The surge in interest in artificial intelligence (AI) and machine learning (ML) has resulted in a scarcity of hardware resources and excessively high costs for cloud services. However, the reliance on centralized players may be challenged by decentralized infrastructure.
Harry Grieve, co-founder of the machine learning compute network Gensyn, spoke exclusively to Cointelegraph during the ETHGlobal event in London about the potential for peer-to-peer computing networks to disrupt Web2 services such as Amazon Web Services.
Gensyn is an upcoming decentralized network that will allow individuals to connect to various devices over the internet for training machine learning models. The company has received support from several Web3 venture capital firms and raised $50 million from Andreessen Horowitz in 2023.
Grieve believes that the network has significant potential as the internet evolves into a more dynamic representation of information, enabling “self-sovereignty and computational liberty online.”
Gensyn has been in development since 2020, with Grieve and co-founder Ben Fielding conducting research on machine learning computing for training and decentralized verifiable systems. They have been working to address a threefold problem with blockchain-based technology.
Grieve explains, “How can you connect with another device and train a machine learning model on that device where A) the device is untrusted, B) your training model cannot fit on that single device, and C) you want the scalability of the entire system and achieve economic outcomes as good as AWS?”
Gensyn’s litepaper describes the protocol as “a layer-1 trustless protocol for deep learning computation.” The network directly rewards participants for providing computing resources to the network and performing ML tasks.
Grieve acknowledges that verifying completed ML work is a challenge, as it requires expertise in complexity theory, game theory, cryptography, and optimization.
Gensyn draws inspiration from the ideals of the Bitcoin protocol, particularly the early days when users could mine BTC using smaller devices. Grieve plans to make Gensyn a tool that allows a wide range of users and hardware to contribute or access computing resources for ML training. However, the initial launch will focus on users with GPUs due to their ability to provide quick feedback.
Grieve envisions a future where individuals with a laptop can download Gensyn’s client and connect to the network, and he expects developers to build more user-friendly applications on top of Gensyn.
Additionally, Grieve highlights the potential of Apple Silicon chips, which could unlock significant global computing resources. Research shows that Apple M2 and M3 chips rival current-generation Nvidia RTX GPUs, providing two major benefits for protocols like Gensyn that can utilize a wide range of devices for their global supercluster.
Grieve believes that the future will see more powerful edge devices and emphasizes the importance of decentralization and device-agnostic verification.
In conclusion, the rise of decentralized infrastructure presents an opportunity to address the shortage of hardware resources and exorbitant cloud service costs in the AI and ML industry. Gensyn aims to provide a decentralized network for training machine learning models, while also rewarding participants for their contributions. With the potential integration of Apple Silicon chips, Gensyn could tap into a broader range of devices and further expand its capabilities.