annualized rate set by the protocol. Currently 5M VVV per year, or roughly 416K per month. The rate has stepped down six times since launch, from 14M/year at TGE to 5M today, with scheduled cuts to 3M on July 1. At the July 1 rate, monthly emissions drop to approximately 250K VVV.
A secondary source is team vesting. Voorhees and the team received 10M VVV at launch, of which 7.5M vests linearly over 24 months. About 4.83M has vested as of May 2026. The remaining 2.67M unlocks gradually through January 27, 2027, after which the team allocation is fully circulating.
Venice also holds approximately 31M VVV across the treasury, Incentive Fund, and other reserves. How and when those get deployed is at Venice's discretion and not on a fixed schedule. This is the largest source of potential supply that I cannot accurately model.
Against current burns of 20K per month, net new supply from emissions is approximately 212K VVV per month, or about 3.2% annualized inflation. That math improves automatically after the July 1 cut and improves further as subscription revenue grows and the programmatic burn scales with it. Net deflation is achievable. It is not quite here yet.
WHAT KILLS VENICE?
Venice’s biggest challenge may be that privacy becomes a feature rather than a product category. If the major AI labs make privacy the default experience for consumers, Venice has a much harder time convincing users to switch platforms.
This is also one of the key risks highlighted in Yan’s competitive framework.
The second biggest risk is privacy-native incumbents (Brave, DuckDuckGo, Proton) collapsing Venice's funnel before it gets the chance to actually scale.
The Big Tech threat is closer than most Venice bulls admit
OpenAI, Anthropic, and Google already offer real privacy at the enterprise tier. No training on your data. Short or zero retention. Legal recourse under contract.
When they eventually offer privacy at the retail level, Venice loses most of its addressable market. Most users do not want privacy plus uncensored content plus crypto-native economics. They just want privacy. If their existing AI provider offers it, the switching friction disappears. The privacy-only segment is Venice's largest audience and the segment most exposed.
What Venice still has after that is narrower but real. Hardware-attested privacy for users with serious threat models (journalists, activists, lawyers handling privileged communications), which Big Tech cannot easily ship at consumer scale outside Apple's locked Apple Intelligence stack. And uncensored model access for the slice that needs it, which Big Tech structurally will not enter because of brand, regulatory, and enterprise consequences.
That is a much smaller pie than the privacy-conscious mainstream user. Whether it is big enough to justify the current valuation is the question that determines whether VVV works as an investment.
Yan’s report also frames the competitive risk as an unbundling problem: different players can attack different parts of Venice’s wedge.
The other threat is privacy-native competitors with much bigger distribution
The Brave, DuckDuckGo, and Proton comparison below is based heavily on Yan’s market map.
Brave is the headline example: over 115M monthly active users, 47M daily actives, and Brave Leo Premium at $14.99/month with no account required and local-only history. Brave is now moving toward hardware-attested privacy through NEAR AI (Venice's own partner). DuckDuckGo AI Chat and Proton Lumo are pursuing variants of the same play from different distribution surfaces - search and the Proton privacy ecosystem.
These three are dangerous because they collapse the funnel. They meet privacy-conscious users inside products already in daily use. They do not need to persuade anyone to seek out private AI. For the casual privacy-aware user, the privacy they offer plus a familiar brand may be enough.
What they do not match is depth. Brave Leo and DuckDuckGo operate at the anonymous-proxy level. Lumo operates at zero-access encryption. None ship the full privacy ladder, none offer uncensored access, and none have crypto-native economics. They are competing for the casual user, not the user with a serious threat model or unfiltered-content needs.
The more likely outcome is death by a thousand cuts. Different competitors attack different pieces of the Venice value proposition, leaving a smaller market for the full package.
Probably not a single competitor kills Venice. Big Tech takes the privacy-conscious mainstream user. Brave and DuckDuckGo take the casual privacy-aware user. Local AI eventually takes the technical user who runs models on their own hardware. Each one chips off a slice.
What is left is the user who needs the full bundle: hardware-attested privacy, uncensored access, and crypto-native economics, all in one product. That market is real and currently uncontested. It is also much smaller than "all of consumer AI."
