Intel: Here Come The Custom Chips

Summary

• Whether Amazon’s Graviton2 will be the game changer for the CPU market remains to be seen, Intel has to be sensitive to the emerging trend of customers’ designing proprietary chips.
• Intel’s large cloud customer Amazon’s Graviton2 stands to take away $120 million immediate revenue from Intel. Google’s TPUs may take $650 million, and Microsoft’s FPGAs may take $700 million.
• The wide uses of custom chips are restricted by their customized nature. Intel has about 2% revenue exposure on the custom chips from the 3 cloud providers.
• Merchants developing their own proprietary silicon, i.e., Apple, should be a more alarming trend. Currently, Intel has about 4%, or $2.5 billion annual revenue exposure potentially to Apple’s SoC/SiP chips.

Large cloud providers such as Amazon (AMZN), Google (GOOG), and Microsoft (MSFT) have been rapidly developing their proprietary processors and AI accelerator chips. For one, Amazon recently fired the first shot at Intel’s (NASDAQ:INTC) x86-based instance by designing a second generation of data center processor chip, Graviton2. Whether Amazon’s deepening chip portfolio becomes a long-term threat to Intel’s CPU business remains to be seen, but the immediate revenue impact can be realistically estimated. In this post, I aim to access the near-term revenue impact of the customized chips on Intel.

Amazon’s Graviton2

This is a custom AWS design that is built using a 7nm manufacturing process. It is based on 64-bit Arm Neoverse cores, delivering up to 7x the performance of the A1 instances. Additional memory channels and double-sized per-core caches speed memory access by up to 5x. The 7nm chip offers up to 40% better performance from comparable x86-based instances, 20% faster than Amazon's first ARM-based chip, Graviton, with a 20% lower cost. AWS services like Amazon Elastic Load Balancing, Amazon ElastiCache, and Amazon Elastic Map Reduce have tested the instances and will move into production next year. At this stage, Graviton2 is still restricted to Amazon’s internal use. Based on the estimates of Bloomberg Supply Chain Database (SPLC), Intel currently has approximately 0.18% or $29 million revenue exposure with Amazon for the most recent quarter (Table 1). It may be safe to speculate that this is the maximum of the immediate revenue impact for Intel with Amazon.

Google’s TPUs

Google’s Tensor Processing Units (TPUs) was created to improve applications that use artificial intelligence to do things like recognize words people are saying in audio recordings, spot objects in photos and videos, and pick up underlying emotions in written text. The chips represent an alternative to Nvidia’s GPUs. Also, if the new version is anything like its predecessor, it will also become accessible to third-party developers through Google’s public cloud service, which could help Google compete with Amazon and Microsoft. The SPLC estimates that Intel has about $162 million or 1.02% revenue exposure from Google (Table 1).

Microsoft’s FPGAs

As the cloud battle heats up, Microsoft develops chips to help developers more quickly perform artificial intelligence computing tasks. An initiative called Project Brainwave lets developers in Microsoft’s data centers use field-programmable gate arrays (FPGAs), which can be customized even after they’ve been plugged into servers. These chips will enable faster processing of images with AI models than what’s available from Amazon Web Services or Google’s cloud. Those companies and others like China’s Alibaba offer an assortment of chips, ranging from GPUs from Nvidia to Google’s TPUs. Microsoft says another advantage to using commercially available FPGAs, is that companies will be able to set up similar technology in their own locations, and not be limited to working in the cloud. Intel appears to have a longer shelf life with Microsoft’s custom chips because Intel is the manufacturer of the FPGA chips. The estimated relationship value is around $173 million or 1.09% of Intel revenue (Table 1).

Apple’s SoC and SiP Chips

Apple (AAPL) has designed System on Chip (SoC) and System in Package (SiP) processors for their mobile consumer devices. They combine a low-power CPU and other components into a physically compact package to meet the stringent power and space constraints common to mobile devices. The Apple "A" series is a family of SoCs used in certain models of the iPhone, iPad, iPod touch, and the Apple TV digital media player. They integrate one (or more) ARM-based CPU, a GPU, cache memory, and other electronics necessary to provide mobile computing functions within a single physical package. Currently, the A series is manufactured by Samsung and TSMG.

Apple is planning to leave Intel and transition to Mac chips starting in 2020, based on multiple rumors that Apple will transition to custom ARM-based chips next year. According to Axios, developers and Intel officials are expecting Apple to begin using ARM-based chips in 2020. The move to ARM-based chips is said to be part of Apple's effort to make Macs, iPhones, and iPads work together and run the same apps. Bloomberg reported that by 2021, Apple wants developers to be able to create one app that will work on iPhones, iPads, and Macs. Apple has been a major Intel customer, responsible for approximately $629 million or 3.98% of Intel's quarterly revenue (Table 1). Obviously, the shift has a significant revenue implication to Intel.

Rough Estimates of the Revenue Impact

As pointed out by previous authors, it is almost impossible to find out the actual number of new servers that large cloud providers installed each year. However, since I am only interested in the relationship between the cloud companies and Intel or AMD, the revenue value that Intel derives from each of the cloud customers should be a good point to start. In terms of “relationship revenue value,” Bloomberg’s Supply Chain Database provides the most accurate estimates. In Table 1, I summarize the quarterly revenue that Intel receives from each customer in both dollar and percentage terms. For example, Amazon gives Intel $29 million or 0.18% of Intel’s revenue per quarter. (Intel’s $29 million from Amazon each quarter, or $116 million annually, is different from the previous estimate of $2 billion.)

In total, for the most recent quarter, Intel has received $364 million (2.3%) quarterly revenue from the 3 large cloud providers. Obviously, the relationship revenue estimates should be the rough estimates for the potential revenue risk that Intel may face as a result of the custom chips since the current revenue is not restricted to only server CPU processor chips, and Bloomberg may have missed some other sources of revenue. But for all practical purposes, these numbers should be a the same order of the actual revenue exposure of Intel to the large cloud companies.

Takeaways

While Amazon’s Graviton2 will be the game changer for the CPU market remains to be seen, the traditional X86ers got to be sensitive to the emerging trend of designing proprietary chips which are specific to individual needs and at a lower cost. Though, the advantage of custom chips that they provide tailored needs to the specific users also becomes the limitation for the wider uses. For no more than 2% revenue exposure, at least for the time being, Intel investors should not worry too much about the immediate revenue impact of the custom chips developed by the large cloud providers. However, merchants developing their own proprietary silicon (like Apple) should be an alarming trend to Intel.

Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. 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.