Research report
VVV is the governance and utility token of Venice AI, a privacy-first decentralised AI inference platform founded by Erik Voorhees. It represents the largest position in the Arxonic portfolio at 51.43% allocation, rated High Conviction.
The investment thesis centres on three pillars: growing demand for private, uncensored AI inference; a tokenomic structure trending toward net deflation; and the DIEM utility floor, which provides a quantifiable minimum value anchor for the entire VVV ecosystem.
Key catalysts over the next six months include confirmed emission reductions from 6M to 3M VVV/year across May–July 2026, an accelerating revenue-driven burn mechanism, and continued platform growth toward and beyond 2 million users. Primary risks are platform dependency, admin key concerns, and relatively thin liquidity.
Return on capital deployed at time of writing: +494%.
Venice AI is a generative AI platform that provides private, censorship-resistant access to leading open-source and proprietary AI models for text, image, code, and video generation. The platform was founded in May 2024 by Erik Voorhees, the former CEO of ShapeShift, and Teana Baker-Taylor, former VP at Circle.
Venice differentiates itself from centralised AI platforms (ChatGPT, Claude, Gemini) through three design principles: conversations are processed without data logging on centralised servers, no content filtering or censorship is applied to outputs, and users can interact with the platform anonymously without creating an account.
The platform supports a broad range of models including Llama 3.1 405B, DeepSeek R1 671B, Qwen 2.5 VL 72B, and integrations with proprietary models such as Claude, GPT, and Grok variants. Venice also offers image generation through FLUX and Stable Diffusion, and has expanded into video generation as part of the Venice V2 platform update.
As of early 2026, Venice has surpassed 2 million registered users and processes over 45 billion LLM tokens daily. The consumer product offers a free tier with daily limits and a Pro subscription at $18/month for unlimited text prompts, up to 1,000 images per day, and additional features including PDF analysis, system prompts, and API access.
The VVV token launched on 27 January 2025 as an ERC-20 token on the Base network. It functions as an access key to the Venice ecosystem — staking VVV provides a pro-rata share of the platform's inference capacity and earns emissions-based yield.
The thesis for VVV rests on three interconnected pillars.
Pillar 1: Demand for permissionless AI inference is growing.
As centralised AI providers tighten content restrictions, increase surveillance, and impose ideological constraints on model outputs, demand for uncensored alternatives is rising. Venice addresses this demand directly. The platform's growth from launch in May 2024 to over 2 million users by early 2026 demonstrates real adoption, not speculative interest. The expansion of the model lineup to include video generation, character interaction, and developer APIs broadens the addressable market beyond text-only use cases.
The target audience is expanding from individual privacy-conscious users to include autonomous AI agents, developers building applications that require unrestricted inference, and businesses seeking predictable AI compute costs. The staking model — where VVV holders receive perpetual API access rather than paying per-request — is particularly attractive for high-frequency automation use cases.
Pillar 2: Tokenomics are trending toward net deflation.
VVV's supply dynamics are structurally improving through two mechanisms: emission reductions and revenue-driven burns.
Annual emissions have been reduced three times since launch: from the initial 14M/year (at token generation) to 10M (August 2025), then to 8M (October 2025), and most recently to 6M (February 2026). A further confirmed reduction schedule will bring emissions from 6M down to 3M across May–July 2026, representing a 50% cut in new supply over three months.
Simultaneously, Venice launched a revenue-based buyback-and-burn programme in December 2025. Under this mechanism, a portion of monthly platform revenue is used to purchase VVV on the open market and permanently destroy the tokens. Monthly burns have included approximately 57,000 VVV in December 2025 (~$64K at ~$1.13/token), 45,000 VVV in January 2026 (~$94K at ~$2.07/token), 38,000 VVV in February 2026 (~$70K at ~$1.86/token), and 22,000 VVV in March 2026 (~$112K at ~$5.14/token, based on February revenue).
As of early March 2026, over 33.7 million VVV tokens have been removed from circulation, representing approximately 42.5% of the initial total supply. This figure includes both the revenue-driven burns and a large burn of unclaimed airdrop tokens in March 2025.
