NVIDIA Finds a New Gold Mine in AI
In a recent interview with The Guardian, Michael Kagan, the CTO of well-known GPU manufacturer NVIDIA, made a bold statement about cryptocurrencies. According to him, despite the fact that the industry has brought the company huge profits in 2021, cryptocurrencies do not provide any value to society.
It seems like Michael Kagan is making these statements to grab the attention of other clients. With Ethereum moving to a different consensus algorithm, graphics cards are no longer needed to support its network.
As a result, he now believes that computational power is better suited for AI programs like ChatGPT. Interestingly, this bot was trained on a supercomputer that contained around 10,000 NVIDIA graphics cards, which were not used by gamers or professionals in need of powerful computers during the profitable days of ETH mining.
It's worth noting that in the same year, 2021, NVIDIA began limiting the power of its graphics cards for use in mining. However, it's unclear whether this was a noble gesture that reduced sales to those in the cryptocurrency industry. Miners still actively bought the cards while it was profitable, and when profits decreased significantly, new software appeared mysteriously that removed the limitations.
It's worth noting that NVIDIA also produced special versions of its graphics cards designed solely for mining and unsuitable for use in video games. These cards were essentially the same as regular cards but without video outputs and sold at the same price.
There are even more contradictions to Michael Kagan's words. Recently, we wrote about a CoinWire study that showed NVIDIA is one of the companies betting on the Web3 industry. And as we know, Web3 and cryptocurrencies are very closely related concepts.
It seems that the chipmaker has simply found a more interesting niche for itself. Recently, Microsoft announced the acquisition of tens of thousands of Nvidia A100 graphics processors to support OpenAI's workload. Additionally, 20,000 H100 chips were sold to Amazon's cloud computing service and 16,000 units to Oracle.