ROKO DAO Governance Proposal: Dynamic TAO Allocation of 884 TAO

Proposal No. [39]

Overview of Dynamic TAO (dTAO)

Dynamic TAO (dTAO) represents a fundamental shift in Bittensor’s tokenomics and governance model, introducing a market-driven approach to token emissions. In this framework, TAO holders can stake into specific subnets in exchange for that subnet’s native “alpha” tokens. This structure enables efficient capital allocation, directing token emissions toward subnets that provide real utility and demand-driven services, rather than distributing rewards without performance-based considerations. Key advantages of this model include:

• Market-Driven Emission Model – Token emissions are determined by supply and demand for subnet services rather than a fixed schedule. This ensures that rewards flow to active and valuable subnets, avoiding inefficient allocations to idle validators.

• Incentivizing High-Utility Subnets – Each subnet operates its own micro-economy via its alpha token. As a subnet demonstrates growth and usage, its alpha token gains value, incentivizing further development and innovation. The most useful subnets attract the most stake, creating a self-reinforcing ecosystem of high-performance AI services.

• Permission-less Scaling – Subnets are empowered to scale organically based on demand, eliminating the need for centralized intervention in their economic growth.

In summary, dTAO optimizes TAO emissions by rewarding valuable AI subnets, leading to a more efficient and competitive network that promotes sustained long-term development.

Allocation Strategy

This chart shows how TAO emissions gradually shift from the root network (blue) to subnet alpha tokens (red) over time. Initially, most newly minted TAO goes to the root, but as more subnets register, a growing share of emissions flows into the subnets.

ROKO DAO aims to balance stable returns and growth exposure by allocating its 884 TAO stake under the dTAO framework as follows:

• Root Subnet (75% = 663 TAO): The majority of the stake will remain in the root network (Subnet-0) to maintain a stable yield and support Bittensor’s core. Keeping 75% in the root subnet provides the DAO with consistent staking rewards from the base network and helps secure the overall protocol. This large root allocation ensures the DAO can benefit from front-loaded TAO emissions while still supporting the main chain’s operations and security.

The remaining 25% (221 TAO) will be allocated across selected subnets as follows:

• SN51 – Celium (5% = 44.2 TAO): Subnet-51 (Celium) is chosen for its strong AI compute utility and rapid growth potential. Celium is a decentralized marketplace for high-performance GPU computing, enabling users to rent out and access industrial-grade AI compute power. It has quickly demonstrated traction – for example, SN51 attracted over 500 H100 GPUs in just its first month of operation(TAO Times #13: Bittensor's Biggest Week). This indicates massive demand for its service and suggests significant upside as AI workloads increasingly seek scalable, decentralized compute. By allocating a portion of our stake to SN51’s alpha token, the DAO will gain exposure to the growing AI compute economy and support an essential infrastructure subnet in the ecosystem.

• SN64 – Chutes (5% = 44.2 TAO): Subnet-64 (Chutes) is a serverless AI computing platform that allows developers to deploy and run cutting-edge AI models in seconds. Chutes provides on-demand “compute containers” for AI (from large language models to image processors) in a decentralized manner. This subnet has shown extraordinary usage metrics – it recently processed over 1 billion organic AI tokens in a single day via its platform, highlighting both its performance and adoption. The DAO’s stake in SN64’s alpha token capitalizes on the projected growth of serverless AI services and supports a subnet that is pushing the limits of scalable AI throughput. SN64’s focus on efficient, scalable model deployment makes it a cornerstone for future AI dApps in Bittensor.

• SN19 – Nineteen (5% = 44.2 TAO): Subnet-19 (Nineteen) is a decentralized AI inference subnet that enables permission-less model execution without reliance on centralized APIs. It provides access to cutting-edge open-source AI models, including large language models (LLMs) and image generation models, allowing users to run inference in a fully decentralized manner. Miners operate their own models, ensuring censorship resistance, while validators autonomously verify responses to maintain reliability. The subnet’s performance-based incentives reward miners based on response speed, volume, and accuracy, fostering a competitive AI marketplace. By allocating a portion of our stake to SN19’s alpha token, the DAO supports a subnet that is pioneering trustless, scalable AI inference, expanding decentralized access to high-quality AI services in Bittensor.

