Key facts:
- Bitcoin mining companies dedicate their GPU power to AI servers.
- The AI industry is one of the industries with the greatest economic projection today.
The growth of the artificial intelligence (AI) industry demands more resources from chip manufacturers and data centers every day. Servers for AI applications are proliferating at breakneck speed. Hence, Bitcoin miners respond to the call of an expanding business, which is emerging as one of the industries with the greatest projection in the coming years.
In this sense, the emergence of generative artificial intelligence models represents an opportunity for many cryptocurrency miners who have GPU farms and who, since the end of mining on Ethereum, have been overcoming obstacles to generate profits. In fact, some of these miners found it necessary to turn off their equipment or mine other cryptocurrencies. This is the case of Hive Digital, a company that mines Bitcoin and that in the past had diversified its activity to mine Ethereum, which with only 2 % of the 38,000 graphics cards it owns dedicated to AI servers obtained profits of USD 250,000 in the first trimester.
This is also the case of Northern Data, the largest Bitcoin mining company in Europe, which through an investment of EUR 575 million from Tether will acquire graphics cards to use as servers for AI in the cloud. And so, others like Bit Dig i tal, Applied Digital, and Bitdeer also join the wave.
All of these companies have reasons to get into this business. The main one could be the precipitous drop in Bitcoin mining profitability, which has decreased by at least 85% since November 2021 with the network’s difficulty and hashrate increasing and prices falling from their all-time high.
To understand how companies that mine Bitcoin are making a leap into the generative artificial intelligence industry, one idea about this activity must be demystified: it is not just about connecting equipment to a network and sitting back and waiting for rewards. On the contrary, a large majority of people involved in Bitcoin mining specialize in systems engineering (with or without an academic degree). And this quality is what ultimately provides the skills for numerous companies to diversify their business and venture into other areas that require data center operations.
This leads me to think that it is precisely the Bitcoin mining companies that could take over from traditional data centers, which currently face numerous challenges to sustain an industry that demands more energy and computing power every day.
First, Bitcoin miners have developed better strategies to optimize their operations by using renewable energy. In addition, many of these companies generate their own electricity and even have the capacity to supply electricity to towns near the mining farms. In a world where resources are scarce, this could be an advantage.
Here other factors come into play that contribute to the emergence of this business opportunity. The growth in demand for servers to run artificial intelligence applications and the growth of the AI market, which is estimated to reach 300 billion by 2025, could be among the most notable.
This projection generates different problems, both for data center operators and for manufacturers of semiconductor chips (such as those used by GPUs) or integrated circuits (such as those used by ASIC devices to mine Bitcoin).
Specifically, one of the problems experienced by the GPU manufacturing industry arises from the difficulty of predicting how far the development of more powerful chips that could serve the AI industry (and others such as video games or mining) will go. cryptocurrencies), because manufacturers are reaching the physical limit of the materials to create photolithography plates or wafers, from which the chips are extracted. The most efficient and powerful devices of this type today are those of 2 and 3 nanometers, manufactured by companies such as IBM and TSMC, and the complexity of producing the substrate for these chips is such a complex process, and it is in so few hands. , which generated shortages between 2020 and 2022, a bottleneck in the industry that is difficult to overcome.
In 2023, the outlook took a turn, as the shortage eased. Among the reasons is the decrease in demand for GPUs with the end of mining on Ethereum. Furthermore, with the growth of second-hand markets for graphics cards, which were no longer used to mine cryptocurrencies, the chips began to drop in price and produced the current situation. Companies that mine Bitcoin (and data centers) are taking advantage of favorable market conditions to acquire cheap equipment and GPU manufacturers, such as Nvidia, are suffering a decline in their profits, as reported by Market Times.
Another factor that we can take into account to understand why Bitcoin miners are entering the artificial intelligence business has to do with the capabilities of data centers today. Data center facilities will also reach their own limits to sustainably meet the computational requirements of AI applications.
Let’s take into account that it is estimated that data centers currently demand between 200 Tw/h and 250 Tw/h, added together, the equivalent of 1% of all the electricity we consume on the planet (these estimates could even be doubled and according to estimates would grow up to 30% in just 7 years). Furthermore, the operations of these buildings full of computers require enormous amounts of water to cool the equipment (up to 25 million liters for each building) and, in that sense, we are also at the limit of what is sustainable.
Greater water and electricity requirements to meet demand could mean in the future that data center buildings will have to make better use of existing resources, due to global environmental policies and the disposition of those resources on the planet, something that Bitcoin miners have been working on for some time.
A final factor to take into account is related to the shortage of people trained to operate in data centers. The engineers and operators of many companies that mine Bitcoin have the potential to work on two fronts, an alternative so that the sector has options in the near future.
