2. The Volunteer Computing Network

GPU resources today are a commodity that are increasingly owned by hyperscalers and big tech companies, such as Microsoft, Amazon, Google, amongst other smaller but by no means insubstantial well-funded data centre players and startups in the industry.

As AI has captured the zeitgeist and becomes the first major innovation tech delivers to the world in over a decade, demand for high performance computing is set to increase exponentially, and as such access to the such machines is becoming increasingly expensive.

There exists, however, a vast untapped resource in the world today: individual GPUs, or small clusters of GPUs owned by individuals, whether for gaming, video rendering or crypto mining sitting either underutilized or idle, not yielding anything for their owners. In the crypto example, the move from POW (Proof of Work) to POS (Proof of Stake) has made large numbers of machines 'obsolete' in their originally intended use cases. However, these machines can provide great value in machine learning model training and inference.

NetMind power is built upon the concept of Volunteer Computing. Volunteer Computing is a system that allows owners of these GPUs that are underutilized or otherwise sitting idle, to contribute them to the NetMind Power network in return for rewards (more details are given in the NetMind Token section of this white paper).

This network of globally distributed GPUs provides the computing power necessary to operate the Training and Inference platforms, which form the basis of the user facing features of NetMind Power.

The platform will route training and inference requests to the most suitable Volunteer Computing nodes on the network, depending on the needs of the training and inference job in question.