Intel Unveils Vision for the Open AI Era and Strategies to Challenge Nvidia's Dominance

Intel recently unveiled its vision for enterprise AI during the Vision 2024 event in Phoenix, Arizona, highlighting an open AI ecosystem and introducing its new Xeon 6 data center processors.

In his keynote address, CEO Pat Gelsinger emphasized Intel's commitment to fostering a secure and open enterprise AI landscape. Based in Santa Clara, California, Intel showcased its expanding roster of customers and partners across the AI software and hardware domains.

Together with industry partners, Intel aims to establish an open platform for enterprise AI that enhances the deployment of secure generative AI (GenAI) systems, powered by retrieval-augmented generation (RAG).

Intel's scalable systems strategy is designed to cater to every AI segment within enterprises, marking a departure from its 1990s dominance in the PC market. Today, with Nvidia leading in AI chip technologies, Intel is positioning itself as a competitive open alternative.

Gelsinger stated that Intel aspires to become the second-largest foundry for chip manufacturing by the end of the decade. This ambition is supported by significant investments, including an $8.5 billion grant from the U.S. Commerce Department, aimed at fueling the AI revolution, which Gelsinger deems the most significant technological advancement since the internet.

“Intel is on a mission to bring AI everywhere,” Gelsinger asserted. The company has already shipped five million AI chips for PCs, projecting the total to reach 40 million by year-end.

Intel introduced its forthcoming processors for data centers, cloud, and edge applications under the new Xeon 6 brand. The Xeon 6 processors will feature Efficient-cores (E-cores), boasting a 2.4 times performance-per-watt improvement compared to earlier models, while the Performance-cores (P-cores) will enhance AI performance significantly.

Katti explained that Intel is prioritizing openness across all system layers, stating, “Our CPUs and accelerators will integrate seamlessly using open Ethernet-based networking standards.” This robust approach extends to Intel's software offerings, with the oneAPI and oneDNN supporting open-source frameworks to eliminate hardware restrictions.

Among Intel's flagship projects is the Gaudi 3 AI accelerator, designed to outperform Nvidia's H100 AI computer with 50% better inference and a 40% improvement in power efficiency—all at a lower price point.

Intel's Gaudi 3 accelerator customers include notable companies like Bosch, CtrlS, and IFF, aimed at expediting GenAI integration within enterprises. Gelsinger noted that only 10% of enterprises managed successful GenAI deployments last year, underscoring the need for efficient solutions.

“Every company is swiftly becoming an AI company," Gelsinger remarked. “Our Gaudi, Xeon, and Core Ultra platforms provide flexible solutions to address evolving customer needs.”

Intel's roadmap includes plans for an open and scalable AI ecosystem. The company is committed to fostering collaboration among hardware manufacturers, software providers, and system integrators to support enterprise-specific GenAI requirements.

Additionally, Intel announced partnerships with Google Cloud, Thales, and Cohesity to enhance confidential computing services. These collaborations will enable customers to run their AI models in trusted environments, leveraging Intel’s security features like Trust Domain Extensions (TDX) and Software Guard Extensions (SGX).

Overall, Intel's strategic thrust toward an open enterprise AI ecosystem, anchored by its advanced hardware and collaborative approach, aims to transform the way businesses harness the power of AI.