In a bid to challenge Nvidia's dominance in the artificial intelligence (AI) chip market, Intel has announced its latest offering, the Gaudi 3 chip. With Nvidia currently holding an estimated 80% market share, Intel aims to capture a significant portion of the growing demand for AI semiconductors.

The Gaudi 3 chip is designed to train and deploy large AI models, such as OpenAI's ChatGPT, and boasts impressive performance capabilities. Intel claims that the Gaudi 3 is over twice as power-efficient as Nvidia's H100 GPU and can run AI models 1.5 times faster. The chip comes in various configurations, including a bundle of eight Gaudi 3 chips on one motherboard or a card that can be integrated into existing systems.

During testing, Intel utilized the Gaudi 3 chip on models like Meta's open-source Llama and the Falcon, backed by Abu Dhabi. The company stated that the Gaudi 3 chip is capable of training or deploying models, including Stable Diffusion and OpenAI's Whisper model for speech recognition.

One of the key advantages highlighted by Intel is the lower power consumption of their chips compared to Nvidia's offerings. However, Nvidia's graphics processors (GPUs) have been the preferred choice for AI builders in the past year, driving the company's market dominance.

Intel plans to release the Gaudi 3 chips to customers in the third quarter. Major companies such as Dell, HP, and Supermicro will partner with Intel to build systems incorporating the new chips. Although Intel did not disclose the pricing range for the Gaudi 3, it expects it to be highly competitive with Nvidia's latest chips.

The data center AI market is expected to experience significant growth as cloud providers and businesses invest in infrastructure for AI applications. This suggests that there is room for competition, even if Nvidia continues to dominate the market. With the high costs associated with running generative AI and buying Nvidia GPUs, companies are actively seeking alternatives to reduce expenses.

Alongside Intel and Nvidia, other players in the AI chip market are also looking to expand their offerings. AMD, for instance, introduced the MI300X data center GPU last year and counts Meta and Microsoft among its customers. Nvidia, on the other hand, recently revealed the B100 and B200 GPUs, which promise improved performance and are expected to ship later this year.

Intel is taking a different approach to compete with Nvidia by partnering with chip and software giants like Google, Qualcomm, and Arm. Together, they aim to develop open software solutions that are not proprietary, providing software companies with the flexibility to switch between chip providers.

Notably, the Gaudi 3 chip is built on a cutting-edge five nanometer manufacturing process, indicating that Intel may be working with an external foundry for its production. In addition to designing the Gaudi 3, Intel plans to manufacture AI chips, potentially for external companies, at a new factory in Ohio set to open in 2027 or 2028, as confirmed by CEO Patrick Gelsinger.