As a world-first, Shenzhen Data Exchange (SDE) facilitated a transaction that introduced decentralized, community-contributed AI data into real industrial application scenarios through Web3 infrastructure. As the largest national data trading platform for data marketization and cross-border circulation in China's digital economy, SDE provides a full range of services including compliance assurance, circulation support, supply and demand docking, and ecological development to help enterprises and users efficiently complete data transactions.

I have repeatedly emphasized in previous articles that data will become the next battleground in the global AI competition. This article will explore in depth how Shenzhen Institute of Data Science can take a key step in cross-border data collection through a commercialization model driven by decentralized AI (DeAI).

The bottleneck of AI: data

First, we must reiterate the main bottleneck currently facing the global AI industry: data shortage.

As more and more industries and companies rely on AI models to innovate, the demand for high-quality training data is rising sharply. This challenge affects all areas:

• Healthcare: AI-driven diagnostic systems require massive amounts of medical images and medical record data to accurately identify diseases.

• Autonomous driving: Autonomous vehicles require millions of miles of diverse driving data to safely handle complex real-world scenarios.

• Financial modeling: AI algorithms used for fraud detection or market prediction rely on large amounts of transaction data.

• Smart Manufacturing: High-resolution data such as images of equipment, materials, and defects are key to computer vision models in automation and quality control.
  
  

So, the core question becomes: How do we obtain such a huge amount of data on a large scale?

Traditional centralized data collection methods have many limitations:

• Geopolitical and regulatory restrictions make cross-border data sharing difficult;

• Data privacy regulations such as the European GDPR impose strict requirements on data collection and use;

• Centralized data sets lack diversity, which can easily lead to bias in AI models;

• Access to data is unequal, with only large tech companies able to control high-value data lakes, creating a “walled garden” effect.

Pragmatic breakthroughs achieved by DeepMind using DeAI

While the chip war has dominated the headlines, an equally important data war is quietly underway. Recently, Shenzhen Institute of Data Science and Technology facilitated a far-reaching business transaction between Shenzhen Intellifusion, a leading listed AI company in China, and OORT, a decentralized AI solution provider.

Intellifusion has been developing industry-specific AI solutions to enhance its smart factory capabilities, including the need for industrial data such as specialized protective masks and ventilation ducts in small spaces. OORT has enabled the collection of this data through its product OORT DataHub, which distributes data tasks to a global community in more than 130 countries. Participants can contribute data and receive cryptocurrency incentives, a mechanism that is not possible with traditional banks or Web2 platforms.

This transaction marks the realization of the first truly commercially viable, global decentralized data collection model, bringing a major breakthrough in cross-border data services.

Limitations of traditional platforms such as ADX

While platforms such as Amazon’s AWS Data Exchange (ADX) are well known, they have significant limitations in driving the next phase of AI globalization:

• Lack of open, community-contributed data: ADX is primarily a B2B market dominated by commercial enterprises, excluding valuable data from developing regions, local communities, academic research, and open source organizations. For example, citizen science data on water quality in rural India or agricultural data collected by farmers in Africa can be extremely valuable for training AI models.

• Cross-border compliance challenges: In jurisdictions with strict data localization laws such as China, India and the European Union, cross-border data transfers face many restrictions.

• Centralized access model: Users must rely on AWS accounts and Amazon's infrastructure and policies. There is a lack of built-in data ownership verification or self-sovereign identity, and publishers are forced to rely on AWS for subscription and billing management.

• Global underrepresentation: Data providers are mainly concentrated in the United States and Europe, while small and medium-sized enterprises and researchers in Africa, Latin America, and Southeast Asia, as well as indigenous data holders and community/device-generated data (such as rural IoT data) are seriously underestimated.

• Limited interoperability: Although ADX is well integrated with the AWS ecosystem, it lacks open interoperability with other cloud platforms and Web3 tools, which hinders its integration with Google Cloud, IPFS/Filecoin, decentralized computing layers, and native blockchain applications.

DeAI moves beyond hype and toward practical application

Against this backdrop, the field of decentralized AI (DeAI) is making significant progress, working to build a more open future in an AI landscape dominated by large companies.

Recently, two DeAI alliances were announced on the same day:

First is HumanAIx, initiated by 13 Web3 entities including OORT, YGG, NEO and io.net, which launched an open protocol designed to seamlessly connect all parties. Each member provides key components - verification, storage, computing and data, and jointly builds a permissionless, scalable and verifiable DeAI infrastructure. The protocol adopts a three-layer architecture - interface layer, protocol layer (integrating computing, storage and data) and security layer, combining industry experience to create an open DeAI development environment.

At the same time, another group of Web3 leaders, including NEAR, Aethir and Coinbase, jointly formed the Open Agents Alliance (OAA), which is committed to ensuring the security, open source, economical and fair use of AI.

Despite the current sluggish crypto market and the hype bubble that AI is inevitably experiencing, it is gratifying that serious players in the industry have begun to advance far-reaching and sustainable solutions. Ultimately, only projects with viable business models will survive. Shenzhen Institute of Data has taken a key step with decentralized data collection, marking a shift in the global data landscape. It reminds us that it is time to rethink the way data is collected, verified, and managed in the development of AI.