Cryptominers have enriched themselves on AI technologies
Date: 2023-07-04 Author: Karina Ziganova Categories: BUSINESS
The demand for high-performance chips, provoked by the development of artificial intelligence (AI), has given new life to cryptocurrency mining technology
Two years ago, during a bull market, miners tried to build and buy hardware equipped with GPUs to mine cryptocurrency. However, the collapse of the market has led to the fact that many devices have stopped using. With the growing demand for AI technology, miners are extracting GPUs from idle machines and reselling them for use in AI.
Despite the fact that converting mining rigs is not always easy or cheap, this option is more affordable compared to AI infrastructure from leading mining companies.
What the experts think
According to Vipul Veda Prakash, CEO of alternative cloud provider Together, about 20% of the chips released can be repurposed to train AI models.
Saurabh Vij built an artificial intelligence business on chips that were previously used for mining. His startup Monster has access to more than 30,000 chips — most of which were once used for mining — that its customers can rent.
However, the head of CoreWeave, Michael Intrator, believes that miners will need to prepare for radical and expensive upgrades when working with AI.
"Most miners don't realize how much they need to invest in machines so that they can handle AI workloads," said entrepreneur Mark D'Aria.
A device with lower-end graphics cards requires additional components worth up to $5,000 to give it the processing power to perform AI tasks, he said.
Nvidia switches to AI
Leading U.S. GPU maker Nvidia has also switched to artificial intelligence. The company believes that other products, such as the artificial intelligence chatbot ChatGPT, are more useful than cryptocurrency mining.
At the same time, the company is already rapidly making money on the wave of hype around AI. In particular, Microsoft bought tens of thousands of AI-focused processors from Nvidia to power the OpenAI platform workload. The company sold another 20,000 processors to Amazon, and 16,000 chips to Oracle.
