Original title: Bittensor: New Trends in Usage
Original author: Sami Kassab
Original translation: Joyce, BlockBeats
Editor's note: As an AI project supported by powerful investment institutions such as Polychain and DCG, Bittensor has attracted much attention, with a market value of over $4 billion, and is regarded as a leading project in the AI track. On April 11, Bittensor was listed on Binance, and TAO hit a high of $700 on the same day, driving a series of AI coins to rise. A month later, the price of TAO was $377, a drop of 47% from its high point.
Even the general recovery of the AI sector cannot stop Bittensor's decline. This is mainly due to the obvious advantages and disadvantages of Bittensor's "moat". The Subnet incentive mechanism that once amazed the community is currently caught in controversy such as "ineffective competition and low quality". Recently, some AI projects such as Myshell and Virtual Protocol have successively launched their subnets on Bittensor, bringing vitality to the Bittensor ecosystem, but it is still unknown whether they can help TAO return to its peak.
Sami Kassab, a member of O$$ Capiτal and former Messari researcher, has always been optimistic about Bittensor. Recently, Sami analyzed the advantages and development potential of the subnet mechanism from the perspective of the AI projects that have settled in Bittensor. BlockBeats compiled as follows:
Image generated using Corcel, powered by Subnet 19
Bittensor is often described as a decentralized platform for issuing "data commodities". While data commodities are often thought of as only computing, storage, and network resources, Bittensor covers much more than these categories. In a broader sense, a digital commodity network can refer to any network that provides standardized digital tasks or services and is governed by a clear and consistent incentive and verification framework.
This means that in addition to traditional subnets centered around web scraping, data storage, and cloud computing, Bittensor also supports subnets dedicated to specialized tasks and services, such as creating AI models for specific modalities, fine-tuning open source, and generating 3D content, images, and trading signal generation.
Bittensor incentivizes miners to provide specific services and adopts a standardized incentive and verification framework, which enables teams to become more creative in using Bittensor. Two emerging trends are:
1. Outsourcing technology innovation
2. As an incentive layer for independent networks
Outsourcing Technology Innovation
There has been a recent trend where cryptocurrency teams are leveraging Bittensor to outsource the development of the underlying technology that powers their product or service. Rather than maintaining an in-house R&D team, these entities are turning to Bittensor. Both centralized and decentralized projects are creating competitive marketplaces as subnets, incentivizing contributors to solve specific problems that they define.
OpenKaito Subnet
Take Kaito, for example, a centralized AI search engine for the crypto industry. Their goal is to make information in crypto more accessible by indexing crypto content and converting unstructured data into a searchable and actionable format.
There are many complexities in building a search engine, including data acquisition, indexing, ranking, and knowledge graph development. To address these challenges without maintaining a large internal R&D department, the Kaito team launched the OpenKaito subnet on Bittensor. Here, the challenge of search relevance is defined as a miner-validator problem. Miners on the subnet submit ranking results for search queries, and validators apply a reward model to evaluate the quality of these miner responses.
This approach enables Kaito to outsource important R&D tasks, leveraging the collective expertise of contributors with specific domain knowledge to build a decentralized search engine. Kaito's goal is to develop a search and analytics product on the subnet, which it intends to monetize.
MyShell and Virtual Subnets
MyShell and Virtual are two decentralized projects that adopt similar strategies. MyShell focuses on the AI consumer layer, allowing users to make personalized chatbots, and to enhance their chatbot interaction experience, the team plans to add voice capabilities. However, given that text-to-speech (TTS) technology is still in its infancy and there is a lack of solutions suitable for custom voice models, MyShell launched a subnet to incentivize the development of open source TTS models. This move allows them to shift their focus from machine learning problems to other important aspects of the network.
Virtual followed suit, but with a subnetwork focused on the development of an incentivized audio-to-animation model.
Why outsource to Bittensor?
Both MyShell and Virtual incentivize contributors through their protocols to contribute data and models to develop personas, custom chatbots, and complete other tasks critical to their platform products and services. So why do they use Bittensor to drive the development of key AI models that underpin their platforms, rather than doing it through their own protocols?
There could be several reasons:
Easier to attract contributors: Attracting experts with specific domain knowledge to contribute to early-stage projects is a challenge, especially machine learning experts. However, Bittensor has a strong brand and a wide network of miners/contributors with different expertise. Among these contributors are experts in machine learning, who can seamlessly choose to contribute to subnets of projects such as MyShell and Virtual to help them achieve their goals.
Immediate value for contributors: Contributors prefer to be rewarded immediately for their work in a valuable currency. For example, TakeMyShell does not have a token, and while it is possible to offer points to contributors, it is unlikely that serious contributors will commit to substantial work based solely on the promise of future tokens without knowing its potential value. Even in cases where small projects do have tokens, with Bittensor, contributors can earn TAO (a relatively mature token with a fair amount of liquidity), which allows contributors to be compensated immediately in a stable manner.
