Intel vision 2024 Announced Enterprise AI with Gaudi 3 for Partners

During the Intel Vision 2024 customer and partner conference, Intel launched the Intel Gaudi 3 accelerator aiming towards improved performance, openness, and options for enterprise generative AI (GenAI). Additionally, Intel revealed a range of new open scalable systems, next-gen products, and strategic partnerships, all geared towards expediting the adoption of GenAI.

Given that only a small fraction (10%) of enterprises effectively transitioned GenAI projects to production last year, Intel's recent offerings are supposedly designed to tackle the hurdles businesses encounter when scaling AI initiatives.

“Innovation is advancing at an unprecedented pace, all enabled by silicon – and every company is quickly becoming an AI company,” said Intel CEO Pat Gelsinger. “Intel is bringing AI everywhere across the enterprise, from the PC to the data center to the edge. Our latest Gaudi, Xeon, and Core Ultra platforms are delivering a cohesive set of flexible solutions tailored to meet the changing needs of our customers and partners and capitalize on the immense opportunities ahead.”

Enterprises aim to transition GenAI projects from pilot phases to full-scale production. Achieving this goal necessitates access to readily accessible solutions, utilizing energy-saving processors such as the Intel Gaudi 3 AI accelerator. Furthermore, these solutions would tackle challenges related to complexity, fragmentation, data security, and compliance requirements.

Introducing Gaudi 3 for AI Training and Inference

The Intel Gaudi 3 AI accelerator has designed AI systems to enable connections of up to tens of thousands of accelerators via the ubiquitous Ethernet standard. Offering an advancement, Intel Gaudi 3 has AI compute capacity for BF16 and an increase in memory bandwidth compared to its predecessor. 

This accelerator reportedly has AI training and inference capabilities, catering to global enterprises seeking to implement GenAI on a large scale.

Intel Gaudi 3 is anticipated to achieve a 50% faster time-to-train on average across Llama2 models with 7B and 13B parameters, as well as the GPT-3 175B parameter model. 

Moreover, the inference throughput of the Intel Gaudi 3 accelerator is reportedly expected to surpass predecessors by an average of 50%, with a 40% improvement in inference power efficiency across Llama 7B and 70B parameter models, as well as the Falcon 180B parameter model.

Intel Gaudi 3 offers open, community-driven software and utilizes industry-standard Ethernet networking. Enterprises scale seamlessly from individual nodes to clusters, super-clusters, and mega-clusters comprising thousands of nodes. Such scalability supports inference, fine-tuning, and training operations at an extensive scale.

The Intel Gaudi 3 will be accessible to OEMs, including Dell Technologies, Hewlett Packard Enterprise, Lenovo, and Supermicro, starting in the second quarter of 2024.

Bharti Airtel: Utilizing Intel's advanced technology, Airtel aims to harness its extensive telecom data to strengthen its AI capabilities and enhance customer experiences. These initiatives align with Airtel's plans for technological innovation and facilitating the development of new revenue channels in the swiftly evolving digital realm.

Infosys: A strategic partnership was unveiled to integrate Intel technologies, encompassing 4th and 5th Gen Intel Xeon processors, Intel Gaudi 2 AI accelerators, and Intel Core Ultra, into Infosys Topaz. This AI-centric suite comprises services, solutions, and platforms that leverage generative AI technologies to accelerate business value.

Ola/Krutrim: To pre-train and fine-tune its initial India foundational model, which boasts generative capabilities across 10 languages, Krutrim aims to achieve better price/performance compared to existing market solutions. Currently, Krutrim is pre-training a larger foundational model using an Intel Gaudi 2 cluster.

Intel also revealed partnerships with Google Cloud, Thales, and Cohesity to utilize Intel's confidential computing capabilities within their cloud instances. This encompasses features such as Intel Trust Domain Extensions (Intel TDX), Intel Software Guard Extensions (Intel SGX), and Intel’s attestation service.