Intel: Dark Horse In AI Processing

Summary:

• AI is a large enough application space, there can and should be more than one leader.
• Intel has technical experience and had made strategic acquisitions to support an AI business.
• Their Habana Labs’ Gaudi AI processors, in particular, appear to be competitive to Nvidia GPU.
• If they can maintain their AI focus, Intel can emerge as a significant player in the AI boom.

Nvidia (NVDA) has had an amazing run, but any emerging technology, such as AI, which is bottlenecked by a single company will have issues in growth. Consulting firm McKinsey has pegged the AI market to be worth $1 trillion by 2030, but also that it was in an experimental and in early phases of commercial deployment.

While Nvidia will likely retain its leadership in GPU hardware as applied to AI for the foreseeable future, it is likely that other hardware solutions for AI systems will also be successful as AI matures. While technologist may quibble on specifics, all major AI hardware today are based on GPU architectures, and as such I will use the terms and concepts of AI hardware and GPU architecture somewhat interchangeably.

One likely candidate for AI related growth may be AMD (AMD), which has had GPU products since acquiring ATI in 2006.However, unlike Nvidia, which had a clear vision for of general-purpose GPU products (GPGPU), historically, AMD had largely kept its focus on the traditional gaming applications. AMD has developed an AI architecture called XDNA, and an AI accelerator called Alveo and announced its MI300, an integrated chip with GPU acceleration for high-performance computing and machine learning. How AMD can and may evolve in the AI may be subject of a different article.

Another contender for success in the AI applications using GPU is Intel (NASDAQ:INTC), who is the focus of this article. Intel has maintained a consistent, if low key focus on GPU hardware focused on AI applications over the last decade. Intel’s integrated HD Graphics is built into most modern processor ICs; however, these are insufficient compared to dedicated GPUs for high-end inferencing or machine learning tasks.

It has 2 primary GPU architectures in production release:

In 2019 Intel Corporation acquired Habana Labs, an Israel-based developer of programmable deep learning accelerators for the data center for approximately $2 billion. Habana Labs’ Gaudi AI product line from its inception focused on AI deep learning processor technologies, rather than as GPU that has been extended to AI applications. As a result, Gaudi microarchitecture was designed from the start for the acceleration of training and inferencing. In 2022 Intel announced Gaudi2 and Greco processors for AI deep learning applications, implemented in 7-nanometer (TSMC) technology and manufactured on Habana’s high-efficiency architecture. Habana Labs benchmarked Gaudi2’s training throughput performance for the ResNet-50 computer vision model and the BERT natural language processing model delivering twice the training throughput over the Nvidia high end A100-80GB GPU. So, Gaudi appears to give Intel a competitive chip for AI applications.

Concurrent with the Habana Labs’ Gaudi development, Intel has internally developed the Xe GPU family, as dedicated graphics card to address high-end inferencing or machine learning tasks as well as more traditional high-end gaming. Iris® Xe GPU family consists of a series of microarchitectures, ranging from integrated/low power (Xe-LP) to enthusiast/high performance gaming (Xe-HPG), data center/AI (Xe-HP) and high-performance computing (Xe-HPC). The architecture has been commercialized in Intel® Data Center GPU Flex Series (formerly codenamed Arctic Sound) and Intel® Arc GPU cards. There is some question on Xe GPU future and evolution. Intel has shown less commitment to the traditional GPU space compared to Gaudi. Nonetheless, it does demonstrate Intel ability to design and field complex GPU products as its business requires.

Intel has many other AI projects underway. The Sapphire Rapids chips implements AI specific acceleration blocks including technology called AMX (Advanced Matrix Extensions), which provides acceleration inside the CPU for efficient matrix multiplications used in on-chip inferencing and machine learning processing by speeding up data movement and compression. Intel has supporting technologies such as Optane, which while cancelled as a production line, is available for their needs of a high-performance non-volatile memory, one of the intrinsic components in any AI product.

Based on the above, Intel appears to have competitive hardware solutions, however if we look at Nvidia success in AI, it is a result of a much a software and systems focus as it is the GPGPU hardware itself. Can Intel compete on that front. Ignoring for the moment that Intel has a huge software engineer (approx. 15,000) resource, it also has- access to one of the leading success stories in perhaps the most competitive AI application – self driving cars.

Mobileye, who was acquired by Intel in 2017, has been an early adopter and leader, with over 20 years of experience in automotive automated driving and vision systems. As such, Mobileye has a deep resource of AI domain information that should be relevant to many applications. Mobileye has announced that it is working closely with Habana, as related divisions within Intel. While Intel is in the process of re-spinning out Mobileye as public company, Mobileye Global Inc. (MBLY), at present Intel still owns over 95% of shares, keeping it effectively an Intel division.

In looking at Intel, we have a company with the history, resources, and technology to compete with Nvidia and infrastructure. They have made significant investment and commitment to the emerging AI market, in times when they have exited other profitable businesses. It should also be understood that AI related product are a small percentage of overall Intel revenues (INTC revenue are more than twice NVDA, even if NVDA has 6x its market cap), and continues to keep its primary business focus on its processor and foundry business.

Hopefully for shareholders, Intel continues to push their AI technology and business efforts. Their current position is that this is strategic, but Intel is in a very fluid time and priorities may change based on business, finances, and of course the general interest and enthusiasm for AI. It is always worth noting that AI as a technical concept is mature, and appears to be cyclical, with interest in the technical community rising and falling in hype and interest once every decade or so. I remember working on AI applications, at the time labeled as expert systems in the 1980s. If we are currently at a high hype point, this may be temporary, based on near term success and disappointment in what AI does achieve. Of course, as always, “this time is different” and the building blocks of effective AI systems currently exist, where for previous iterations, it was more speculative.

As always, caveat emptor. I make no recommendations. I speak only for myself in sharing analysis done for my own personal use. All the above is based on my interpretations of information that I reviewed from public sources. I may have gotten it wrong, errors may have crept in. In talking about the future, you are almost always likely to get the best parts wrong. That’s what makes it interesting.

Editor’s Note: This article was submitted as part of Seeking Alpha’s Best AI Ideas investment competition, which runs through August 15. With cash prizes, this competition — open to all contributors — is one you don’t want to miss. If you are interested in becoming a contributor and taking part in the competition, click here to find out more and submit your article today!

Analyst’s Disclosure: I/we have a beneficial long position in the shares of INTC, AMD, NVDA either through stock ownership, options, or other derivatives. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.

Seeking Alpha’s Disclosure: Past performance is no guarantee of future results. No recommendation or advice is being given as to whether any investment is suitable for a particular investor. Any views or opinions expressed above may not reflect those of Seeking Alpha as a whole. Seeking Alpha is not a licensed securities dealer, broker or US investment adviser or investment bank. Our analysts are third party authors that include both professional investors and individual investors who may not be licensed or certified by any institute or regulatory body.