DGrid said its Genesis membership program brought in more than $23 million in the first half of 2026, with paying members topping 15,000, a figure the company presents as evidence that decentralized AI is starting to win direct, paid adoption rather than relying only on concept-driven market narrative.
The article argues that DGrid is not built as a pure decentralized compute platform or a basic model-routing tool. Instead, it describes the project as a verifiable intelligence network hub linking AI developers, enterprise users and crypto-native users. Its focus is split between two practical goals: making inference more verifiable and making the payment loop work in a real market setting. In DGrid's framing, the first addresses trust, while the second tests actual demand. That is the backdrop for its Proof of Quality, or PoQ, mechanism.
Genesis membership used as a commercial test
The Genesis membership program is presented as DGrid's clearest business result so far. Users pay an annual fee of $1,580 to join. In return, they receive $300 in monthly model usage credits, access to exclusive NFT benefits, one-click deployment through DClaw, and DGAI token mining rewards.
Measured against the listed annual fee, the monthly model access available to members works out to roughly 44% of the official price, according to the article. DGrid described that pricing as highly competitive.
The piece also draws a contrast with traditional Web2 AI subscriptions. DGrid's membership payment records are kept on-chain, which the article says gives the revenue figure a public and transparent basis. In that framing, the more than $23 million in revenue is tied to real payment activity rather than to projections shown in a pitch deck.
Just as important for DGrid, the more than 15,000 paying members are described as proof of sustained demand. Users choosing to pay directly for AI services is used as a sign that the product has delivered practical value, something the article says many token-led projects have struggled to demonstrate.
Outside the membership system, DGrid also uses AI Arena to collect blind-test ratings from users and build a pool of high-quality human preference data. That data is then fed back into routing optimization to improve Gateway service quality. The article describes this as a reinforcing loop: paid demand validates value, and better product performance supports continued usage.
DGrid's seed round was $5 million, the article says. It argues that turning a $5 million seed financing into more than $23 million in revenue is unusual capital efficiency for the sector and offers another signal about the health of the business model.
The article still notes several conditions that matter for the model over time, including retention, credit consumption, model cost control and whether DGAI incentives can keep functioning. Even so, it says DGrid has already taken an important step in decentralized AI commercialization.
Product stack built around actual usage
DGrid links the Genesis revenue to a broader product structure built around user pain points.
AI Gateway serves as a one-stop intelligent routing layer. Based on cost, speed and historical performance, it can automatically call more than 200 mainstream models, including Claude, GPT, Gemini, MiniMax and GLM. The goal is to let users access high-quality AI services through one API instead of integrating separate vendor endpoints.
AI Arena gathers human preference data through blind scoring of model outputs. Users anonymously compare two responses and rate which one is better. Those results can be used to optimize intelligent routing and can also become commercializable labeling assets. The article says Arena has already drawn more than 300,000 participants.
DClaw is designed for fast deployment of local AI assistants at the minute level. Users do not need to configure keys to call top-tier models directly, and the product includes persistent memory plus hot-swappable skill plugins. It is built to work across Telegram, WeChat and Discord and is aimed at longer-term, stable usage scenarios.
Model Marketplace allows AI models to be listed freely, priced independently and tokenized as assets. Model providers can join directly and compete in the market, giving them a more direct path to revenue.
DGrid has also launched Dori, an intelligent recommendation agent that lets users describe their needs in natural language and then receive model and call-plan suggestions. The stated aim is to reduce the barrier to using multi-model AI services.
Across these products, the article says the central question is not feature accumulation. It is how to make AI cheaper to use, easier to trust and easier to access.
BNB Chain integration extends agent capabilities on-chain
If the product stack addresses whether AI is usable, the article says DGrid's integration with BNB Chain tackles a separate issue: whether AI agents can operate autonomously on-chain.
By connecting to Agent Registry and x402 payment capabilities, DGrid gives AI agents on-chain identity and pay-per-request settlement.
