Recently, the narrative in the Web3 space has begun to shift towards “AI +,” especially following the OpenAI launch of Sora. Many have clearly sensed that AI is not just a fleeting trend within a hype cycle, but truly has the potential to become the next foundational technology layer, much like mobile internet did. Ultimately, whether it is text generation, image creation, or large model inference and training, the essence of AI still revolves around one key term: computing power.
This is precisely where the problem lies. The AI services we use today may appear “free” and “intelligent,” but behind the scenes they are built on an extremely centralized resource structure: high-performance GPUs are monopolized by major corporations, leaving ordinary developers with little access; data sovereignty is lost the moment data is uploaded, and “energy consumption anxiety” has become a new source of digital pollution. Web3 has always claimed to “reshape value relationships,” so is there any project in the AI space that offers a genuinely sincere solution?
Coincidentally, I recently tried out a new project called Gplusck. Rather than positioning itself as an “AI platform,” Gplusck introduces the concept of a “decentralized AI operating system,” which I found quite intriguing. I would like to share my experiences and insights from using and exploring the project.
Initially, I was drawn in by its “carbon footprint NFT” feature. Simply put, each time you contribute computing power, the system generates a traceable carbon certificate based on your local electricity type (such as solar, hydro, or thermal power). This carbon certificate can be traded, destroyed, or used as a green incentive in DeFi participation. For the first time in Web3, I saw “green energy” not just as a slogan, but truly written into on-chain smart contracts.
Upon deeper exploration, I discovered that Gplusck has made several technical architectural innovations. For example, its computing power sharding mechanism breaks down large AI training tasks into “computational microcells,” which are intelligently scheduled based on node performance, latency, and energy type. Data transmission employs quantum encryption and invisible sharding, with built-in data compliance sandboxes for various jurisdictions—a significant plus for project compliance. Even model training incorporates AutoML and federated learning, ensuring data “never leaves the local environment.”
Moreover, Gplusck is not just about ideals. The project offers a low-code tool called GemForge. I tested it and found that you can genuinely build a “prediction market + AI oracle” dApp simply by dragging and dropping components—very user-friendly for non-technical users. I also browsed the official ModelVerse model library, which already provides over 800 pre-trained models for use, covering fields such as healthcare, finance, and content moderation.
Of course, there are still areas for improvement. Currently, the network does not have a large number of access nodes, and computational latency in some scenarios is less than ideal, especially with multi-node coordination—there remains room for optimization. The team claims they will deploy a 50MW solar-powered computing center in the Sahara this year, which I am looking forward to seeing in practice. Additionally, from a tokenomics perspective, GPUCK employs a typical deflationary and staking governance model. While the design is sound, its success depends on the actual computing power market and data trading scenarios—otherwise, it risks falling into the “lots of talk, little use” trap.
Overall, Gplusck is not the kind of project that immediately amazes me, but the more I explore, the more I appreciate its solid architecture and real-world potential. If you are interested in AI and computing power, or in Web3 infrastructure, I recommend taking some time to learn more about it.
Finally, a question for you: Have you been following any “AI + Web3” projects lately? Are there any you think are truly addressing fundamental issues? Feel free to share your thoughts in the comments!
(Disclaimer: This article reflects personal opinions only and does not constitute investment advice. DYOR.)
