Author: TaxDAO-Nandin Wu, Leslie
In January 2024, digital asset fund management company CoinShares pointed out that as Bitcoin halved, more miners are expected to turn to artificial intelligence in energy-secure areas in pursuit of higher income. Among them, mining companies such as BitDigital (BTBT), Hive (HIVE) and Hut 8 (HUT) are already using artificial intelligence for profitable activities. At the same time, TeraWulf (WULF) and Core Scientific (CORZ) are planning to expand in the field of AI or have already carried out related activities. This article briefly sorts out the initiatives of mining companies to turn to artificial intelligence and cloud services.
1. Specific action strategies of different mining companies
1.1 BitDigital
On October 23, 2023, Bit Digital announced the launch of Bit Digital AI. Bit Digital AI is a new business line that will provide professional infrastructure to support generative artificial intelligence ("AI") workflows.
Currently, BitDigital has launched the operation of Bit Digital AI. According to the agreement between the company and its customers, Bit Digital will provide customers with rental services of a minimum of 1024 GPUs and a maximum of 4096 GPUs.
Bit Digital also purchased 132 FusionOne HPCs from xFusion Digital Technologies Co., Ltd. ("xFusion") for approximately US$35 million. The FusionOne HPC solution is an integrated HPC platform, and each configuration ordered by Bit Digital includes NVIDIA HGX H100 8-GPUs, for a total of 1056 GPUs. These GPUs will be delivered to the company by the end of 2023 and will be deployed in Tier-3 data centers.
1.2 Hive
According to relevant detailed reports, HPC HIVE's AI team is currently working with strategic partners across North America to advance the forefront of open source AI large language models (LLMs).
In addition, HIVE has acquired two new Nvidia GH200 Grace Hoppers, which are designed to "handle the world's most complex generative AI workloads, covering large language models, recommendation systems, and vector databases."
HIVE officials said: "In short, we have been building HIVE's GPU and AI business unit to become the ultimate AI infrastructure choice and "trump card" play. As a leader in the technology field, we believe that we are underestimated as a company with efficient Bitcoin mining operations and a thriving GPU computing business unit."
1.3 Hut 8
Hut 8 claims that it currently has more than 36,000 square feet of geographically diverse data center space and cloud capacity, connected to a power grid powered by a large amount of renewable energy and zero-emission resources, and is completely transforming traditional assets to create the first hybrid data center model to serve traditional high-performance computing and emerging gaming, artificial intelligence and machine learning industries.
1.4 Core Scientific
Core Scientific, which has been operating mining since 2017, began diversifying its business in 2019. The company owns and hosts Nvidia DGX systems and GPUs for AI computing, and has built and deployed specialized facilities dedicated to high-value computing applications at its data center campus in Dalton, Georgia. At the same time, Core Scientific is also working with cloud service provider CoreWeave to provide infrastructure for use cases such as machine learning.
These combined capabilities will support AI and high-performance computing workloads, with revenue expected to reach $100 million, although the company said the total potential revenue is much higher because they have a huge infrastructure that can host the most advanced GPU computing on the market.
1.5 Bitdeer
On November 10, 2023, Bitdeer reached a cooperation agreement with US chip giant Nvidia to become the first Cloud Service Provider (CSP) partner in Asia to provide DGX H100 SuperPOD cloud services. BitDeer will enjoy NVIDIA's priority supply rights and provide GPU cloud services for the development and training of AI models. The news was announced at a press conference at NVIDIA's Singapore office, showing BitDeer's emphasis on the title of "NVIDIA Partner". Founded in 2018, BitDeer is one of the world's largest owners of self-operated mining data centers. As of October 31, 2023, it owns 6 large data centers with a managed computing power of 21.2 EH/s.
BitDeer's AI development roadmap includes infrastructure construction, cloud platform, software support, and application API services. Through cooperation with NVIDIA, BitDeer will continue to invest in infrastructure construction such as GPU, DGX, SuperPod, and launch cloud computing platforms, artificial intelligence training and reasoning platforms, AI software tools, and application programming interfaces to provide all-round support for artificial intelligence development.
2. The impact of Bitcoin halving on the strategic adjustment of mining companies
At 8:09 Beijing time on April 20, 2024, Bitcoin successfully completed its fourth halving. This has brought new challenges to various mining companies. First, the halving of computing power means that the number of bitcoins obtained by miners will decrease, which directly affects the profitability of mining. Miners need to invest more resources and computing power to obtain the same number of bitcoins, which may cause some high-cost miners to have a decline in profitability or even exit the market.
On the other hand, the halving of computing power will also have an impact on the market supply and price of Bitcoin. As the mining reward is halved, the supply of Bitcoin will also decrease. Many people believe that this may drive up the price of Bitcoin.
In general, the halving of computing power not only affects the profitability of miners and the market competition environment, but also affects the market supply and price of Bitcoin, triggering market changes and adjustments to investment strategies.
In general, the impact of mining companies shifting their computing power to AI and cloud computing on various companies is multifaceted, mainly including the following aspects:
(1) Diversified income sources and risk diversification
Using computing power in the field of artificial intelligence can help companies diversify their income sources. By providing computing power services to AI service providers, companies can obtain stable income from the AI field and reduce their dependence on Bitcoin price fluctuations.
At the same time, by conducting business in different fields, companies can achieve risk diversification. At present, the profits of most mining companies are strongly related to the volatility of the crypto market, and the shift to AI and cloud computing can help mining companies obtain a part of stable income and resist the risks brought by business and cash flow fluctuations in the bear market.
(2) Seizing market share in the AI field
The market demand in the AI field is extensive and growing rapidly. By providing AI computing services, companies can attract more customer groups, expand market share, and further improve profitability. At the same time, as revenues fell across the board overnight, the strategic response of each miner and how they adapt are likely to determine who is ahead and who is behind.
(3) Cost considerations
Currently, the cryptocurrency market is very different from a few years ago. In the past, most mining activities were scattered around the world. However, in recent years, most mining activities have moved to the United States, which has intensified the competition for electricity in the United States. The emerging artificial intelligence industry is attracting a lot of capital, which makes it more likely for mining companies that turn to AI to obtain preferential electricity prices from utility companies.
(4) Improving technical reserves and human resources construction
Turning to the AI field requires more technical reserves and talent teams. The company's investment in the AI field can not only improve its own technical strength and competitiveness, but also promote the transformation of the company's talent team, promote team building and business development.
On the other hand, the mining company's turn to AI is not smooth sailing. The ASIC chips used by the current mainstream mining machines are specially optimized chips for mining algorithms. Directly deploying mining machines on AI cloud computing cannot obtain ideal computing power. This means that mining companies cannot "seamlessly connect" mining equipment to AI scenarios, but still need long-term strategic deployment. In addition, since mining itself requires a lot of computing power to form economies of scale, the increased difficulty of mining brought about by the halving of Bitcoin makes the economies of scale more prominent, which will make the mining business more and more concentrated in leading enterprises, and the motivation of small and medium-sized mining companies to seek transformation to the AI track will be stronger.
On the whole, by entering the AI track, mining companies can not only achieve business diversification, but also enhance their ability to resist the cyclicality of the cryptocurrency market. However, the transformation to this track is still a long process.