AMD: The Autonomous Drive to Profitability
By: Sankalp Hariharan & Remmy Martin Kilonzo
The Ivey Business Review is a student publication conceived, designed and managed by Honors Business Administration students at the Ivey Business School.
Advanced Micro Devices (AMD) competes in the semiconductor industry and creates the chips that provide the computational power and intelligence for technologies around us. Its role in supplying companies like Microsoft and Lenovo with processors has landed AMD’s product at the core of essential goods such as gaming consoles and laptops. After reaching its zenith in market share and innovation back in 2006, a decade of product delays and uncompetitive products has resulted in the company’s failure to turn an annual profit since 2011. Despite refreshing its product line-up in 2017, AMD’s current strategy lacks the oversight to put the company on a trajectory of dominance. The rise of artificial intelligence (AI) threatens to upend much of the semiconductor industry, and while competitors have invested heavily to reposition themselves, AMD has remained stagnant and complacent. To continue providing its robust and reliable products, AMD must reposition itself as the leading provider for the promising world of autonomous vehicles (AVs).
The Heart of Modern Technology
Semiconductors are the heart of every electronic device; from smartphones to computers, semiconductors have been essential in pioneering the information age. AMD, Nvidia, and Intel are the major players in the industry, with Intel creating central processing units (CPUs), Nvidia creating discrete graphics processing units (GPUs), and AMD capable of designing both. CPUs provide the logic and computational power for devices to process the workloads that applications demand. As the first-mover in the industry, Intel made the most powerful CPUs and came to dominate the market for desktops and laptops, holding a 96.5-per-cent market share and a near monopoly over data centre CPUs, as of 2016. In contrast to the flexibility of CPUs, GPUs are optimized to perform the calculations needed to render images on a screen. This can range from watching high-definition videos to playing graphically demanding video games. The market has been historically dominated by Nvidia, which controlled about 76 per cent of the market as of 2016. AMD is the only company that creates both types of chips and is consequently the sole provider of accelerated processing units (APUs)—a chip which combines a powerful CPU and GPU. This integration is superior to a separate GPU and CPU because it does not require connections between the two chips, thus, increasing the speed of data flow and improving energy efficiency.
AMD’s Failure to Meet Changing Industry Trends
Within the last decade, the uses for GPUs have increased as big data and AI have gained momentum. While historically GPUs were used for image rendering, they are now at the heart of various AI applications. For example, voice recognition applications like Apple’s Siri or Amazon’s Alexa use GPUs to power their development. Likewise, GPUs power Facebook’s facial recognition abilities. With GPU sales increasing 40 per cent from 2015 to 2016, soaring sales of these chips are the clearest sign of the newly emerging markets that require these powerful products. In this changing environment, Nvidia has been the largest beneficiary; it has invested significantly throughout the last decade to establish its GPUs as the de facto standard for all these new technologies. Even Intel and Qualcomm—who do not produce GPUs—have tried to benefit from this secular trend through various acquisitions. Unfortunately, AMD has fallen behind the pack. Its current focus on desktops and laptops, gaming, and data centres is unsustainable and fails to gain market share in the multi-trillion dollar AI market.
The Current Game Plan
Desktops and laptops were historically one of the most significant end-products for GPUs and CPUs. Unfortunately, with the rise of cheaper and more portable forms of computing, this has led to a decline in unit sales of these products at a negative two per cent compound annual growth rate over the past five years. With such a grim outlook and steep competition from Intel and Nvidia who dominate the market, it seems unlikely that AMD will realize much growth. To make matters worse, AMD has had negative operating margins in this segment. Over the last three years, its average selling prices were 30 and 23 per cent lower than the competitors for CPUs and GPUs respectively, reflecting its historical position as a low-cost, economical chip-maker.
Currently, AMD is the sole provider of custom chips for the two major gaming consoles, PlayStation 4 and Xbox One. However, mobile gaming has transformed the market by putting cheap and often free games in the hands of anyone with a mobile device, reducing the demand for consoles. This shift has increased mobile gaming revenues by more than 42 per cent in 2013 alone, leaving console gaming to more serious gamers. With just 2.3 per cent anticipated growth in 2019, AMD should not expect large gains in this industry and should focus on new markets that have the chance to boost revenues significantly.
Data centres appear to be the most promising part of AMD’s current strategy with the release of a new line of CPUs. For AMD to succeed here, it would be more feasible to expand the market by finding new customers, rather than trying to steal market share from Intel, which has a near monopoly over the industry. While AMD’s new CPUs are on par with Intel’s, the company certainly faces an uphill battle considering this segment is Intel’s gold mine; data centre CPUs provide Intel with more than 58 per cent of its operating income and have the highest margins among other segments. Intel has spent the last decade building relationships, brand recognition, and investing in R&D in this space, and can easily continue to outperform AMD in any of these areas.
