Intel, the leading global chip manufacturer, introduced a suite of innovative products at its recent “AI Everywhere” event, with a standout being Gaudi3, tailored for generative AI applications. Set to hit the market next year, Gaudi3 aims to rival offerings from Nvidia and AMD, targeting robust and energy-intensive AI models.
Generative AI, a facet of artificial intelligence that crafts novel content like images, text, music, and videos from existing data, has witnessed significant traction. Nvidia’s GPUs, as employed by popular models such as OpenAI’s ChatGPT, have fueled the company’s stock surge by nearly 230% this year. In response, Intel and AMD have unveiled their chips for generative AI to challenge Nvidia’s dominance.
While specifics about Gaudi3 remain undisclosed, Intel indicated it would compete with Nvidia’s H100 and AMD’s upcoming MI300X, slated for a 2024 release. Intel initiated the development of Gaudi chips in 2019 following the acquisition of chip startup Habana Labs.
Intel CEO Pat Gelsinger, at a launch event in New York, highlighted the prominence of generative AI in 2023. He also emphasized Intel’s vision for AI PCs, where the newly revealed Core Ultra processors, introduced alongside Gaudi3, are expected to play a pivotal role.
The Core Ultra processors, tailored for Windows laptops and PCs, incorporate a dedicated AI component known as an NPU, enabling faster execution of AI programs. This aligns with the broader trend among traditional chipmakers, including AMD and Qualcomm, adapting products to leverage the rising demand for AI-driven applications.
The 7-nanometer Core Ultra chips boast enhanced energy efficiency compared to their predecessors, aligning with Gelsinger’s plan to catch up with Taiwan Semiconductor Manufacturing Co. in chip manufacturing technology by 2026. Offering superior gaming performance and graphics capabilities, these chips can accelerate applications like Adobe Premier by over 40%.
Simultaneously, Intel unveiled fifth-generation Xeon server chips, integral to the infrastructure of cloud companies and large organizations. While the pricing remains undisclosed, these processors, commonly paired with Nvidia’s GPUs, excel in training and deploying generative AI models.
Gelsinger clarified in a recent interview that Intel has no plans to spin off its contract chip manufacturing business. The Intel Foundry Services (IFS) operates as an internal business unit within Intel, with Gelsinger emphasizing its advantages in the current environment. Despite future plans to break out IFS’s financials, Gelsinger believes the integrated model, with chip design and manufacturing working in tandem, provides confidence to IFS customers as a “clean supplier” of manufacturing capacity.
