In the wake of a decline in cryptocurrency mining, many cryptocurrency miners have found new opportunities in the booming field of artificial intelligence (AI). As the demand for high-end chips rises in the AI industry, companies like Satoshi Spain have repurposed their idle crypto mining rigs to handle AI workloads for startups, universities, and individual developers across Europe. This trend, known as “mining 2.0,” allows crypto miners to leverage their existing equipment and tap into the AI boom.
Converting mining rigs to handle AI boom
Graphics processing units (GPUs), originally designed for gaming, are crucial components in both cryptocurrency mining and AI systems. These powerful chips are capable of performing the intensive calculations required for cryptocurrency mining as well as the computational workloads involved in training and running AI models that generate complex text and images. The viral success of AI-powered chatbots like ChatGPT has further fueled the demand for GPUs, leading to a shortage in supply and driving companies to seek alternative solutions.
During the cryptocurrency upsurge, miners rushed to build or purchase hardware fitted with GPUs to mine digital coins. However, with the subsequent decline in cryptocurrency prices and changes in mining dynamics, many of these machines were left idle, creating a significant untapped resource. Startups and miners are now repurposing these “dark GPUs” for AI workloads. Some companies reprogram and repurpose their machines, while others extract the GPUs from dormant rigs and resell them for use in AI computers.
Challenges and potential of crypto miners in the AI industry
While the mechanics of cryptocurrency mining differ from AI training, the refurbished mining rigs present an affordable and accessible option compared to the infrastructure provided by major cloud companies. Startups and universities, often facing challenges in obtaining AI computing power from the cloud giants like Microsoft and Amazon, are turning to repurposed mining rigs as an alternative. The high requirements for training AI models, coupled with the surge in user interest, have strained even the largest players in the AI industry.
The recent change in the mechanics of mining Ethereum, the second-largest cryptocurrency after Bitcoin, has led to a surplus of used chips in the market. Approximately 20% of these freed-up GPUs can be repurposed for training AI models, offering a viable solution for the surging demand. Together, an alternative cloud provider is leveraging former crypto miners and their GPUs to build virtual server farms. By leasing thousands of GPUs, Together’s cloud service supports AI developers. This approach provides cost-effective computing power, particularly for startups that require more affordable options than mature cloud infrastructure.
A new lease on life for crypto miners
Entrepreneurs like Demi Guo, founder of Mellis AI, have turned to repurposed hardware available through cloud providers like Together due to pricing and capacity limitations at leading cloud providers. The availability of repurposed chips offers a more cost-effective solution. Similarly, startups like Monster, led by Saurabh Vij, offer leasing options for the vast number of GPUs acquired from Ethereum miners. These GPUs, often the most expensive parts in mining machines, can now be put to use in AI applications.
However, it is important to note that repurposing mining rigs for AI workloads may require significant investments and upgrades. While mining rigs are optimized for cryptocurrency mining, they may not possess the computing power for AI tasks. Companies like CoreWeave have been refurbishing mining rigs since 2019 for image rendering, science analysis, and AI applications. Rebuilding the machines with high-end Nvidia data center chips involves substantial investment, but the excitement surrounding AI has enabled companies like CoreWeave to secure substantial funding.
As the cryptocurrency mining industry experienced a decline, crypto miners have found a new lease on life in the AI industry. Repurposing idle mining rigs for AI workloads has become a viable option, leveraging the power of GPUs originally used for mining. The shortage of high-end chips and the demand for AI computing power have created opportunities for miners and startups to repurpose these GPUs. However, the transition to AI requires careful consideration of upgrades and investments to meet the computational demands of AI tasks.