Changing Gears
With industry trends pushing GPUs to more exotic and innovative applications, AMD should position itself as a major supplier of these chips in these alternative markets. One of the most widely-accepted applications of AI is within the world of AVs. The opportunity cost of AMD neglecting to serve the AV market is incredibly high; with a total addressable market of more than 18 million vehicle units expected to be sold annually by 2035, the market opportunity for AMD is around $23 billion. AMD needs to venture into this nascent market in order to find new revenue sources with high-growth potential, increase its appetite for risk by venturing out of the safety of its traditional markets, and cement its leadership in a new market just as it did with gaming consoles. With the industry evolving from a competition amongst individuals towards new competitive interactions in the form of partnerships, it is imperative that AMD starts immediately in forming relationships with the leading companies in AV technology.
Two opportunities exist for the company in the world of AVs. The first lies in providing the in-car technology platforms to original equipment manufacturers, namely OEMs such as Toyota and BMW, as well as auto parts suppliers. These companies require custom hardware to power their AV technology and AMD’s experience in providing semi-custom chips and APUs will offer a compelling value proposition. Second, data monetization will be a crucial differentiator among vehicle manufacturers. This requires data analytics and computational power which the OEMs and auto parts suppliers often lack as this is not one of their competencies. With AMD’s new line of data centre CPUs, this will be an ideal opportunity for them to tap into this new emerging market.
Computers on Wheels
Each AV uses a collection of radar, lidar, cameras, and sensors to allow the vehicles to perceive the world around them. Interpreting the tremendous amount of data generated from these sensors requires powerful computers embedded in each AV. These machines are made up of GPUs that perform complex calculations and help the vehicles make safe decisions. Unfortunately, with no standard approach to developing AVs, there cannot be a one-size-fits-all hardware solution. For example, Google exclusively uses lidar while Tesla relies solely on cameras and radar. These different strategies imply that custom solutions are necessary for each of these approaches. Therefore, the primary parties involved in creating AVs—OEMs, auto parts suppliers, and software giants—need a personalized and uniquely tailored AV solution. For software giants like Google, creating their own in-house optimized chip is not a problem; unfortunately for OEMs and auto parts suppliers, manufacturing chips are not one of their core competencies, and thus, they must seek third-party assistance.
This gap in the market is where Nvidia has positioned themselves and has had a first-mover advantage in forming relationships. While Nvidia is the dominant leader, AMD’s core competencies should enable it to have a compelling value proposition and foster partnerships with OEMs and auto parts suppliers. AMD has extensive experience in creating personalized chips for consoles and forming long-term relationships with console manufacturers like Microsoft and Sony. This dedication and commitment is something that OEMs and auto parts suppliers value highly, given how important it is for them to capitalize on AVs.
Each computer within each vehicle will require both a powerful GPU and CPU, and AMD is the only company that can provide this solution in a single chip via an APU. GPUs will be used to analyze the information from sensors and make decisions. CPUs, on the other hand, will be useful for providing in-car entertainment services and media. Entertainment services are expected to increase as telecommunication service providers look to expand their connectivity with vehicles. Internet access will lead to a proliferation of services such as Netflix or Spotify as the vehicles are transformed into entertainment platforms. These services would be accessed by installing software on the AV computer, for which a CPU is needed. In-car features will become significant differentiators for OEMs and auto parts suppliers, as vehicle owners’ attention moves away from speed and performance as they are no longer the ones driving.
Stepping Outside the Vehicle and Into the Data Centre
AVs are expected to generate a substantial amount of data, not only regarding the vehicle and its environment, but also around rider statistics. Gathering and analyzing this data can further improve the overall experience for riders. In order for OEMs and auto parts manufacturers to leverage the data, they have two options: the first is to simply use cloud computing services such as Amazon Web Services or Microsoft Azure. The second is to build the infrastructure in-house to perform the big data analysis. While there are advantages and disadvantages to both options, it is clear that some companies would prefer developing the infrastructure in-house: Ford has announced plans to build a $200-million data centre to support its data-driven strategy. As data centres require both CPUs and GPUs to analyze and process the vast quantities of data, the greatest opportunity for AMD lies with companies like Ford that require both types of processors. Furthermore, this would allow the company to grow in the data centre market without competing with Intel for existing market share. In fact, for companies that do follow this strategy, AMD’s value proposition increases significantly: OEMs and auto parts suppliers can use one chip manufacturer for all their computational needs inside the AV as well as in their data centres.
The Road to Profitability
For more than 250 years, the fundamental driver of economic growth has been technological innovation. With many industry professionals and academics touting AI as one of the most consequential new technologies of the modern era, the demand for AMD’s products and computational power will only increase. Autonomous vehicles are a catalyst in enabling AMD to overtake its competition and ultimately lead the charge in powering the future.