JAKARTA Nvidia CEO Jensen Huang will reveal the latest details regarding the latest artificial intelligence (AI) chip from the company from Santa Clara, California at its annual software developer conference, Tuesday, March 18.
Nvidia shares have more than quadrupled in the last three years. This is driven by its success in supporting the development of advanced AI systems such as ChatGPT, Claude, and others.
Most of Nvidia's success comes from the last decade when the Santa Clara, California-based company built software tools to attract AI researchers and developers. However, the largest contribution in Nvidia's annual revenue of US$130.5 billion (Rp2.1 quadrillion) last year came from its data center chips, each for tens of thousands of dollars.
Huang has previously hinted that Nvidia's latest flagship chip will be named Rubin. The chip will be part of a family of chips that include GPU (unital graphics processing), CPU (central processing unit), and network chips, all designed to work in a large data center to train the AI system. Analysts expect Rubin chips to start production this year and will be released in large volume starting next year.
However, Nvidia faces various challenges in its efforts to introduce new flagship chips every year. The current chip, called Blackwell, is experiencing a delay in launch due to design defects that cause production problems.
In addition, the AI industry is also facing challenges where conventional methods of processing large data at data centers full of Nvidia chips are starting to show a reduced return.
SEE ALSO:
Nvidia shares also fell this year after Chinese startup DeepSeek claimed to be able to create competitive AI chatbots with less computing power, thereby reducing dependence on Nvidia chips.
Responding to the claim, Huang emphasized that the latest more complex AI model would further increase demand for Nvidia chips. According to him, the next AI generation will think more about it before it produces answers, and the Nvidia chip is the fastest in generating tokens, fundamental units in AI processing.
"When the ChatGPT was first launched, the speed of the token generation only needed to be as fast as humans read," Huang was quoted as saying by VOI from Reuters last month. "But now, AI is thinking faster than humans, as it has to generate many possibilities before giving the best answer. That's why our chips will remain the most important."
The English, Chinese, Japanese, Arabic, and French versions are automatically generated by the AI. So there may still be inaccuracies in translating, please always see Indonesian as our main language. (system supported by DigitalSiber.id)