At the confirmed 3M/year emission rate with the current burn percentage, net new supply entering the market drops to approximately 1.7M VVV annually. If burn rates continue accelerating alongside platform revenue growth, a net deflation crossover — where more tokens are destroyed than created — becomes achievable by late 2026.
Pillar 3: The DIEM utility floor provides a quantifiable value anchor.
Each DIEM token provides exactly $1 per day in Venice API inference credits in perpetuity. DIEM can only be minted by locking staked VVV, creating a direct economic link between VVV supply lockup and the tokenised compute market. This gives VVV something that most crypto tokens lack: a downstream asset with hard, observable utility value that does not depend on speculation.
The DIEM utility floor means that as long as Venice's AI inference service operates and honours the $1/day credit commitment, there is a minimum demand-side anchor for VVV — because DIEM can only be created from locked VVV.
Initial allocation:
The VVV token launched with an initial supply of 100 million, distributed as follows: 50% (50M) airdropped to the Venice community and AI-related projects on the Base network, 35% (35M) allocated to Venice.ai for development and growth, 10% (10M) designated for the Venice Incentive Fund, and 5% (5M) reserved for liquidity provision on decentralised exchanges.
Emission history:
Annual emissions have followed a consistent downward trajectory since launch. The initial rate of approximately 14M VVV per year was reduced to 10M in August 2025 when DIEM was introduced, then to 8M in October 2025 as part of the "Venice is Burning" tokenomics update, and to 6M on 10 February 2026 — a permanent 25% reduction.
Confirmed upcoming reductions:
Venice has announced a further 50% emission reduction across three monthly steps:
- 1 May 2026: 6M → 5M per year
- 1 June 2026: 5M → 4M per year
- 1 July 2026: 4M → 3M per year
- December 2025: ~57,061 VVV (~$64,000 value at time of burn, VVV ~$1.13)
- January 2026: ~45,303 VVV (~$94,000, VVV ~$2.07)
- February 2026: ~37,826 VVV (~$70,000, VVV ~$1.86)
- March 2026 (February revenue): ~21,778 VVV (~$112,000, VVV ~$5.14)
Staking:
VVV holders can stake their tokens to earn emissions-based yield and gain access to a pro-rata share of Venice's API inference capacity. The current staking APR is approximately 18.52%, though this rate is variable and depends on total staked supply and platform utilisation.
Venice distributes staking emissions using a utilisation-based model. At approximately 50% platform utilisation (the target efficiency point), Venice itself receives maximum emissions to fund infrastructure expansion. This incentivises the team to maintain optimal capacity while ensuring growth headroom.
Dual-tier staking strategy:
This position employs a dual-tier approach: approximately 4% of the position is staked at full APR for maximum yield, while approximately 96% is locked to mint DIEM. The locked portion continues earning 80% of the normal staking yield while simultaneously generating DIEM tokens.
DIEM minting:
DIEM was introduced in August 2025 as a secondary ERC-20 token on Base. It can only be minted by locking staked VVV (sVVV), and each DIEM token provides exactly $1 per day in Venice API inference credits when staked. This credit renews daily and covers access to all models available through the Venice API.
The mint rate is dynamic — Venice controls a target DIEM supply variable, and the number of VVV required to mint one DIEM adjusts accordingly. Higher target supply makes minting cheaper; lower target makes it more expensive.
DIEM can be freely traded on Aerodrome (the primary DEX on Base). Burning DIEM unlocks the original locked sVVV at any time, providing an exit mechanism for holders who no longer need API access.
Key innovation:
Before DIEM, VVV stakers received a fluctuating daily API allocation based on their share of total staked supply. DIEM fixed this — each token provides a predictable $1/day regardless of how many other users are staking. This predictability is critical for developers and AI agents that need to budget compute costs reliably. Venice has also optimised capacity allocation to divide resources among active API users (those who made at least one API call in the past seven days), resulting in approximately 14x improvement in effective capacity per staker.
Venice competes across two dimensions: decentralised AI compute and private AI inference.
Decentralised compute competitors:
The primary comparables are Render (RNDR), Akash (AKT), and Bittensor (TAO). However, these projects serve fundamentally different functions. Render is a GPU rendering marketplace primarily serving visual effects and 3D rendering workloads. Akash is a decentralised cloud computing marketplace with broader infrastructure ambitions. Bittensor is a decentralised machine learning network focused on model training and validation.
