hapabapa
The momentum in semiconductor stocks in recent weeks marks a stark shift from last year’s landslide losses. Investors have largely been celebrating the improving prospects of a “less hawkish” Fed ahead of upcoming rate decisions as inflationary pressures continue to ebb. Not only has this lifted pressure from the discounting factor on tech stocks that have much of their valuation outlooks pinned on future cash flows, but also signal a potential “soft landing” that could tame inflation without stifling growth.
Yet, the recent run of mediocre earnings forecasts and warnings of a worsening slowdown across industry constituents continue to underscore that impact from the weakening consumer is only just beginning with potential for further deterioration ahead. This is corroborated by not only the slowdown in processor demand across consumer end-markets like PC and smartphone applications, but also in the previously more resilient enterprise end-markets, as the impact of the deteriorating macroeconomy spread upstream to impact IT investments.
And this is exactly the case being observed at Advanced Micro Devices, Inc. (NASDAQ:AMD) coming out of its latest earnings release. Consistent with our previous coverage on the stock, AMD’s data center sales – a relative bright spot – remain vulnerable to faltering hyperscaler demand as industry monikers warn of macro-driven moderation ahead. For instance, Microsoft Corporation (MSFT) warned of deceleration in its Azure cloud-computing business in the months ahead as demand normalizes on the back of growing enterprise calls for “cloud optimization” – a quest to enhance “applications, performance, and business needs in the cloud while eliminating costs and inefficiencies.” This is consistent with market leader Amazon Web Services’ (AMZN) deceleration observed in the third quarter, marking a largely expected feat given its sprawling share of the industry. Server chip market leader Intel (INTC) has also recently confirmed that the “overall consumption TAM [in the server market] has [grown] modestly in calendar year 2022, albeit at diminishing rates as the year progressed.”
Despite being one of the hardest hit cohorts in 2022, the semiconductor sector is likely far from ripe for a rebound from current levels, as industry and market headwinds stiffen. And as bad as it sounds, the current market climate has created potential opportunities to buy-in cheaper on stocks that continue to exhibit a sustainable trajectory of longer-term upside potential. Considering AMD’s continued fundamental outperformance – marked by consistent market share gains, a resilient bottom-line (despite one-time amortization costs pertaining to Xilinx acquisition that weighed on fourth quarter margins), and ongoing commitment to innovation – the stock’s vulnerability to persistent market volatility in the coming months creates a compelling opportunity to partake in its run as a potential gainer from key longer-term growth trends ahead.
AMD Remains a Victim to the Slowing Consumer
Despite mounting macroeconomic uncertainties still, spanning persistent inflation that runs well above the Fed’s liking, an ongoing rate hike cycle, and a worsening consumer slowdown that is heralding an impending downturn, investors have been eager in “scooping up tech stocks” this year. Specifically, the tech-heavy Nasdaq 100 has been on a roll in reversing some of last year’s disheartening downtrend, with year-to-date gains of as much as 12%. Growing optimism in recent weeks that the Fed may be nearing the end of the aggressive rate hike cycle that started last year, paired with AMD’s fourth quarter double beat has also spurred the stock’s post-earnings rally in late market trading, despite management’s conservatism on the company’s near-term forward outlook with expectations for a revenue decline in the current quarter due to continued “weakness in the PC and gaming segments.”
But the “hope for better days” likely remains premature, as the earnings season so far has been marked by a repeating theme of grim forecasts from management in hopes of tempering investors’ expectations in response to low market visibility ahead. This is consistent with the flurry of earnings estimate reductions over the fourth quarter, which averaged -6.5% compared to historical trends of -2.5% over the past five years. Full year earnings estimates for the next 12 months have also been slashed by -4.4% on average, worsening from historical trends of -0.2% over the past five years as deteriorating macroeconomic conditions sow a “challenging environment for corporations to grow profits.” And a deeper dive into each of AMD’s core segments would be explicable of management’s conservatism as the broader industry inches further into an uncertain macro backdrop:
Data Center (Enterprise)
AMD’s data center sales have been critical to its bottom-line in recent quarters given the segment’s higher margin nature. Yet, the segment’s slight deceleration in the fourth quarter underscores its vulnerability to a materializing hyperscale slowdown over the near-term, a headwind to its market-leading margins (despite Intel’s leading market share in server processors, AMD beats its rival to the curb on margins by at least 10 points), and inadvertently, the stock’s near-term outlook, potentially erasing recent gains.
