With the development of artificial intelligence (AI) gaining more attention globally, cloud-based solutions as well as decentralized networks have been presented as solutions that can enable the advancement of such technology. The primary concern with cloud-based solutions is the high costs associated with the training of these AI models.
Messari’s research analyst Sami Kassab recently released a report which outlines blockchain projects and cryptocurrency solutions that could present viable alternatives to these high costs. The report dives into the mechanics behind these AI models and their training methodologies, as well as potential cost-cutting solutions.
According to the report, it cost Stability.AI, an open-source generative AI company, a whopping $50 million to use the 4,000 Nvidia graphics processing units (GPUs) running in Amazon Web Service’s cloud to train their models. As the demand is increasing for computer power for AI, its supply cannot keep up resulting in a bottleneck to the development of this technology.
To address this issue, crypto projects that are decentralized and utilize blockchain technology, in order to create a marketplace for people to transact with providers of AI, could reduce the cost of training a machine learning model.
Different crypto projects that could provide support for the stages of AI development, according to this report, include Filecoin, Arweave, Sia, Storj, Akash, Gensyn, Together, Cudos, iExec, Exabits, Flux, and Golem. For example, Cudos and iExec can currently provide support for the stage of deploying the model while other projects such as Akash, Exabits, Flux, and Golem are developing the capacity to further meet these needs.
The Decentralized Compute Networks that Kassab is referencing provide the potential to allow anyone with GPU or CPU hardware to lend their computing power to facilitate the development of AI models. This would incentivize the use of these resources, using correlating cryptocurrency as token rewards for the providers of such services.
From a larger standpoint, this will allow smaller firms and startups to enter the AI race without having to pay hefty fees for AI training, by taking advantage of cloud-like networks. This could potentially break down the monopolizing power held by large tech firms and allow for a much-needed competitive, collaborative, and transparent environment for the growth of AI.
In summary, the development of AI is now at the forefront of blockchain projects and the utilization of cryptocurrency. This report outlines the rise of decentralized compute networks that could open up opportunities for developers to access these services at lower costs, while also incentivizing the use of resources such as GPUs and CPUs. With the utilization of such technology, we are also beginning to see a potential break of the monopoly held by large technology firms.
