One reason the crypto market—and others—finds itself in this concerning situation is the rise of AI. Promises of incredible future returns and the potential to replace human labor—thereby cutting personnel costs—have driven companies (and, consequently, investors) to pour significant energy and capital into this sector in recent years.
Billions have been invested in information-processing hardware, data center infrastructure, and power generation to get this technology up and running. The problem, however, is that many of these AI companies are currently operating in the red; they offer free plans that consume tokens—representing computational power—without generating revenue.
Amidst this market euphoria, it is crucial to recall a fundamental concept from accounting and economics: depreciation. Every facility where processing takes place—including the graphics cards, cables, and the entire infrastructure enabling the system to function—has a finite useful life.
Some companies, such as Google, have begun to see the first negative indicators resulting from this massive bet on AI. For the first time, Google has burned through cash due to multibillion-dollar investments in infrastructure—including the chips and the systems required to handle AI processing tasks.
I'll be much appreaciated if you Votes, Reblogs or give any Tips. Thanks and have good reading! 🍀
The Lifespan of Technology in the AI Market
With each passing year, the lifespan of this equipment shrinks; if it fails to generate enough revenue to offset financial losses incurred during the user-acquisition phase, the market will have to recalculate. Even though we are currently in a long period of growth, eventually, someone will sit down with pen and paper to crunch the numbers and weigh the risks. The moment risks outweigh potential returns—and we are talking about potential returns, not actual ones—the bubble could burst.
To illustrate the concept of depreciation: in some countries, it is calculated as an annual percentage of the value invested in fixed assets, typically ranging from 4% to 20%. Notably, in the AI sector, a major expense involves the graphics cards used for processing; these are the fixed assets that depreciate the fastest.
Furthermore, there is a hidden risk that seems to be overlooked: innovation. This could take the form of more modern graphics cards or entirely new types of equipment. Google, for instance, has developed a specialized AI processor called a TPU (Tensor Processing Unit) to replace the GPU, whereas many existing data centers were built using thousands of standard graphics cards. Just imagine the cost of retrofitting one of those facilities.
An interesting perspective on this issue comes from history itself. It is often said that during the Gold Rush, while everyone else was scrambling to acquire machinery for a chance to strike gold—risking their capital on renting or buying that equipment—the manufacturer of the machinery was the one consistently profiting; after all, they simply sold the gear without bearing the operational risk. This is the case with Nvidia selling graphics cards to companies seeking computing power. However, the gold rush will eventually end; without it, will the "gold-mining machines" continue to sell at the same rate? That is a good question...
