From Crypto To AI: Repurposing Data Centers For The Future

The power demands of AI computing are already huge and skyrocketing by the week. In 2022, AI data centers accounted for over 4% of U.S. electricity consumption, and by 2030 it’s estimated that AI may consume 3% to 4% of total data center demand worldwide, on par with entire countries like Japan or Russia.
As more and more AI foundational models are being trained and coming online, each one seemingly requiring more computing power than the last, it’s clear that something needs to change to keep up with this surge. That’s why some companies are looking to the world of cryptocurrency, where meeting massive power needs is already a reality.

Everyone Needs Energy

Cryptocurrency mining is incredibly power-intensive because of its proof-of-work (PoW) consensus mechanism, in which miners compete to solve complex mathematical puzzles. Bitcoin alone already consumes more electricity than Norway.
Crypto companies are exploring ways to use their expertise in running large, efficient data centers to support AI research and applications, and crypto facilities are already being repurposed fully or partially for this new use. The potential revenue rewards are high. According to some estimates, AI compute can be up to 25 times more profit-dense than crypo mining.
MORE FOR YOU
Today’s NYT Mini Crossword Clues And Answers For Friday, September 20th
‘Don’t Move’ Directors On Work With Sam Raimi As Trailer For Netflix Movie Debuts
‘Arcane’ Season Two Release Date Revealed In New Trailer
AI systems also carry less upfront capital risk, since GPU orders can be timed closer to shipping or delivery. Crypto, on the other hand, requires up-front deposits long before delivery, which makes top-line revenue more volatile. At the same time, crypto mining earns revenue from the moment the machines are turned on, with no need for a sales force or even clients, while AI requires securing customers to utilize the compute capacity.
In terms of hardware, the bones of both types of systems are similar, but there are a few key differences. AI data centers have higher requirements for uptime and therefore need more redundancy for power and networking. They also need better temperature control systems than more “rugged” crypto-mining facilities. Think of it this way: If a crypto data center is like a pickup truck, an AI system needs a facility that is more like a Ferrari.

The Future Of AI Data Centers

It’s clear we have to figure out how to handle the insatiable demand for computer power needed to train increasingly sophisticated AI models—not just chatbots, but multimodal AI that can handle text, images and video.
Of course, repurposing crypto infrastructure isn’t the only way to approach the soaring energy demands of AI. Advances in neural networks and other technology developments are helping make AI data centers more energy efficient without affecting performance. Other green computing initiatives include liquid cooling systems, renewable energy sources and waste heat recovery.
The trend toward edge AI, where processing occurs closer to the data source (such as in IoT devices) reduces the need for centralized data centers and can improve efficiency. Like blockchain, decentralized AI networks can distribute computation across a network of smaller nodes, reducing the strain on large data centers.
In the meantime, however, the intersection of AI computing and crypto-mining infrastructure can be one of many solutions, and the overlap of technologies has the potential to drive innovations in both fields.

Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?

Original Source: https://www.forbes.com/councils/forbestechcouncil/2024/09/19/from-crypto-to-ai-repurposing-data-centers-for-the-future/