According to Michael Kagan, chief technical officer of NVIDIA, cryptocurrencies do not bring anything useful to society. He believes that other ways to use computing power, such as the chatbot with artificial intelligence ChatGPT, would be more useful than cryptocurrency mining. The first version of ChatGPT was trained on a supercomputer with approximately 10,000 NVIDIA graphics cards.
“Cryptographic technologies needed parallel processing, and the technique [NVIDIA] was the best, so people just programmed it for that purpose. We bought a lot of equipment, and then in the end everything fell apart, because it does not bring anything useful to society. AI – yes. Kagan said. “With ChatGPT, now everyone can create their own machine, their own program: you just tell it what to do and it will work. And if it doesn’t work the way you want, you say, “I want something else.”
Cryptocurrency, in contrast, was more like high-intensity trading (HFT)
HFT – high-speed automated stock trading – is an industry that was in the business interests of Mellanox, a company founded by Kagan before it was acquired by NVIDIA.
“I never believed that [криптовалюта] can benefit humanity. You know, people do crazy things, but they buy your products and you sell them to them. But you’re not redirecting the company to support anything.”
Just under a year ago, NVIDIA was forced to pay $5.5 million in fines as part of a US Securities and Exchange Commission settlement. The company did not inform investors about how mining affects GPU sales. In addition, according to the SEC statement, NVIDIA was aware that cryptocurrency mining was part of their business, and even claimed that NVIDIA senior management planned to enter the cryptocurrency mining market.
NVIDIA started out as a company that made powerful graphics cards, data center hardware, and supercomputers. After the end of the cryptocurrency mining boom, its equipment was at the center of the rapid development of generative AI technologies.
Two weeks ago, Microsoft said it had bought tens of thousands of A100 processors from NVIDIA to power the OpenAI workload. NVIDIA also sold 20,000 H100 processors, the chip’s successor, to Amazon for its cloud computing service AWS, and another 16,000 were sold to Oracle. In addition, NVIDIA offers the rental of computing power directly, and quite expensive: almost $37,000 for a cluster of eight H100 connected together in a “cluster”.
Recently, NVIDIA CEO Jensen Huang compared the importance of ChatGPT to the advent of the iPhone and noted that generative AI could “reinvent some industries.”
NVIDIA Introduces Hardware Accelerators for ChatGPT, AI Video Processing, Generative and More
Source: The Guardian
