Today we continue to talk about projects in the 200-300 range to see if we can find super potential stocks, because the market value in this range is low, around 200-500 million, and a 100-fold increase is very possible. The two projects we talked about today are from the Chia-Depin track, which is also a copycat of Filecoin, and the second project is Netmind, which is also a shared GPU project. There are too many GPU projects recently, and it seems to be a Chinese project, and the technical team should be from Tsinghua University.

1. Divide

This project was launched in 2021. It reached 1,500 US dollars at its peak and is now 30 US dollars. Just imagine how many lives were buried there. Moreover, it has hardly moved in the past year and is quite stable. Its current market value is 300 million US dollars, so this project has been criticized to death on the Internet.

Introduction

Founded by legendary programmer Bram Cohen, Chia Network can use cheap and redundant unused hard drive storage space to verify its blockchain. CHIA relies on file storage space for verification, and combining it with proof of time eliminates concerns about large-scale attacks.

Chia was founded in August 2017 to develop an improved blockchain and smart transaction platform. Chia aims to build an enterprise-grade digital currency. Chia is using the first new blockchain consensus algorithm since Bitcoin, it is called "Proof of Spacetime", it was created by Bram Cohen, the inventor of BitTorrent and the best network protocol engineer alive. Chialisp is Chia's new smart transaction programming language, which is powerful, easy to audit and secure. Currently available reference smart transactions include: atomic swaps, authorized payees, recoverable wallets, multi-signature wallets and price limit wallets.

The difference between Chia and Filecoin

The above picture is the conclusion derived by Google's AI, but I would like to add that mining Chia is easier than Filecoin, so when it goes online, a large number of people will flock to join the mining army.

Token Economy

The current supply is 31,501,075. Chia will pre-mine 21 million Chia when the network is launched. It has a mining mechanism, and the annual selling pressure is still relatively large.

Chia coin mining halving plan

64 chia: will be generated every 10 minutes for the first three years after the mainnet launch.

32 chia: From the fourth to sixth year after the mainnet launch, chia will be generated every ten minutes

16 chia: From the 7th to the 9th year after the mainnet launch, 16 chia will be produced every 10 minutes.

8 chia: From the 10th to the 12th year, every 10 minutes will generate

4 chia: After the twelfth year, every ten minutes every year

The following figure is a selling pressure plan in the white paper. You can see that it is expected to provide 22 years of mining rewards and releases. Currently, it should be 32 Chia. This year should be the fourth year, so the current release is 11.77 million + 1.68 million = 13 million. So you can see why the price of the currency is not growing, and the inflation is too high. And its mining cost is very low, so if everyone comes to mine, it will definitely not go up.

2. Netmind

If we look at this project over a one-year period, it has only doubled in value. In fact, you can think that it has not increased much. It is also a GPU sharing project with a current market value of US$300 million, ranking 250+. FDV is around US$1.5 billion.

Training Platform Introduction

The training platform is the foundation of NetMind Power's decentralized computing ecosystem. It allows users to leverage idle GPUs from global participants to train and fine-tune models in an efficient and cost-effective manner. The platform's architecture is built on advanced technologies and methods to enable distributed AI model training.

-Decentralized Architecture: The platform leverages a decentralized network of connected devices to distribute training workloads across multiple GPUs. This decentralized approach reduces reliance on centralized resources and keeps the cost of model training low.

-Decentralized Architecture: The platform leverages a decentralized network of connected devices to distribute training workloads across multiple GPUs. This decentralized approach reduces reliance on centralized resources and keeps the cost of model training low.

- Resource Allocation and Scheduling: NetMind Power's intelligent resource allocation system dynamically allocates tasks to the most suitable GPU in the network. This ensures optimal performance and reduces training time.

Data parallelism and model parallelism methods are used, depending on the specific needs of the AI ​​model being trained. After processing these small parts, the platform aggregates the results of each device to form the final trained AI model, ensuring the best learning effect. Techniques such as federated learning and parameter averaging are used to merge model updates from different devices while maintaining data privacy.

