Skip to main content

The Necessity of Confronting Compute Costs Head-On

The BBC article “Every Bitcoin payment ‘uses a swimming pool of water'” left many puzzled. The vision of paying for everyday items with Bitcoin, embraced by some stores and countries with Bitcoin ATMs, takes an unexpected turn. Picture this: you’re buying coffee, and as the barista brews, you find yourself watching a series of pools being drained in the  desert. Whether you’re an environmentalist or not, the apparent waste overshadows the benefit. If credit cards operated similarly, most would likely be abandoned.


Yet, the issue goes beyond this. In a world where profits matter, the old adage that there’s no such thing as a free lunch holds true. Costs invariably find a way to be covered. Sooner or later, the transaction cost of paying in bitcoins falls on the consumer. This is the moment when the rubber hits the road.


Strangely enough, we often overlook such costs until they become unavoidable. The issue of technical debt in constructing advanced AI models is frequently a neglected topic in many organizations. Transaction costs, whether related to cutting-edge technologies like Bitcoin and AI or the most routine tasks, tend to be an afterthought. Pose a question to a financial institution about the unit costs of bank statements or database queries, and observe the struggle to explain why such considerations haven’t been given much thought.


I deliberately selected database queries as an example because, in today’s analytics, reporting, and web interactivity landscape, the global volume of queries might surpass the number of consumer transactions. While this might seem counterintuitive, consider the numerous queries or searches made before making a purchase. We can reasonably assume a ratio of 50 to 1. Fortunately, advertisers cover these costs. If consumers had to bear the expense, the preference for in-store shopping and strolling through aisles might outweigh online purchases.


Consider all the operational queries for reports that numerous stakeholders require. Take, for instance, an organization with 10,000 merchandisers generating daily reports. If you pay for query consumption on a pay-as-you-go basis, and each merchandizer runs 3 reports with 10 data manipulations in each, you’re looking at 30 queries per person. That sums up to 300,000 queries for the day with just 3 reports. This not only strains compute power consumption but also introduces unpredictability. How do you guarantee it will be 300,000 and not 500,000 queries? The cost disparity is substantial. While statistical averaging over time might be argued, try explaining that to accountants who favor the strict logic of double-entry accounting over statistical averaging.


This is where we advocate for considering decentralized compute to alleviate the burden of compute costs. Although decentralization and distributed computing have a longstanding history in software, cloud technology has somewhat overshadowed them. Bringing decentralization back into focus, especially when combined with cloud computing, enhances the advantages of the cloud and amplifies cost savings. Imagine pushing the 30 queries per person to each individual’s browser using distributed in-document compute. The total queries consuming compute power plummet to 30,000, covering the queries required to generate the 3 reports for each merchandizer. The remaining queries— the difference between 30,000 and 300,000— are executed in local browsers. That translates to a whopping 90% reduction in query costs. While it might not make paying with bitcoins less expensive, it represents a substantial step forward in reducing compute costs, with significant impacts on various applications running in the cloud or on-premise.


To Learn more about how to create interactive bank statements, contact us.

Dr. Rado

read it on Linkedin

Leave a Reply