- Tech manufacturers constantly evaluate insourcing versus outsourcing product components based on performance, costs, control, etc. Apple Inc. announced it will design its own microprocessors, and Microsoft Corp. reportedly will do the same.
- Advanced reduced instruction set computer machine (ARM)-based processors--primarily used in smartphones and tablets--are gaining traction against x86 microprocessors where there are high growth prospects and by large cloud and web-based service providers, which can benefit from economies of scale.
- Intel Corp. announced it might outsource manufacturing of its most advanced microprocessors following multiple product delays. Impact on its long-term manufacturing roadmap and processor market share, which has declined in recent years, are key credit risks we continue to monitor.
- We see positive credit trends at Nvidia Corp. due to its considerable technological leadership position in graphics processing unit designs with secular growth trajectory in the data center market, and at Advanced Micro Devices Inc. because of continued share gains in the PC and server processor markets.
Apple Inc. recently announced it will design its own advanced reduced instruction set computer machine (ARM)-based processors for its Mac products, instead of relying on Intel Corp.-based x86 processors. This should not come as a surprise as Apple has long deployed its tightly integrated iOS software and in-house developed ARM processors to drive greater performance, faster development cycles, and an enhanced user experience in its iPhones and iPads.
By bringing microprocessor design in-house, Apple can customize chips to optimize power efficiencies and integrated graphics. It would also further its brand and production differentiation with unique features such as facial recognition, augmented reality, and neural engine (to accelerate certain artificial intelligence software for processing images and speech), in addition to allowing mobile applications to run seamlessly on both its iOS (iPhone and iPad) and MacOS (laptop) operating systems.
Microsoft Corp., according to media reports, is also designing an ARM-based server processor to run its Azure cloud services and another chip for its Surface PCs. In addition to the economies of scale benefits, these decisions are a result of Intel's well-documented production cycle delays by over two years in its latest 10-nanometer (nm) processor node. It has also lost its leadership position in advanced manufacturing process technologies to Taiwan Semiconductor Manufacturing Co. Ltd. (TSMC) since mid-2018.
Both situations could be a hit to Intel's dominance in the microprocessor market.
|Chip Design Insourcing Pros And Cons|
|Insourcing||Complete control over tech specifications (i.e., custom designs) and intellectual properties.||More expensive if volume sub-optimal.|
|Brand and product differentiation.||Access to chip design talent in-house needed.|
|Less costly if at scale.||Reliant on foundry partners for chip manufacturing.|
|Execution risk (i.e., could take longer to design and verify than industry standard chip designs).|
|Outsouring||Benefit from best-of-breed industry designs from established chip suppliers.||More difficult quality control.|
|Usually less costly due to research and development costs spread over more units at chip suppliers.||Reliance on third party for critical component.|
|Opportunity to improve flexibility and streamline production processes.||Added risk of dealing with supply chain and performance monitoring.|
Synopsis Of PC And Data Center Markets
- Historically, Intel has dominated the market for x86 PC central processing units (CPU), with Advanced Micro Devices Inc. (AMD) taking varying share as a second-tier provider. Apple's switch from PowerPC architecture in 2006 to Intel-provided chips further cemented Intel's lead.
- AMD disrupted this with the launch of its Zen architecture-based Ryzen CPU in 2017, a redesign of its core x86 offering that competed with Intel's highest performing consumer offerings at compelling prices. AMD has since introduced four generations of Ryzen PC CPUs and more than doubled its share of the PC market to approximately 20%.
- This surge was further aided by its manufacturing arrangement with TSMC, which implemented and deployed the 7nm process node at scale, giving AMD chips an additional edge in speed and power consumption. Given its scale differential with Intel, we do not believe AMD could have deployed its 7nm product on this timeframe without this partnership.
- Meanwhile, the unexpected stumble in Intel 10nm chip development delayed its node migration by 2-3 years and constrained supply to its PC and, to a lesser extent, server chip customers. This also attributed to AMD's recent share gains. Intel lost its process manufacturing leadership to TSMC a few years ago and is likely to remain behind over the intermediate term.