- Security and Privacy: The platform uses advanced encryption and secure multi-party computing technologies to ensure that user data is protected. In addition, techniques such as differential privacy can be applied to add an extra layer of protection for training data.

Introduction to the Inference Platform

The inference platform is designed to complement the training platform, providing users with a seamless way to deploy and run their own models, models created by others, and open source off-the-shelf models.

Main functions and technical details of the inference platform

-Model deployment: Users can deploy their trained AI models on the inference platform, making them accessible to other users and applications through APIs. The platform supports containerization, allowing AI models and their dependencies to be packaged into lightweight, portable containers that can be easily deployed to the network.

-Scalability: The inference platform is designed to handle different workloads, automatically expanding or shrinking according to demand. It uses distributed computing and load balancing technology to distribute inference workloads across multiple GPUs in the network, ensuring efficient use of resources and minimal latency.

-Cost optimization: By leveraging the decentralized nature of the platform and the idle resources of participants, the inference platform provides cost-effective access to computing power for running AI models. This reduces operating costs for users while maintaining high performance. NetMind Power's resource allocation algorithm dynamically allocates inference tasks to the most suitable GPU in the network based on factors such as computing power, latency, and availability to optimize cost and resource utilization.

-Security: The inference platform uses state-of-the-art security measures to protect AI models and processed data. This includes encryption technology, secure enclaves for model execution, and secure multi-party computing technology to maintain data privacy and model integrity during inference.

NetMind Chain

6.1 NetMind Chain

The NetMind Power network is built on and managed by the NetMind Chain blockchain. NetMind Chain decentralizes all tasks, transactions, and functions on the platform. It should be clear that the training process is performed on the local machine used for training, not on the blockchain.

NetMind Chain is based on the most mature Ethereum 2.0 technology, such as the POA (Proof of Authority) consensus mechanism and smart contracts that are fully compatible with the Ethereum chain.

6.3 NetMind Chain Protocol

NetMind Chain is a public blockchain based on the Ethereum protocol, adopts the POA (Proof of Authority) consensus algorithm, and is fully compatible with Solidity contracts.

Compared with POW (proof of work), POA does not need to consume a lot of resources to maintain network performance, making the maintenance cost of this platform extremely low. Unlike POA, in POS (proof of stake) and DPOS (delegated proof of stake) consensus algorithms, the more tokens a user owns, the more likely they are to become a node and be responsible for generating blocks. However, in POA, the validators responsible for processing transactions and verifying blocks must go through a series of reviews and must ensure their own reliability.

Dashboard

The data currently seen on the official website's dashboard shows that there are a total of 2098 GPUs and the GPU utilization rate is 91%.

Token Economy

This project was launched in February 24, with a total of 147 million, 33,450,208 in circulation, a circulation rate of 22%, and a current exchange rate of 8.8 US dollars. The peak was $16.0389 (2024-03-10). As can be seen from the token distribution above, mining rewards account for 39.9%, token issue is added later, the team accounts for 6.8%, strategic investment accounts for 5.1%, the ecosystem accounts for 11.2%, staking accounts for 11.2%, and GPU contribution accounts for 3.3%. However, this wave of additional issuance has a bad impact, because what is the difference between arbitrary additional issuance and the concept of currency in WEB2? This is what web3 hates the most.

Finally, let me summarize these two projects. The first project, Chia, actually has a good concept and good technical capabilities, but the annual emission is too large, so it is difficult to compare prices. In addition, so many people were buried in the previous wave, and there is no chance to get out of this wave. The principle is the same as filecoin. The second project is another GPU project. There are too many such projects. The project team is not real-name. I checked the information and it seems to be a Chinese project. The technical aspect is not special. The number of GPUs is only 2,000, which is not good enough. This volume can only be counted as the volume of the third echelon, so these two projects are not very good overall. #内容挖矿