In the ever-growing field of artificial intelligence (AI), major tech players like Intel, Nvidia, and AMD are constantly vying for dominance. On Thursday, Intel unveiled its latest computer chips, including the Gaudi3, an AI chip targeted towards generative AI software. This article will delve into the details of Intel’s new releases and their competitive position against rival chips, specifically Nvidia’s offerings.
Gaudi3, set to launch next year, poses a direct challenge to Nvidia and AMD’s AI chips that currently power large and power-hungry AI models. Nvidia’s GPUs have long been the preferred choice for AI models like OpenAI’s ChatGPT, which runs in the cloud. This dominance is evident in Nvidia’s impressive stock growth of nearly 230% year-to-date, in stark contrast to Intel’s 68% increase. Sensing an opportunity to capture AI companies’ attention away from Nvidia, both AMD and Intel have announced rival chips in a bid to disrupt Nvidia’s stronghold in the market.
Intel’s Gaudi3 specifically competes with Nvidia’s H100, the preferred chip for companies building large AI chip farms. Additionally, AMD’s MI300X, scheduled for customer shipments in 2024, is another rival in the AI chip market. Intel’s journey into developing Gaudi chips began in 2019, when it acquired a chip developer named Habana Labs. Intel CEO Pat Gelsinger highlighted the growing excitement around generative AI and expressed his belief in AI PCs becoming the standout performer in the upcoming year.
Alongside Gaudi3, Intel also unveiled its new Core Ultra processors, designed specifically for Windows laptops and PCs. These processors come equipped with a specialized AI component called an NPU, capable of running AI programs at a faster pace. This strategic move by Intel aligns with the industry trend, as it indicates traditional processor manufacturers like Intel, AMD, and Qualcomm are adapting their product lines to cater to the rising demand for AI models.
While Core Ultra chips may not possess the power to run heavy-duty AI applications like ChatGPT without an internet connection, they excel at handling smaller tasks. For example, Intel highlighted that its chips can efficiently handle Zoom’s background-blurring feature. The Core Ultra chips are built using Intel’s 7-nanometer process, ensuring greater power efficiency compared to their predecessors. This aligns with Intel CEO Pat Gelsinger’s strategy to catch up to Taiwan Semiconductor Manufacturing Company’s (TSMC) chip manufacturing prowess by 2026.
Intel’s Core Ultra chips not only boast AI capabilities but also offer enhanced gaming features. The added graphics muscle can significantly improve the performance of programs like Adobe Premier, reducing render times by over 40%. This advancement in gaming capabilities caters to the ever-growing demand for high-performance gaming systems.
Intel also introduced its fifth generation Xeon processors, primarily targeted at large organizations and cloud companies that heavily rely on server power. While Intel did not disclose pricing details, previous Xeon processors have commanded prices in the thousands of dollars range. These processors are commonly paired with Nvidia GPUs in systems utilized for training and deploying generative AI models. With some systems incorporating up to eight GPUs alongside one or two Xeon CPUs, the new Xeon processor is particularly well-suited for inferencing, a less power-hungry process than training AI models.
Intel’s latest computer chip offerings, including Gaudi3 and Core Ultra processors, demonstrate the company’s determination to compete with industry giants like Nvidia and AMD in the AI chip market. With Nvidia’s stronghold in AI, Intel aims to attract AI companies and shift the balance of power in its favor. The future of AI supremacy is still uncertain, but as Intel continues to innovate and develop cutting-edge chips, the race for dominance in the AI market is set to intensify.