- Intel is expected to make a formal announcement regarding its outsourcing strategy in early 2021. We believe Intel will likely partner with foundry leader TSMC on 5nm CPU-related products to stem rapid share losses. Key questions and concerns revolve around its longer-term manufacturing strategy and whether a foundry partnership is permanent. If it's temporary, we expect Intel to continue to invest heavily in its chip manufacturing roadmap with a goal of retaking its lead. However, this strategy introduces significant execution risk and potential for major share losses. If it's permanent, it would indicate a loss of manufacturing leadership, which we view as Intel's most important competitive advantage. That would, however, allow the company to stem meaningful near-term share losses. In either scenario, we expect Intel's dominance over the PC and server markets to erode over the intermediate term.
- AMD's success in iterating and improving on its original Zen-based products, continued share gains, and positive response from customers indicates staying power. AMD's significantly improved financial position will bolster its ability to invest in chip design and consolidate market-share gains through a compelling product pipeline. However, we expect each incremental gain in market share to become more challenging.
- Intel's business loss from Apple's move in-house appears meaningful. Apple uses Intel x86 processors for all Mac products, accounting for about 7% of worldwide PC shipments in 2019 and 2020, according IDC. Nevertheless, we expect the impact to be far from detrimental as Intel will maintain its dominant position in the x86 PC processor market. Lost business from Apple will only represent a low-single-digit percentage of its total revenues.
While Apple's announcement is primarily related to PCs, the data center processor market is subject to higher risk of chip design insourcing by original equipment manufacturers (OEM) and original design manufacturers.
Data center market
- Similar to the PC processor market, this market is dominated by Intel with AMD a distant No. 2. However, it is in secular growth, unlike the PC market, which we view to be flat to slightly declining over the longer term.
- The server market is rapidly changing with two distinct factors that could alter the status quo over the longer term. First, as workloads continue to migrate to the public cloud, cloud service providers with significant scale such as Amazon Web Services (AWS), Microsoft Azure, and Google Cloud continue to enjoy strategic and economic benefits in designing proprietary silicon. These providers already design their own servers (versus purchasing from branded manufacturers such as Dell Inc. and Hewlett Packard Enterprise Co.) with processors purchased from Intel or AMD. But they are increasingly turning to ARM-based designs for their internal processor development. ARM-based silicon products are generally more energy efficient, and their slower speed compared to x86 architecture has narrowed in recent years.
- Indeed, AWS has designed its custom ARM-based processors since 2018 and claims its newest Graviton2 delivers 40% better performance at 20% reduced cost compared to Intel chips. Microsoft Azure uses Intel-based processors, but is also seeking to develop ARM-based processors. Google LLC introduced the TensorFlow processing unit (TPU) in 2016 and uses the chip in both data center processing and to make its TPU-powered artificial intelligence (AI) platform available to customers via its cloud-services business.
- In all, ARM-based servers make up a small portion of the market despite compelling performance and cost benefits. But as cloud service providers attempt to take the lead in growth areas such as AI training, 5G network graphics, and autonomous driving, we expect greater custom chip development over time.
- As Moore's law--the observation that the number of transistors in a dense integrated circuit doubles about every two years--begins to bump up against physical limits and the cost of developing new process nodes rises exponentially, enterprises are increasingly looking to alternatives to CPUs to handle expanding and increasingly complex compute workloads.
- While we expect CPUs will remain at the core of compute architecture in data centers, enterprises are responding to progressively more difficult CPU performance improvements by deploying an increasingly diverse array of compute architectures to achieve performance gains through chip designs optimized for specific workloads. So far, the greatest example of this is the use of graphics processing units (GPU) in data centers at scale to accelerate high-performance computing tasks, most significantly the training of deep-learning AI algorithms.
- Beyond GPUs, expect increasing use of field-programmable gate arrays (FPGA), AI-focused application-specific integrated circuits, and other specialized architectures to supplement CPUs in the post-Moore's law world. Nvidia Corp.'s announced acquisition of semiconductor design firm ARM Holdings PLC and AMD's acquisition of Xilinx Inc. illustrates this point. Both companies seek to compete in the data center market by offering a complete stack of core CPU compute products and specialized acceleration chips.
- We expect Intel will remain dominant in the server processor market, but could face market share loss over the near term. Its recent steps to develop a compute-focused GPU and acquisition of Altera Corp. further highlight the need to offer more than a CPU to compete.