Venice differentiates by operating at the application layer rather than the infrastructure layer. While Render, Akash, and Bittensor provide raw compute resources, Venice delivers a consumer-ready AI product that non-technical users interact with directly. The platform has over 2 million registered users — none of the infrastructure-layer competitors have comparable consumer adoption.
Private AI alternatives:
Users seeking private AI access can run local models on their own hardware, use centralised APIs through VPNs, or access various open-source model hosting services. Venice competes on convenience — it provides genuine privacy (no data retention) with the ease of a web application, avoiding the hardware investment of local deployment and the technical overhead of self-hosting.
Economic model differentiation:
Most competitor utility tokens operate on a spend-to-use model — users actively pay tokens to receive services. Venice's model inverts this: users stake VVV and receive ongoing API access without spending the token. The DIEM layer adds a further innovation — hold-to-access with predictable daily credits. This is genuinely novel and has no direct comparable in the crypto utility token space.
Moat analysis:
Venice's competitive position strengthens through the dual-token flywheel. As platform usage grows, DIEM demand increases, which requires more VVV to be locked (reducing circulating supply), which supports VVV price, which attracts more stakers, which generates more DIEM supply for the compute market. Competitors would need to replicate not just the inference platform but the entire staking-to-utility economic loop.
User growth:
Venice reached 1 million registered users by early 2026 and surpassed 2 million users by late January 2026, as confirmed in Venice's own changelog.
Inference volume:
Venice founder Erik Voorhees stated in February 2026 that the platform was processing over 45 billion LLM tokens daily, representing a doubling from earlier in the month. This metric demonstrates real computational throughput rather than just registered accounts.
Model integrations:
The platform has expanded well beyond its initial open-source model offerings. Venice now provides access to proprietary models including Claude (Anthropic), GPT series (OpenAI), Grok (xAI), and numerous open-source options. Video generation capabilities were added in October 2025 as part of Venice V2, with music generation tools following.
Developer ecosystem:
Venice's API has gained traction with the developer and AI agent community. Notably, OpenClaw (openclaw.ai) — a decentralised AI agent framework — named Venice as its primary recommended model provider in March 2026, listing Venice models as default configurations in its documentation. This type of integration embeds Venice directly into developer workflows and creates structural demand for VVV staking.
Revenue signals:
While Venice does not publish detailed revenue figures, the monthly buyback-and-burn programme provides an observable proxy. The burns represent a portion of platform revenue, meaning actual revenue exceeds the burn values. The Pro subscription at $18/month with 2M+ users and the API's growing developer adoption suggest meaningful revenue generation.
VVV resists simple valuation models because its value derives from multiple sources: staking yield, DIEM minting rights, governance, and speculative premium on the decentralised AI narrative. The framework below addresses each layer.
Layer 1: DIEM utility floor.
Each DIEM provides $1/day in API credits, equating to $365/year in observable utility value. Since DIEM can only be minted from locked VVV, there is a mathematical relationship between VVV price and DIEM economics. At any given VVV price and mint rate, you can calculate the implied DIEM payback period — how long the daily API credits take to equal the value of the VVV locked to mint one DIEM.
Layer 2: Staking yield.
At the current ~18.52% APR, staked VVV generates meaningful yield. This yield is paid in additional VVV tokens, which can be compounded through restaking. The yield component provides a floor return for holders who are not minting DIEM, analogous to a dividend on an equity.
Layer 3: Emission cut catalyst.
The confirmed reduction from 6M to 3M VVV/year across May–July 2026 is the most significant near-term catalyst. This 50% supply reduction, combined with the ongoing burn programme, structurally changes the supply-demand dynamics. Historical data from both Bitcoin halvings and previous VVV emission cuts suggests that supply reductions with stable or growing demand lead to price appreciation, though past performance is not predictive.
Scenario analysis:
Bear case: Venice fails to grow beyond its current user base, emission cuts reduce staking yields without corresponding demand growth, and DIEM trades at a persistent discount to its utility value due to lack of API adoption. VVV stagnates or declines, though the staking yield provides some downside cushion.
