In the rapidly evolving landscape of technology, two sectors are capturing significant attention: AI/HPC compute and Bitcoin mining. A couple of weeks ago, Digital Mining Solutions published an article delving into the key similarities and differences between these two domains, highlighting their unique infrastructure requirements for optimal efficiency. But what lies beneath the surface of their operational frameworks? How do their business models compare? In this article, we will take a closer look at the business models of Bitcoin mining and AI/HPC.
Bitcoin Mining Business Models
Self-Mining
Hosting Service Provider
Cloud Mining
AI/HPC Business Models
In-house Data Center to Power Internal Services
Co-Location Services
Cloud Computing
Infrastructure-as-a-Service (IaaS)
Platform-as-a-Service (PaaS)
AI-as-a-Service (AIaaS)
PREMIUM Members Only
Unpacking the AI/HPC Strategies of Public Miners
Hut 8
Core Scientific
A couple of weeks ago, Digital Mining Solutions published an article exploring the key similarities and differences between AI/HPC computing and Bitcoin mining, along with the unique infrastructure requirements for efficiently operating these two types of data centers. Make sure to check it out.
Bitcoin Mining Business Models
The primary business models in Bitcoin mining include self-mining, offering hosting services, and engaging in cloud mining. But what do these models entail? Let’s take a closer look at each model and uncover the nuances.
1. Self-Mining
Self-mining or direct mining means mining Bitcoin for your own account. A self-miner owns the ASIC mining hardware and put these to work by operating or utilizing hosting services. Self mining should not be confused with solo mining. A solo miner operates mining hardware independently without joining a mining pool. In solo mining, all rewards from successfully mining a block go entirely to the miner, without sharing it with others. A self-miner can either join a mining pool or solo mine independently. The size of a self-miners can vary from someone who owns a single machine to a multi-megawatt operation.
Self-mining businesses can be operated in two ways: by operating a mining facility or by utilizing hosting services to operate your mining hardware. Both options have their advantages and challenges. Choosing between hosting services and running your own Bitcoin mining facility depends on your financial capacity, technical expertise, and risk tolerance. Hosting services offer a convenient, lower-risk option, require less capital and implies less operational hassle. The flipside is generally higher ongoing cost and less control. Running your own facility provides greater control and potentially lower long-term costs but requires a significant initial investment and operational expertise. Marathon Digital Holdings (MARA) started out as an asset light business but has acquired mining sites over the past year to limited the potential third-party risk of using hosting providers.
2. Hosting Service Provider
In Bitcoin mining, hosting services are arrangements where individuals or companies can place their mining hardware in a facility operated by a third-party service provider. The mining hardware is owned by the client but physically housed and maintained by the hosting provider in their data centre or facility. Next to the mining machines, the hosting provider takes care of the physical infrastructure, such as cooling, electricity, security, and internet connectivity. Clients typically pay a fee to the hosting provider for these services, which may be structured as a flat rate or based on factors like power consumption. Hosting providers typically offer the option of buy & host, also referred to as Mining as a Service (MaaS), and/or co-locating machines that are already purchased by a client.
Hosting services offer operators a relatively stable and predictable income compared to self-mining. This stability arises from contracts that are often paid in fiat currency or fiat-pegged stablecoins. Additionally, these contracts typically have a minimum duration of 12 months. Establishing a hosting service business is less capital-intensive since the primary investment is in the infrastructure rather than in ASIC hardware. A significant difference between hosting and self-mining is that hosting is a client-based business.
3. Cloud Mining
Cloud mining services enable investors to purchase mining contracts, typically spanning fixed durations (ranging from 30 to 360 days), from companies operating large-scale mining facilities. Investors pay upfront fees for hashing power and, in return, receive a portion of the mining rewards generated by the contracted hashing power. These platforms manage all facets of the mining process, including deploying miners, connecting to mining pools, and maintaining operations, ensuring a continuous and stable cryptocurrency output for users. Users can easily commence mining by paying for hash power and service fees online.
Cloud services generally provide operators with stable, predictable, and recurring income. Since customers do not need to invest in hardware and contracts can be short-term, the barrier to entry is quite low. As a result, the potential client base is broader compared to hosting businesses.
AI/HPC Business Models
Revenue in an AI/HPC (High-Performance Computing) data center typically comes from offering computational services to clients, which can be delivered in several ways.
1. In-house Data Center to Power Internal Services
Big companies like Amazon, Microsoft, and Google run their own AI/HPC operations within their data centers. Each of these companies has developed extensive AI/HPC capabilities to power both internal projects and cloud-based services for external clients. Tesla also runs it own data centers to support the company’s advanced technological infrastructure, including electric vehicle production, energy storage systems, and AI-powered initiatives like self-driving cars. These centers process vast amounts of data, enabling real-time analysis for Tesla’s autonomous driving software and other AI applications. This business model compares to a self-miner owning and operating its own infrastructure. Big difference is that a AI/HPC is 10 – 20 times more capital extensive and not as modular as Bitcoin mining operations. Therefore this in-house data center business model is only financially feasible for bigger companies.
2. Co-Location Services
Some data centers lease out space and power for clients to deploy their own AI or HPC servers. This model appeals to organizations that want physical control over hardware but prefer to outsource infrastructure management, including power, cooling, and connectivity. These services allow organizations to leverage the infrastructure of specialized facilities without the need to build or maintain their own data centers. The business model has a lot of similarities with the hosting service in Bitcoin mining.
3. Cloud Computing
Cloud computing is renting out computing power usually in the form of Pay-Per-Use. This can be in the form of a lease or rental model, customers lease HPC hardware, infrastructure, or software licenses for a fixed period, typically on a monthly or yearly basis. Alternatively customers pay for AI/HPC resources based on their usage (similar to how consumers pay for utilities like electricity). These are the three most common cloud computing services offered in AI/HPC:
Infrastructure-as-a-Service (IaaS): AI/HPC data centers offer raw computing power, storage, and networking resources to customers on a pay-as-you-go basis. Companies like AWS, Microsoft Azure, and Google Cloud provide this model, allowing businesses to scale their AI or HPC operations without building their own data centers. Alongside these major tech companies, Bitcoin mining firms are also adopting this model. Notable examples include the GPU-as-a-Service announced by Hut 8 and the GPU computing services provided by Core Scientific’s partner, CoreWeave.
Platform-as-a-Service (PaaS): In this model, the data center provides an integrated platform that includes AI and HPC tools, frameworks, and managed services for development, testing, and deployment. This makes it easier for companies to build AI models or run simulations without managing the underlying infrastructure. Google Cloud AI and Amazon SageMaker are examples of PaaS offerings.
AI-as-a-Service (AIaaS): Providers offer pre-trained AI models, APIs, or tools to businesses, allowing them to integrate AI functionality into their operations without developing the technology in-house. Examples include Google Cloud’s AI APIs or Microsoft Azure Cognitive Services.
And now we move on to the next content for our Premium Members:
Unpacking the AI/HPC Strategies of Public Miners
Hut 8
Core Scientific