Impact On Wafer Fab Equipment Vendors
We expect no material change to overall wafer fab equipment (WFE) spending over the intermediate term as a result of potential outsourcing by Intel or proprietary chip designs by OEMs. We expect a modest increase by TSMC given its increasing list of customers, including Apple, with relatively flat spending by Intel during fiscal 2021. However, Intel's capital expenditures could decline in subsequent years if it outsources more of its manufacturing operations.
We expect good growth in both DRAM and NAND spending, but native China spending could decline in 2021 due to the ongoing U.S. ban on supplying semiconductor capital equipment to Chinese companies, including its largest chip manufacturer, Semiconductor Manufacturing International Corp. Key U.S. WFE providers, including Applied Materials Inc., Lam Research Corp., and KLA Corp., derive 20%-30% of their revenues from China (both native and multinationals). China accounts for more than 20% of overall WFE spending, hence its relationship with President-elect Joe Biden's incoming administration could play an important role in the overall industry growth trajectory over the intermediate term.
Impact On Leading-Edge Semiconductor Firms
OEM insourcing PC or server processor designs would be a headwind to x86 chip vendors such as Intel and AMD. In our view, given their secular growth prospects and rising data center server needs, cloud and web service providers (such as AWS, Microsoft, Alphabet Inc., Facebook Inc., Alibaba Group Holding Ltd., Tencent Holdings Ltd., and Baidu Inc.) would be more inclined to design their own chips to optimize use cases and energy and cost efficiencies. ARM Holdings is one of the largest providers of semiconductor intellectual property, and it designs the microprocessors used in over 90% of smartphones worldwide. The company already licenses to vendors such as Apple and semiconductor design firms such as Qualcomm Inc., and Nvidia. It should increasingly benefit from higher license sales to cloud and web service providers. ARM-based processors require fewer instructions and operate on lower power compared to x86 processors, have attributes such as low latency (they work faster on a single operation), and can be designed in a systems-on-chip architecture with other analog, mixed-signal, or memory chips to maximize performance.
We believe business demands would increase for the world's largest semiconductor foundry service provider as in-house chip design gains traction. TSMC is the clear leader in most advance nodes, with market share estimated at over 80% at the most advanced node. Samsung Electronics Co. Ltd. should benefit along with the other major foundry service providers.
It could be another beneficiary, regardless of its pending acquisition of ARM Holdings. Nvidia has a relatively short history in the data center market, but the increasing use cases for its GPUs in machine learning and data center processor acceleration should allow it to outgrow the market over the intermediate term. If the acquisition is approved, it will provide Nvidia a boost in growth areas such as AI, self-driving vehicles, and smart devices.
Given its lack of exposure to ARM-architecture markets, we expect limited near-term impact from the Apple move. It could pose a challenge if this is a harbinger of a broader shift away from x86 chips in PCs and the data center market. We think AMD will at least sustain its share of PC CPU markets over the next two years. The integration of Xilinx should provide additional ability to compete in the data center market, where it has historically lagged Intel by greater margins than in PCs. Increasing use of diverse, application-focused compute silicon should bolster AMD's market position, as it is the only such vendor capable of offering a full stack of CPU/GPU/FPGA products.
We expect Intel will maintain a dominant position in both PC (near 80% market share for laptops) and server (over 90%) processor markets, but its market shares will continue to gradually erode over the near to intermediate term as it loses ground to AMD despite its architectural leadership and ecosystem advantage. Intel's likely use of TSMC as a foundry partner may help stem rapid share losses, but manufacturing leadership will be difficult to recapture over the intermediate term given how quickly it fell behind TSMC and Samsung. On the server side, AMD's success with its Epyc processors and hyperscale cloud service providers' continued move to bring chip design in-house using ARM architecture will reduce reliance on Intel processors over time.
Recent negative news notwithstanding, Intel has successfully diversified beyond the PC market, with non-PC revenues accounting for about half of total revenues through the first nine months of 2020 from 42% in fiscal 2015 and just 28% in fiscal 2010. Its revenues have also increased nearly $30 billion over the past 10 years due mostly to growth in data center-related businesses. We also note Intel's market share has swung over time with AMD gaining meaningful PC shares, followed by rapid reversals. Lastly, its non-PC business consists of not just server processors but also nascent businesses in autonomous driving, 5G networking, AI training, and advanced graphics. These are modest in scale today but enhance its overall total addressable market and growth potential.