Whether that smaller market justifies a $813M token is the valuation question I want to look at next.
WHAT IS VVV WORTH?
At $17.4, VVV trades at $813M market cap and $1.39B fully diluted.
On $50M of current revenue, that is 16x on market cap and 28x on FDV. On forward revenue of $150-200M, the multiples compress to 4x and 7x.
The market’s closest benchmark today is probably OpenRouter, another fast-growing AI infrastructure business that sits between users and model providers. OpenRouter raised $113M at a $1.3B valuation on May 26 (which included names like Alphabet's growth fund, NVIDIA, a16z, plus the venture arms of MongoDB, Snowflake, and Databricks) on roughly $50M ARR. That is 26x revenue. VVV at 16x looks cheaper. But that comparison doesn't stand once you look at gross profit.
Yan also makes the important distinction that OpenRouter and Venice are not margin-identical businesses.
OpenRouter is a pure pass-through marketplace. It takes a 5% markup on Claude, GPT, and other API calls, with the other 95% flowing back to model providers. Its $50M of revenue is the markup on roughly $1B of inference spend, which is why gross margins clear 90%.
Venice is the seller of record. When a user pays $18 per month, the full $18 is gross revenue, and Venice pays the inference costs out of it. Queries routed to Claude or GPT carry a thin margin. Queries routed to Venice-hosted open-source models run on GPU rental time and carry a much higher margin. Venice has not disclosed actual gross margins. Public estimates run 40-60%. Let’s go with 40%.
The multiples flip on gross profit. OpenRouter trades at 27x. Venice trades at 37x on market cap and 63x on FDV. OpenRouter also routes $1B of inference spend per year, which justifies some premium on top.
Another key point from Yan’s report: VVV should not be treated exactly like equity.
The token benefits from platform growth, but it does not carry the legal rights or protections that equity holders receive. Whereas OpenRouter investors own equity with residual claims and governance rights.
Venice is unusual in one respect. Voorhees self-funded the company, so no preferred equity stack sits ahead of VVV holders. That is the most common failure mode for VC-backed crypto projects, and Venice does not have it. The token-versus-equity discount should be smaller than average. It should not be zero.
SO IS VVV CHEAP?
On headline revenue, yes. On gross profit, the discount disappears. VVV trades at roughly 37x gross profit on market cap. OpenRouter trades at 27x. The token-versus-equity adjustment narrows the gap further.
The forward case is more interesting. If Venice sustains $2M of weekly ARR additions, twelve-month forward revenue lands near $250M. At today's market cap, that is 3x forward revenue and roughly 8x forward gross profit. For a business doubling revenue in twelve months, that is not stretched.
So the bull case for VVV is not "the token is mispriced today." It is "the growth trajectory makes today's premium look reasonable by mid-2027."
At this point, I think the market understands the story. What it doesn’t know yet is whether Venice can keep compounding fast enough to justify today’s valuation. That’s what I’ll be watching over the next few quarters.
MY POSITION
To keep reading my take on Venice, you’ll need to join Milk Road PRO. It’s literally just one buck for a quick 7-day trial, and if you hate it you can cancel anytime.
Although I’m not buying Venice just yet, I did add two new assets to my Watchlist this week that I’m pretty excited about.
BITE-SIZED COOKIES FOR THE ROAD 🍪
Is this one of crypto's most underrated real-adoption plays? XDC sits right where RWAs, cross-border payments, and institutional money are all colliding.*
That’s my booooy! Our lead macro analyst / all-round ETH bull, John Gillen, just debated David Hoffman on the ‘ETH is money’ thesis.
Jamie Coutts: The setup for crypto's next leg is hiding in plain sight.
Ok, but why? PMI hit 54.5% on AI CapEx demand while employment contracted for the third straight month.
Your Bitcoin can hand you cash without selling a single sat → That's Ledn's whole pitch, and our full review breaks down everything you need to know about the platform.*
*this is sponsored content.
MILKY MEMES 🤣
ROADIE REVIEW OF THE DAY 🥛
VITALIK PIC OF THE DAY
