BNR – Advanced Micro Devices (AMD) is scheduled to announce updated information regarding an AI “Superchip” on Tuesday. Analysts believe the new chip will be a significant competitor to Nvidia Corp, whose chips now monopolise the AI sector.
AMD CEO Lisa Su will deliver the opening talk at a San Francisco event on the company’s data centre and AI plans. Analysts anticipate new information on the MI300, AMD’s most sophisticated graphics processing unit (GPU).
The chip falls under the type of chips used by businesses such as OpenAI for developing products such as ChatGPT.
AMD Superchip Takes Aim at Nvidia’s Market Dominance
According to estimates, Nvidia controls 80% to 95% of the AI market.
Nvidia’s market value temporarily surpassed $1 trillion last month. This occurred after the company stated that it projected a revenue increase as a result of acquiring fresh chip supplies to fulfil growing demand.
There are few large-scale competitors to Nvidia. Nvidia’s competitors include Intel Corp. and other start-ups such as Cerebras Systems and SambaNova Systems.
However, the internal chip efforts at Alphabet Inc’s Google and Amazon.com’s cloud unit have proven to be Nvidia’s most significant sales threat thus far. Both companies license out their customised chips to third-party developers. AMD’s new chip will not change that overnight.
Su informed investors last month that the MI300 would begin selling in the fourth quarter. However, “[it] will be more meaningful in 2024,” Su said.
Furthermore, Nvidia’s edge stems from over a decade of supplying software programmes to AI researchers and learning to predict what they would require in chips that take years to develop.
AMD, according to Kevin Krewell, chief analyst at TIRIAS Research, will have to keep up with both software and research advances.
“They have to connect with the research community,” Krewell said. “To understand what the market is doing and what they need to do, well ahead of when something becomes important.”
