Analysis of Recent Trends and Popular Projects in the Crypto+AI Track

In the past month, the Crypto+AI sector has shown three significant trend changes:

1. The project's technical path is more pragmatic, focusing on performance data rather than solely relying on conceptual packaging.

2. Vertical segmentation scenarios are becoming the focus of expansion, and specialized AI is replacing the position of generalized AI.

3. Capital increasingly values the validation of business models, and projects with cash flow are obviously more favored.

Here is a brief introduction and analysis of several popular projects:

Decentralized AI Model Evaluation Platform

The platform completed a $33 million seed round financing in June, led by well-known investment institutions, with participation from several industry experts.

The platform applies human subjective judgment advantages to the shortcomings of AI evaluation. By using artificial crowdsourcing to score over 500 large models, user feedback can be converted into cash. It has attracted several well-known AI companies to purchase data, creating a real cash flow.

The business model of this project is relatively clear and is not purely a money-burning model. However, preventing fake orders is a significant challenge that requires continuous optimization of the anti-witch attack algorithm. From the perspective of financing scale, capital clearly prefers projects with monetization verification.

Decentralized AI Computing Network

The project completed a $10 million seed round financing in June, led by two well-known capital institutions.

The project has achieved a certain market consensus in the DePIN field of a certain public chain through a browser plugin. Team members come from several well-known Web3 projects. The newly launched data transmission protocol and inference engine have made substantial explorations in edge computing and data verifiability, reducing latency by 40% and supporting access from heterogeneous devices.

The direction of the project is very accurate, precisely hitting the trend of "localization sinking" in AI. However, when dealing with complex tasks, it still needs to compete with centralized platforms in terms of efficiency, and the stability of edge nodes is also a problem that needs to be solved. However, edge computing is both a new demand generated by the internal competition of Web2 AI and an advantage of the distributed framework of Web3 AI, making it worth looking forward to its implementation through specific products driven by actual performance.

Decentralized AI Data Infrastructure Platform

The platform incentivizes global users to contribute multi-domain data (including medical, autonomous driving, voice, etc.) through tokens, accumulating revenues of over 14 million USD and establishing a network of millions of data contributors.

Technically, the platform integrates zero-knowledge proof verification and Byzantine fault tolerance consensus algorithms to ensure data quality, and it also uses privacy computing technology from a well-known cloud service provider to meet compliance requirements. Notably, they have also launched brainwave collection devices, achieving an extension from software to hardware. The economic model design is quite good, allowing users to earn $16 and 500,000 points for 10 hours of voice annotation, while the cost for enterprises subscribing to data services can be reduced by 45%.

The greatest value of this project lies in its alignment with the real demands of AI data labeling, especially in fields such as healthcare and autonomous driving, where data quality and compliance requirements are extremely high. However, a 20% error rate is still relatively high compared to the 10% of traditional platforms, and the fluctuation in data quality is an issue that needs continuous resolution. While the direction of brain-computer interfaces is full of imaginative potential, the execution difficulty is also significant.

Distributed Computing Network on a Public Blockchain

The project completed $10.8 million in financing in June, led by an investment institution.

By aggregating idle GPU resources through dynamic sharding technology, it supports the inference of large language models, costing 40% less than a certain well-known cloud service provider. The design of its tokenized data trading is very creative, directly transforming computing power contributors into stakeholders, which can incentivize more people to participate in the network.

This is a typical "aggregated idle resource" model, which makes logical sense. However, a 15% cross-chain validation error rate is indeed too high, and technical stability still needs to be improved. Nevertheless, it does have advantages in scenarios such as 3D rendering, where real-time requirements are not high. The key is whether the error rate can be reduced; otherwise, no matter how good the business model is, it will be dragged down by technical issues.

AI-driven Cryptocurrency High-Frequency Trading Platform

The platform completed a seed round financing of 3.38 million USD in June, led by a cryptocurrency company.

The core technology of the platform can dynamically optimize trading paths, reduce slippage, and has been tested to improve efficiency by 30%. It aligns with a certain AI financial trend and has found a breakthrough in the relatively blank subfield of DeFi quantitative trading, filling a market demand.

The project direction is fine; DeFi indeed requires smarter trading tools. However, high-frequency trading has very high demands for latency and accuracy, and the real-time synergy of AI predictions and on-chain execution still needs further validation. In addition, certain types of attacks pose a significant risk, and technical protective measures must keep up.