CNN Central News & Network-ITDC India Epress/ITDC News Bhopal: Amid rising demand for artificial intelligence, OpenAI has taken a major step towards developing its own custom AI chip. The company, in partnership with Broadcom, has developed its first AI accelerator chip named ‘Jalapeño’. It has been designed specifically with the inference requirements of large language models, or LLMs, in mind.
The biggest reason behind OpenAI’s move is to gain greater control over the cost, capacity and performance of AI computing. The company says designing a chip according to the needs of its models and systems can help run AI faster, more reliably and more affordably. In early tests, Jalapeño’s performance-per-watt is claimed to be better than current state-of-the-art solutions.
The strategy could also reduce OpenAI’s dependence on external AI hardware. Working with Broadcom, the company took the chip from design to manufacturing tape-out in around nine months. OpenAI’s own AI model was also used in the chip’s development and optimization.
The company plans to develop several generations of such custom AI accelerators in the coming years. According to reports, OpenAI is also increasing collaboration with Samsung in the next phase for next-generation chip development and production.
The move makes clear that OpenAI does not want to remain limited to AI models and software, but is moving towards gaining greater control over the entire AI system, from models to chips and data centre infrastructure.
Hashtags: #ScienceTechnology #Bhopal #DeskSource #OpenAI #AIChip #Broadcom #Jalapeño #LLM #Samsung #OpenAICustomAIChip
