Venice AI's token VVV hit a new all-time high today, briefly breaking above $25. The rally was triggered by a mathematical research controversy that brought the importance of "private inference" into the spotlight.

NYU mathematician Tristan Buckmaster wrote in a public statement that he and his collaborators had fed drafts of their entire project into Codex. After learning that an internal OpenAI team had also made related progress, he asked whether their models had accessed those conversations or been trained on them. He was told the models did not review user data, but his follow-up question about training went unanswered.

OpenAI's response denied that researchers or agents accessed specific user data to solve the problem, while acknowledging that—though unlikely—it could not rule out that de-identified data from the research group's private Codex sessions had helped improve the models.

The community began to question: when the stakes are high enough, can these labs see all your work and beat you to the finish line?

Trader based16z has already made his play. He disclosed going long VVV through spot, perpetual contracts, and OTC call options with a $25 strike price. In his view, this incident has made the importance of private inference widely recognized, and Venice is the most suitable liquid asset to capture this narrative. He also drew a comparison between VVV's circulating market cap and ZEC at $50. That's his bet on a repricing of privacy.

As AI moves from answering general knowledge questions to participating in papers, code, product development, and trading strategies, what users type into the input box has become far more valuable.

Who can let people use AI while retaining control over their work content and execution process? Three Web3 projects have offered their answers.

VVV: Turning Privacy into a Business

Venice AI, founded by Erik Voorhees, offers a chat application for consumers and APIs for developers to call models. It aggregates different models, differentiating itself through privacy and fewer restrictions.

Venice AI's privacy operates on three levels. Venice's "anonymous mode" hides user identity, though upstream models can still see request content. "Zero-retention mode" relies on service providers honoring their commitments. "TEE mode" runs inference inside a protected hardware environment, while "end-to-end encryption mode" encrypts from the user's device and only decrypts once inside the protected environment.

This business has already reached a meaningful scale. Banyan, an investor in the project, disclosed that Venice's annualized revenue grew from $14 million in January of this year to over $100 million by August.

VVV is a token issued by Venice on the Base network. For every $100 of Venice API credits purchased by users, $5 is used to buy back and burn VVV. On the supply side, new emissions of VVV tokens are also declining. On September 1, the annual emission rate was reduced from 3 million to 2.5 million tokens, with plans to further cut it to 2 million tokens on October 1.

Another source of demand for VVV comes from DIEM. Holders can lock staked VVV to mint DIEM, and staking 1 DIEM grants them $1 in refreshed API credits daily. This allows developers and agents to hold an asset that continuously generates usage credits, preparing for future model calls.

On September 14, DIEM's target supply will complete a phased expansion from 38,000 to 40,000 tokens, creating room for additional minting. This move also signals the Venice team's optimism about user growth.

The bullish case for VVV is clear. Amid AI privacy threats, Venice has carved out its market niche—more users paying for Venice drives more buybacks, and more users needing sustained inference credits drives lock-up demand.

NEAR: Verifiable Privacy Inference

Following the trail of Venice AI, we also spot a familiar name: NEAR. This Layer 1 blockchain is expanding the utility of the NEAR token through confidential computing and agent services.

In March of this year, Venice announced an integration with NEAR AI, allowing users to opt for verifiable privacy inference services provided by NEAR AI. When consumers send requests through Venice, the underlying privacy computing power is supplied by providers such as NEAR AI.

The core capability of NEAR AI Cloud is running models within TEEs—trusted execution environments isolated by hardware. By design, plaintext data during computation is confined to protected zones, preventing infrastructure operators from directly reading it. Users can verify hardware attestations to confirm that requests entered the appropriate environment. This offers teams seeking to protect research and commercial data a cloud computing option.

Open-weight models can be deployed within this environment. When closed-source models such as Claude, GPT, and Gemini are invoked via the gateway, requests are still routed to upstream service providers, and NEAR's confidential computing cannot extend coverage to their servers. NEAR's value on this track stems from its ability to protect the computation process and organize model-serving capabilities.

This capability has already been linked to the NEAR token. The staking payments launched on July 30 allow token holders to convert NEAR staking rewards into inference credits, with the amount fluctuating based on staked volume, token price, and yield rates. Users retain ownership of the underlying tokens and can unstake to exit at any time. For teams that continuously call models, this adds another use case for holding NEAR.

Another slice of NEAR's opportunity lies in payments. To complete tasks, AI agents need more than just model calls—they require purchasing data, paying for services, and moving assets across different chains. NEAR Intents offers an intent-based execution model where users submit desired outcomes, and solvers compete to fulfill swaps and executions. This infrastructure can plug complex cross-chain operations into an AI agent's task workflow.

According to DeFillama data, NEAR Intents generated approximately $9.32 million in total fees during Q2 of this year, with the protocol retaining around $1.5 million; the retained revenue is used for market buybacks of NEAR.

Therefore, NEAR's bullish narrative is supported by two business pillars: providing computation for sensitive tasks and delivering execution and payments for cross-chain operations. The former can reach users through applications like Venice, while the latter has the opportunity to grow alongside AI agents.

TAO: Supplying Open AI Models

Bittensor has long been the most high-profile AI project in crypto. TAO is the native token of the Bittensor network.

Bittensor organizes different tasks into independent subnets, where miners provide services such as inference, storage, and prediction, validators assess service quality, and the network distributes rewards according to set rules. A team can organize competition around a specific need, while the entire network accommodates multiple such markets.

As AI demand expands, applications require more interchangeable suppliers, and smaller teams need access to customers, compute resources, and capital. Bittensor attempts to organize this supply side through an open incentive market, allowing different teams to compete on specific tasks.

Some subnets have already begun generating external revenue. Take Chutes, which provides model inference services, as an example—its Q2 revenue this year was approximately $1.37 million, sourced from subscriptions, pay-per-use, and instance services.

How does TAO capture this growth? Each subnet has its own alpha token, paired with TAO to form trading pools. Staking TAO into a subnet effectively converts it into the corresponding alpha token; thus, TAO serves as the base asset for capital allocation across subnets. If more competitive services emerge within the network, the demand for participating in these markets has the potential to expand.

On the supply side, TAO retains the scarcity design familiar to the crypto market. TAO has a total supply cap of 21 million tokens, with its first halving completed in December 2025. It currently issues 0.5 tokens per block, approximately 3,600 tokens per day.

The surge in VVV provides us with a window of observation. As models become increasingly powerful, people entrust them with more sensitive information, and "crypto/privacy AI" has naturally found its product-market fit.

Welcome to join the official BlockBeats community:

Telegram Subscription Group: https://t.me/theblockbeats

Telegram Discussion Group: https://t.me/BlockBeats_App

Official Twitter Account: https://twitter.com/BlockBeatsAsia