Near-Term Rating Impact
Hardware manufacturers and large cloud and web service providers
Chip design insourcing would be highly strategic to hardware manufacturers (e.g., Apple) and large cloud and web service providers (e.g., Amazon, Microsoft, Google, Facebook, Alibaba, Tencent, Baidu), but unlikely a key rating driver near term given the breadth and depth of their businesses. Benefits from brand and product differentiations are easier to achieve. However, it will take time to ascertain whether it's the correct decision due to the high investment requirements, long payback period, and many intangible factors. For example, in hindsight, Apple's decision to develop and maintain its proprietary operating software for Mac desktops was a good one, as it likely led to an easier decision to also develop its proprietary mobile operating software. Similarly, Google's development of the Android mobile operating system based on open-source software was a huge success as it allowed significant first-mover advantage. An estimated 85% of global smartphones have the Android operating system installed today, enabling Google to generate tens of billions of dollars in revenues from mobile advertising and application sales.
Its share gains in the PC and server processor markets and material financial improvements over the past few years led to multiple-notch upgrade to 'BB+' from 'B-' in March 2017. We view favorably the Xilinx acquisition in an all-stock transaction, which adds to AMD's scale and reduces its reliance from core CPU and GPU businesses by adding Xilinx's FPGA products in communications, industrial, and automotive end markets. Furthermore, it should provide greater stability to AMD's historically volatile earnings. Our 'BB+' rating is on CreditWatch with positive implications, reflecting our view that the combined entity will have considerable near-term organic growth trajectory and indicates a likely upgrade by up to two notches when the transaction is completed, which is expected in 2021. We could upgrade the company before then if AMD continues to report strong stand-alone operating performance, including further share gains against Intel and Nvidia, while maintaining a net cash financial position.
We've also recognized its impressive operating performance over the past few years and raised our rating on the company to 'A-' from 'BBB-' in September 2016. We expect the company to sustain its considerable technological leadership position in GPU designs, with increasing use cases in gaming, machine learning, and data center processing acceleration, supporting its secular revenue growth trajectory. Fiscal 2021 (ending in January) revenue will likely exceed $16 billion, over 45% improvement versus the prior year, attributable to strong demand for high-end GPUs in gaming, sales ramp-up of its data center processors, and contribution from its Mellanox acquisition closed in April 2020. Nvidia's pending acquisition of ARM Holdings, announced in September, for $40 billion is pending regulatory approval.
Nvidia is expected to fund the transaction with $12 billion cash and potentially an additional $5 billion cash payment based on performance-based earnouts. The remainder would be funded with stock. We believe Nvidia will maintain a modest net cash position following the close and remain well below our 1x leverage downgrade trigger. The growth trends in GPU use cases for machine learning and artificial intelligence applications in data centers will likely provide a boost for years. Given the company's already conservative balance sheet, we do not expect its improving financial position to be the most likely path to an upgrade. Rather, we find Nvidia's success in lowering its reliance on its core gaming market, currently about 48% of total revenues, and volatile earnings history to be key.
Our stable rating outlook on Intel reflects its leading scale among semiconductor manufacturers, good prospects for continued product diversification, and consistent cash flow generation despite recent market share losses due to manufacturing challenges. We see two paths to potential rating downside. First is if it sustains adjusted leverage above 1.5x due to aggressive financial policies, including significant share repurchases or large debt-funded acquisitions. Intel already repurchased more than $12 billion of shares through the first nine months of fiscal 2020, and we believe it could continue to aggressively repurchase shares given its equity underperformance versus peers. However, given adjusted leverage near 0.5x, we believe share repurchases alone are unlikely to trigger a negative rating action.
Second is a sustained weakening of its competitive position. We take a holistic view of Intel's business over an extended period, including its successful diversification into the data center arena over many years as well as its recent loss of manufacturing leadership. We will continue to monitor Intel's business prospects, with potential for a negative rating action should it lose significant market share with limited prospects for recovery. We view positively Intel's announcement of new CEO Pat Gelsinger given his technical background and deep Intel history, although we are uncertain as to his influence over the near-term decision regarding its product roadmap, which we expect shortly.
