Even if the US chipmaker Nvidia’s potent processors are sold in vast amounts to the industry, the corporation has claimed that cryptocurrencies do not “deliver anything good for society.”
According to the company’s chief technology officer, Michael Kagan, other applications of computing power, such as the artificial intelligence chatbot ChatGPT, are more beneficial than mining cryptocurrency.
Nvidia never extended a warm welcome to the cryptocurrency community. To ensure supplies went to its favoured clients instead, including AI researchers and gamers, the company even produced software in 2021 that artificially restricted the ability to utilize its graphics cards for mining the well-known Ethereum cryptocurrency.
According to Kagan, the decision was reasonable given the minimal utility of employing computing power to mine bitcoins.
Around 10,000 Nvidia graphics cards made up a supercomputer used to train the first iteration of ChatGPT.
“Nvidia is the greatest, so people programmed it to utilize for this purpose. All this crypto stuff needed parallel processing and to be done quickly. They purchased a lot of things, but since they didn’t provide anything beneficial to society, they eventually fell apart. Kagan told the Guardian that AI does.
“Using ChatGPT, anyone can now build their device or program; all they have to do is instruct it what to do. Also, you can tell it, “I want something different,” if it doesn’t function the way you want it to.
Contrarily, high-frequency trading, a sector that brought in business for Mellanox, established before Nvidia acquired it, was more like cryptocurrency.
We were heavily involved in trading as well, he continued. “Wall Street was buying our stuff to save a few nanoseconds on the wire, and the banks pulling the fibres under Hudson taut to make them a little bit shorter, to save nanoseconds between their datacentre and stock exchange.”
“I never thought [crypto] would benefit humanity positively. People are crazy, but they still purchase the goods you sell to them. Therefore, you keep the business the same to support whatever it is.
It was almost by accident that Nvidia’s products found their place at the centre of the AI boom, given that the company was previously best known for making potent graphics cards for PC gamers to play the newest games.
The basic yet powerful processors that gamers had adopted happened to operate substantially faster on the computationally tricky work of training a new AI system, which may require millions or billions of dollars worth of computer power.
Microsoft announced two weeks ago that it had purchased thousands of Nvidia’s A100 GPU AI-focused processors to support OpenAI’s workload. The H100, Nvidia’s replacement for that chip, has been purchased by Amazon for use in its AWS cloud computing service and 16,000 units by Oracle.
With its DGX cloud service, which starts at a little under $37,000 (£30,250) a month for just eight H100s connected in a “cluster,” Nvidia also rents access to the chips directly.
Nvidia’s CEO, Jensen Huang, said the company is the driving force behind “the iPhone moment of AI” and that the “generative AI” his company powers will “reinvent practically every industry” in remarks made last week at the company’s annual conference.
Due to regulatory issues, Nvidia’s $40 billion acquisition of the UK-based technology company Arm last year failed.