On one side, AI agents deployed through DGrid can register on-chain and become discoverable, composable and callable on-chain entities. On the other, when those agents call models on DGrid, service settlement can also happen on a per-request basis.
That shifts the role of AI agents. Rather than remaining purely off-chain tools, they begin to carry on-chain identity, service access and payment functions. The article says that foundation is what could allow agents to evolve from single-purpose tools into intelligent actors that can participate in on-chain economic activity.
PoQ is designed to solve a trust problem in decentralized AI
The PoQ mechanism targets what the article describes as a deeper trust issue in decentralized AI.
In centralized AI services, users typically assume the platform will execute requests correctly, bill accurately and deliver reliably. In decentralized networks, model providers, node operators, developers and users all sit in different roles. Without a verification layer, the article says, low-quality models can be passed off as stronger ones, billing can be falsified and task execution can become hard to trace.
DGrid's Proof of Quality is built around that problem. According to the article, it uses an independent sampling and inspection mechanism to verify the service quality delivered by model providers, then writes verification results on-chain.
The piece stresses one point in particular: PoQ checks whether providers honestly deliver the model service they claimed to offer. The verification process uses the platform's own test sets, does not touch user inference data and does not put user data on-chain.
That distinction matters in open AI service networks. Once AI services move beyond a single platform into a broader network, trust can no longer come only from brand reputation. It needs a mechanism that is verifiable, accountable and billable. In DGrid's framing, PoQ serves as that base trust layer.
The team is also conducting research around PoQ and Optimistic TEE-Rollups. According to public materials cited in the article, core members hold PhD backgrounds from institutions including Stony Brook University, and have published four related papers:
- Proof of Quality: https://arxiv.org/abs/2512.16317
- Optimistic TEE-Rollups: https://arxiv.org/abs/2512.20176
- Cost-Aware Proof: https://arxiv.org/html/2601.21189v1
- PoQ-Judge: https://arxiv.org/pdf/2606.11196
From technical verification to a service network
The article says PoQ began as a way to verify a single inference task, but DGrid's larger goal is not a standalone verification tool. It is trying to build an AI service network connecting model providers, node operators, developers and end users.
In that network, calling, verification, billing and settlement need to operate inside one loop. Users pay fees. Nodes provide services. The system records key processes. Developers and model providers receive revenue. The article argues that only with enough transparency can the network move from early trial usage to broader real-world adoption.
That is also one reason outside institutions have taken interest, according to the piece. DGrid's seed round included participation from Waterdrip Capital, IoTeX, Paramita VC, Zenith Capital and CatcherVC.
The article cites Waterdrip Capital CEO Jademont as saying that if a decentralized network leaves builders unable to understand how data is being processed, it will quickly run into execution bottlenecks.
That comment points to a central challenge in decentralized AI as described in the article: the more open the network becomes, and the more parties it includes, the higher the trust cost gets. If model inference, data processing and node execution remain opaque, the system can become less efficient and less credible as it scales.
DGrid's answer is to combine PoQ, on-chain records and payment rails so AI services move beyond being merely callable and become verifiable, billable and accountable. The article presents that as a key distinction between DGrid and projects built only around compute markets or model aggregation.
A first commercial answer, with open questions ahead
The article concludes that DGrid has shown one thing with more than $23 million in Genesis membership revenue and a product stack that is taking shape: verifiable decentralized AI is no longer only a long-dated concept and has begun to enter a stage of real payment and commercial testing.
It also says the path still requires long-term work. Lowering latency and cost while preserving verification, integrating smoothly with existing enterprise systems, and balancing membership growth with token incentives will all shape how far DGrid can go from here.
The piece ends on a broader market observation. It says the market is no longer satisfied with how big an "AI + Crypto" story can sound. What matters now is how much real revenue it can generate, how many real users it can retain and how many real problems it can solve. In that framing, $23 million is DGrid's first answer. What comes next is whether the model can extend beyond Genesis memberships into a larger developer ecosystem, model marketplace and enterprise-grade use cases.