Serves as the incentive layer of the network
One of the biggest challenges in launching a new network is scaling the supply side (the pool of miners contributing resources) to critical mass before the demand side (users) start using the network’s services. Crypto networks have proven to be an effective solution to this chicken-and-egg dilemma by incentivizing supply sides through the existence and availability of tokens, even if they are not actively participating in user tasks.
However, as AI becomes more prevalent and teams building AI resource networks and general digital commodity networks proliferate, attracting miners and bootstrapping the supply side of the network is becoming increasingly challenging and competitive.
In this environment, Bittensor is uniquely positioned to become an external incentive layer for the network, allowing the network to easily bootstrap its supply side and focus solely on the execution layer of the protocol.
Case Study: Inference Labs
Inference Labs is working to bring AI on-chain through proof-of-inference verification models that leverage zero-knowledge (zk) technology through AVS on Eigenlayer. Importantly, they have also launched a subnet on Omron Bittensor specifically to bootstrap a zk prover and model reasoner for their protocol.
Essentially, Inference Labs is using Bittensor as the incentive layer for the supply side of its network in its initial phase.
The rationale behind leveraging Bittensor is simple: it is much easier to attract contributors to mine subnets on an existing network like Bittensor than to attract them to a new independent network. As mentioned above, Bittensor's ability to provide immediate value to contributors is a major selling point. In addition, the network has thousands of miners who have already contributed to various subnets, and since they are familiar with the resources and tasks required to mine different digital commodity networks, they can seamlessly opt-in to join new subnets.
Launching the subnet on Bittensor thus enabled Inference Labs to leverage an existing pool of skilled miners, accelerating the development and growth of its protocol. And accelerating its development. In just two weeks, assuming miners run at the minimum hardware requirements (which likely underestimates actual capacity), the subnet totaled 1900 CPU cores, 15 TB RAM, and 90 TB storage, positioning the subnet as the largest zkML computing cluster.
In the future, Inference Labs plans to internalize the incentive layer, where miners who contribute directly through its protocol will receive token incentives and network usage fees. However, even if Inference Labs transitions to its own incentive mechanism, the subnet on Bittensor will persist, continuing to replenish the protocol's native supply side indefinitely. In doing so, Inference Labs' network acts as an aggregator, sourcing zkML contributors from a variety of sources, including the Bittensor subnet.
While some networks may eventually choose to merge the incentive layer, others may choose to permanently delegate this function to Bittensor, allowing them to focus on the execution layer.
How do subnet dynamics affect the price of TAO?
In Bittensor, validators typically have exclusive access to the digital goods produced by miners (more on this here). When a team launches a subnet as an external incentive layer or outsources technical innovation of their network, the protocol or team has two options:
Becoming a validator – This involves acquiring TAO and staking it to a specific subnet. The network resources or access to services a validator receives on a subnet is generally determined by the stake they hold. For example, if a team owns 20% of the TAO stake of a subnet to run a validator, it will receive 20% of the resources of the corresponding subnet.
Paying existing subnet validators for resources – Alternatively, teams can choose to pay existing subnet validators for their subnet resources. This payment can be made in a variety of currencies, such as fiat or stablecoins, depending on the validator’s preference. Taoshi is developing a request network that will seamlessly enable validators to monetize their resources, allowing third parties to easily access a subnet’s goods through an API.
As the revenue potential of a subnet expands as demand for its resources or services increases, validators begin competing to acquire TAOs to ensure miners receive additional resources and priority, thereby gaining larger revenue streams and enhancing their operations.
Ultimately, both options will help drive demand for TAO within the Bittensor ecosystem. Given that TAO’s token supply is fixed, this growing demand could cause its value to rise.
As more and more excellent miners join Bittensor's mining network, more and more teams are attracted to start subnets to obtain the professional talents needed to solve their specific problems and obtain the digital resources they are looking for. This creates a positive cycle effect, where excellent miners compete for rewards to attract higher-quality subnet teams, which produce more valuable digital goods, ultimately increasing demand and prompting validators to compete for TAO to ensure that they obtain collateral and obtain corresponding resource allocations.
Final Thoughts
The prevailing theme among teams looking to launch subnets on Bittensor seems to be the desire to harness the power of the mining community. A network of experts in fields ranging from machine learning, data science, trading, cloud computing, resource allocation, and more, all looking for ways to leverage their skills and resources in exchange for a common currency. This seems to be Bittensor’s strongest advantage at this point.
What I find most interesting is that Bittensor can facilitate the development of projects that are not feasible as standalone networks. For example, users can launch specialized subnets to outsource specific components of their technology stack, just like deploying specialized microservices. For example, a decentralized social network could offload recommendation algorithms to Bittensor, similar to how projects currently rely on Bittensor to develop and reason about AI models across different modalities.
As future subnets become more accessible, the prospect of centralized and decentralized projects hosting specific parts of their tech stack on Bittensor becomes increasingly real and viable. I foresee a third category of use cases for Bittensor emerging in the near future.
Original link
