Decentralized Protocols: Paving the Path for the Future of Machine Learning Infrastructure?

The growing interest in artificial intelligence (AI) and machine learning (ML) has caused a shortage of hardware resources and expensive cloud service costs. Decentralized infrastructure could challenge the dominance of centralized players in this field. Harry Grieve, co-founder of machine learning compute network Gensyn, spoke to about the potential of peer-to-peer computing networks to disrupt services like Amazon Web Services. Gensyn is a decentralized network that allows people to connect to devices across the internet to train machine learning models. The company is backed by Web3 venture capital firms and received a $50 million investment from Andreessen Horowitz in 2023.

Gensyn has been under development since 2020, with Grieve and co-founder Ben Fielding researching machine learning computing and decentralized verifiable systems. They aimed to solve a threefold problem with blockchain-based technology: how to peer with an untrusted device, how to train a machine learning model on a device with limited capacity, and how to achieve scale and economic outcomes as good as Amazon Web Services. Gensyn’s litepaper describes the protocol as a trustless layer-1 protocol for deep learning computation. It rewards participants for providing computing resources and performing machine learning tasks.

The challenge for Gensyn is verifying completed machine learning work, which requires a decentralized consensus. This involves complexity theory, game theory, cryptography, and optimization. Grieve compares this to the principles behind Bitcoin, where Satoshi gave people the ability to generate money from a laptop and convert electricity into a valuable asset. Gensyn aims to make its network accessible to a wide range of users and hardware, but the initial launch will target users with more GPUs for quicker feedback.

Apple Silicon, the chips used in Apple devices, could provide significant global computing resources. Research shows that Apple M2 and M3 chips are comparable to mid-tier consumer Nvidia RTX GPUs. This is advantageous for protocols like Gensyn, as more people are likely to have a MacBook than a standalone GPU. The versatility of Apple Silicon chips, which can be emulated by other manufacturers, could lead to more powerful edge devices in the future. Decentralizing and verifying across multiple devices is crucial for the field of AI and ML.

The Solana-based decentralized network io.net plans to incorporate Apple Silicon chip hardware for its AI and ML services.