This report does not constitute a rating action.
|Primary Credit Analysts:||David T Tsui, CFA, CPA, San Francisco + 1 415-371-5063;|
|Andrew Chang, San Francisco + 1 (415) 371 5043;|
|James W Thomas, New York + (212) 438-0181;|
No content (including ratings, credit-related analyses and data, valuations, model, software or other application or output therefrom) or any part thereof (Content) may be modified, reverse engineered, reproduced or distributed in any form by any means, or stored in a database or retrieval system, without the prior written permission of Standard & Poor’s Financial Services LLC or its affiliates (collectively, S&P). The Content shall not be used for any unlawful or unauthorized purposes. S&P and any third-party providers, as well as their directors, officers, shareholders, employees or agents (collectively S&P Parties) do not guarantee the accuracy, completeness, timeliness or availability of the Content. S&P Parties are not responsible for any errors or omissions (negligent or otherwise), regardless of the cause, for the results obtained from the use of the Content, or for the security or maintenance of any data input by the user. The Content is provided on an “as is” basis. S&P PARTIES DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, ANY WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE OR USE, FREEDOM FROM BUGS, SOFTWARE ERRORS OR DEFECTS, THAT THE CONTENT’S FUNCTIONING WILL BE UNINTERRUPTED OR THAT THE CONTENT WILL OPERATE WITH ANY SOFTWARE OR HARDWARE CONFIGURATION. In no event shall S&P Parties be liable to any party for any direct, indirect, incidental, exemplary, compensatory, punitive, special or consequential damages, costs, expenses, legal fees, or losses (including, without limitation, lost income or lost profits and opportunity costs or losses caused by negligence) in connection with any use of the Content even if advised of the possibility of such damages.
Credit-related and other analyses, including ratings, and statements in the Content are statements of opinion as of the date they are expressed and not statements of fact. S&P’s opinions, analyses and rating acknowledgment decisions (described below) are not recommendations to purchase, hold, or sell any securities or to make any investment decisions, and do not address the suitability of any security. S&P assumes no obligation to update the Content following publication in any form or format. The Content should not be relied on and is not a substitute for the skill, judgment and experience of the user, its management, employees, advisors and/or clients when making investment and other business decisions. S&P does not act as a fiduciary or an investment advisor except where registered as such. While S&P has obtained information from sources it believes to be reliable, S&P does not perform an audit and undertakes no duty of due diligence or independent verification of any information it receives. Rating-related publications may be published for a variety of reasons that are not necessarily dependent on action by rating committees, including, but not limited to, the publication of a periodic update on a credit rating and related analyses.
To the extent that regulatory authorities allow a rating agency to acknowledge in one jurisdiction a rating issued in another jurisdiction for certain regulatory purposes, S&P reserves the right to assign, withdraw or suspend such acknowledgment at any time and in its sole discretion. S&P Parties disclaim any duty whatsoever arising out of the assignment, withdrawal or suspension of an acknowledgment as well as any liability for any damage alleged to have been suffered on account thereof.
S&P keeps certain activities of its business units separate from each other in order to preserve the independence and objectivity of their respective activities. As a result, certain business units of S&P may have information that is not available to other S&P business units. S&P has established policies and procedures to maintain the confidentiality of certain non-public information received in connection with each analytical process.
S&P may receive compensation for its ratings and certain analyses, normally from issuers or underwriters of securities or from obligors. S&P reserves the right to disseminate its opinions and analyses. S&P's public ratings and analyses are made available on its Web sites, www.standardandpoors.com (free of charge), and www.ratingsdirect.com and www.globalcreditportal.com (subscription), and may be distributed through other means, including via S&P publications and third-party redistributors. Additional information about our ratings fees is available at www.standardandpoors.com/usratingsfees.
Any Passwords/user IDs issued by S&P to users are single user-dedicated and may ONLY be used by the individual to whom they have been assigned. No sharing of passwords/user IDs and no simultaneous access via the same password/user ID is permitted. To reprint, translate, or use the data or information other than as provided herein, contact S&P Global Ratings, Client Services, 55 Water Street, New York, NY 10041; (1) 212-438-7280 or by e-mail to: email@example.com.