• SN4 – Targon (5% = 44.2 TAO): Subnet-4 (Targon) provides redundant, deterministic verification for large language model outputs. In practice, SN4 takes queries and cross-verifies responses against trusted data or multiple models to ensure AI-generated text is accurate and consistent. This service effectively validates and enhances LLM responses in real-time, which is vital for deploying AI in mission-critical or sensitive applications. By allocating stake to SN4’s alpha token, the DAO is backing a subnet that improves the reliability and integrity of AI outputs across the network. We anticipate SN4’s utility will grow as more applications require trustworthy AI responses (reducing issues like hallucinations from LLMs). Supporting SN4 aligns with the DAO’s interest in a robust, trustworthy AI ecosystem, and the subnet’s contributions to AI output quality can translate into steady value appreciation under dTAO’s reward mechanism.

• SN27 – Neural Internet (5% = 44.2 TAO): Subnet-27, Neural Internet is a decentralized AI compute network that transforms GPU power into a scalable digital commodity. It provides distributed, on-demand GPU cloud resources for AI acceleration, allowing seamless integration across multiple cloud platforms. Unlike traditional compute services, SN27 offers permission-less access to AI computing, ensuring decentralized, trust-less validation of workloads. By leveraging blockchain-backed compute verification, SN27 creates an efficient and transparent marketplace for AI developers seeking scalable, cost-efficient infrastructure. This investment aligns with ROKO DAO’s strategy to support subnets that enhance AI scalability, accessibility, and decentralization within the Bittensor ecosystem.

Execution of the Dollar-Cost Averaging (DCA) Plan

To efficiently enter these subnet allocations, ROKO DAO will deploy 221 TAO over a structured 2-month DCA schedule:

• TAO will be allocated in smaller, frequent tranches, rather than large bulk transactions, optimizing entry prices and minimizing market disruption.

• Purchases will be executed on a daily or weekly basis, adjusting dynamically based on market conditions and subnet advancements to maximize returns.

• Total TAO allocated per subnet over 2 months: 44.2 TAO

• Total TAO deployed per month across all subnets: 110.5 TAO

• Approximate weekly deployment per subnet: ~5.525-11.05 TAO

• Approximate daily deployment per subnet (if used): ~0.79-1.57 TAO

Governance Execution

Authorization: Upon approval, the DAO’s operations team and treasury multi-sig will be granted authority to execute the 2-month DCA plan. This includes reallocating 884 TAO, ensuring fund movements are handled transparently and securely via the DAO’s standard multi-sig approval process.

Implementation: The reallocation occurs in two phases:

• Initial Allocation – 884 TAO is split, with 663 TAO remaining in root staking and 221 TAO set aside for subnet DCA purchases.

• DCA Execution – TAO is gradually deployed into subnet alpha tokens at daily or weekly intervals, tracked and audited for full transparency.

No Interim Votes or Changes: The strategy is locked in for 2 months, with no additional governance votes during execution. Only in extreme cases (e.g., a critical security vulnerability) would modifications be considered via separate governance action. Post-Execution Review: After 2 months, ROKO DAO will assess performance, reviewing:

• Root vs. subnet allocations – Evaluating yield stability and subnet growth potential.
• Subnet performance metrics – Analyzing demand and token appreciation for SN51, SN64, SN19, SN4, and SN27.
• Future adjustments – Determining whether reallocation is necessary based on performance insights.

Structure & Clarity: This proposal establishes a clear, executable investment plan, ensuring a strategic and value-driven deployment of TAO that aligns with Bittensor’s evolving AI ecosystem.

Conclusion By approving this proposal, ROKO DAO commits to:

• A 75/25 stake split between the root subnet and selected AI subnets.
• A structured 2-month DCA investment schedule, with frequent and optimized TAO purchases.
• A transparent execution process, overseen by trusted multi-sig signers.

This proposal positions ROKO DAO for both stable staking yields and exposure to high-growth AI subnets, strengthening the DAO’s role in Bittensor’s next-generation AI economy.

Resources

Bittensor’s Official Documentation Understanding dTAO Comprehensive Analysis of dTAO