Photo credit: www.techradar.com

Intel Emphasizes Jaguar Shores to Strengthen Its AI Capabilities

Gaudi Chips Position Intel for AI Inference Market
Intel’s 18A Node Aims to Boost Manufacturing Supremacy

In an ambitious effort to solidify its place in the competitive landscape of artificial intelligence (AI), Intel has recently introduced the Jaguar Shores AI accelerator. This addition, revealed at the SC2024 supercomputing conference, marks a pivotal step in the company’s ongoing development of AI chips aimed at rivaling industry leaders such as Nvidia and AMD.

Although specifics regarding Jaguar Shores remain limited, it is anticipated to succeed the Falcon Shores architecture, which is set to launch in 2025. The new accelerator will likely focus on AI inference, an area Intel is targeting to recapture market share against its formidable competitors. It is currently unclear whether Jaguar Shores will function as a graphics processing unit (GPU) or an application-specific integrated circuit (ASIC); however, Intel’s existing roadmap hints at a next-generation GPU meant for enterprise-level AI applications.

Navigating Market Challenges

Historically, Intel has struggled to keep pace with its rivals in the AI training space, but the company is now shifting its attention toward AI inference. This shift leverages its existing Gaudi chips along with the forthcoming Falcon Shores and Jaguar Shores technologies.

Nonetheless, to improve its competitive standing, Intel confronts numerous technical and organizational hurdles. The company has encountered various difficulties within the AI hardware sector, particularly concerning its GPU offerings. A prominent example is the abandoned Rialto Bridge project, which failed to attract sufficient customer interest, leaving numerous institutions, including the Barcelona Supercomputer Centre, in a lurch.

Intel’s strategy for the Falcon Shores GPU has undergone multiple revisions; what was once envisioned as an integrated CPU-GPU solution is now being redefined as an independent GPU product.

The AI training market has largely been dominated by Nvidia and AMD, with Nvidia leading the charge through its robust GPUs and the CUDA software ecosystem. Acknowledging Intel’s current position, CEO Pat Gelsinger admitted the company has fallen behind competitors like Nvidia, AWS, Google Cloud, and AMD in the race for AI supremacy.

Intel is counting on its upcoming 18A manufacturing process to provide a competitive advantage over companies such as TSMC. The 18A technology boasts innovations like RibbonFET transistors and backside power delivery systems, which are expected to enhance chip efficiency and overall performance.

For further details, refer to HPCwire.

Related Articles

Source
www.techradar.com