Base case: Venice continues moderate growth, emission cuts create supply shock that drives price appreciation, DIEM demand grows with the developer ecosystem, and the revenue burn programme scales with platform revenue. VVV appreciates meaningfully from current levels.
Bull case: Decentralised AI inference reaches mainstream adoption, Venice captures significant market share from centralised providers, DIEM demand creates a secondary premium above the utility floor, and net deflation accelerates as burns exceed emissions. VVV re-rates as a productive, deflationary asset with real yield.
Platform dependency — HIGH
This is the single most important risk. VVV's value is entirely dependent on Venice AI continuing to operate, maintain its technology, and grow its user base. If the Venice team abandons the project, executes poorly, faces legal challenges, or is outcompeted, the entire investment thesis fails. There is no fallback utility for VVV outside the Venice ecosystem. This risk is monitored weekly through platform activity, team communications, and ecosystem developments.
Smart contract and admin key risk — MEDIUM
CoinGecko has flagged that the contract creator can make changes to the VVV token contract, including disabling sales, changing fees, minting tokens, and transferring tokens. While not uncommon for early-stage DeFi protocols, this represents a centralisation risk that contradicts the decentralised narrative. The team's track record of consistent execution mitigates this concern somewhat, but progressive decentralisation of contract control would strengthen the investment case.
Liquidity risk — MEDIUM
VVV trades primarily on Coinbase and decentralised exchanges with varying daily volumes. While liquidity has improved significantly since launch, the position is sized with the understanding that a large exit would need to be executed gradually over days or weeks to avoid significant price impact.
GPU supply risk — MEDIUM
Venice relies on decentralised GPU providers for its inference infrastructure. Any global shortage of high-end AI chips (such as NVIDIA H100s) could constrain the platform's compute capacity, affecting DIEM utility and API reliability. This is a supply-side risk that is largely outside the team's control.
Regulatory risk — LOW-MEDIUM
Venice's uncensored AI inference model could attract regulatory attention, particularly regarding content generation. The decentralised architecture provides some resilience, but regulatory action against the team, the platform, or the token could impact operations. The privacy-first positioning may also face challenges as governments increase scrutiny of encrypted and anonymous AI services.
Market risk — LOW
Broad cryptocurrency market downturns affect VVV alongside all digital assets. This is accepted as inherent to the asset class rather than a VVV-specific concern. The staking yield provides a partial cushion during drawdowns.
Allocation: 51.43% of portfolio — the largest single position.
Origin: Airdrop + subsequent buy-ins. The initial position was received through the Venice airdrop in January 2025. Additional purchases were made through direct buy-ins.
Return on capital deployed: +494% at time of writing.
Staking strategy: Dual-tier — approximately 4% of the position staked at full APR (18.52% at time of writing), approximately 96% locked at 80% of APR, generating DIEM yield that subsidises development costs.
Position sizing policy: Allocation percentages are disclosed publicly. Absolute position sizes, token counts, and specific dollar values are not disclosed for security and privacy reasons.
Current status: INTACT
What would change the thesis:
This research reflects the author's personal analysis and investment position. It is not financial advice. The author holds VVV and DIEM as disclosed above. All information is believed accurate as of the publication date but may change. Do your own research. Arxonic — arxonic.com — @Arxonic
- Venice team abandoning or significantly pivoting away from the AI inference platform
- Emission reduction schedule being delayed, reversed, or cancelled
- Smart contract exploit or admin key misuse resulting in loss of funds
- Sustained decline in platform usage metrics (users, daily tokens processed)
- A competing decentralised AI platform achieving significantly superior adoption
- Regulatory action that prevents Venice from operating in major markets
- Monthly burn quantities and implied revenue trajectory
- Platform user growth and daily inference volume
- Emission reduction execution in May, June, and July 2026
- DIEM secondary market pricing relative to the $1/day utility floor
- Developer ecosystem growth, particularly AI agent adoption
- Any changes to staking mechanics, DIEM minting rates, or token contract permissions
This research reflects the author's personal analysis and investment position. It is not financial advice. The author holds VVV and DIEM as disclosed above. All information is believed accurate as of the publication date but may change. Do your own research. Arxonic — arxonic.com — @Arxonic