Specifically, Microsoft – a customer of AMD’s EPYC server processors – has recently warned of deceleration in its Azure cloud-computing business, as the enterprise sector “hits the brakes on spending”:
“During the pandemic there was rapid acceleration. I think we’re going to go through a phase today where there is some amount of normalization in demand,” [CEO Satya Nadella] said in an interview at the World Economic Forum in Davos, Switzerland, earlier this month. “We will have to do more with less — we will have to show our own productivity gains with our own technology.”
Source: Bloomberg News.
This is also consistent with AWS’ recent warnings of an “uptick in customers focused on controlling costs” in response to ongoing macroeconomic uncertainties. While both, alongside other tech giants, remain committed to maintaining investments towards the expansion of data center capacity that “deliver cloud services” over the longer-term, the near-term industry conservatism remains an overhang for the demand outlook on AMD’s server processors. This is further corroborated by rival and industry leader Intel’s recent observation for deceleration in data center processor demand that will likely worsen over the coming months as industry digests excess capacity built over previous years in response to the looming economic downturn:
We expect Q1 server consumption TAM to decline both sequentially and year-over-year at an accelerated rate with first half 2023 server consumption TAM down year-on-year before returning to growth in the second half.
Source: Intel 4Q22 Earnings Call Transcript.
But despite the anticipated near-term contraction in server processor demand, AMD’s share continues to show signs of gradual increase. Notwithstanding the near-term slowdown in cloud-computing demand due to both market-driven uncertainty in IT budgets and a general shift in corporate preference for cloud optimization, which could further impact AMD’s near-term data center sales, the appetite from high-performance computing (“HPC”) remains resilient and could be a potential offset, which is in line with AMD’s expectation for continued growth in the segment. The 4th Gen EPYC processors have already started shipping, counting key hyperscalers like Azure, Google Cloud Platform (GOOG, GOOGL) and Oracle Cloud as key customers. The EPYC server processors can now also be found in “101 supercomputers in the latest Top500 list of the most powerful supercomputers in the world,” up from 73 in the prior year.
Specially, AMD’s newest generation of EPYC sever processors continue to hold a technology advantage over equivalents made but key rival Intel. The 4th Gen EPYC processors – a line-up marked by Genoa, Bergamo, Genoa-X, and Siena variations optimized for different applications – are manufactured on TSMC’s (TSM) 5 nm process. Not only do the latest generation of EPYC server processors boast 25% more performance-per-watt and 35% better performance overall compared to the preceding 3rd Gen Milan EPYC processors, they also outperform Intel’s equivalent – the fourth generation Sapphire Rapids Xeon server processors – which are still being manufactured on the 7nm process. As discussed in a previous coverage, smaller nm processors are more powerful because it reduces the space between the high volume of transistors within a single chip, and effectively lowers the “distance travelled by electronics to perform work,” thus enhancing energy efficiency while enabling faster computing. Although AMD has only recently started to ship the 4th Gen line-up of EPYC processors to customers, the company is already said to be working on the 5th Gen, codenamed “Turin,” which will likely be manufactured on TSMC’s 4nm and 3nm processes to unlock greater performance capabilities that will further leave rival offerings behind by wide margins, and bolster continued market share gains as secular growth trends in HPC persist.
In addition to data center CPUs, AMD’s growing prowess in data center GPUs also underscores its reach into growth opportunities stemming from burgeoning AI momentum. As discussed in a previous coverage on the stock, AMD’s latest Instinct MI200 series accelerators operates at more than double the performance of rival Nvidia’s (NVDA) “Ampere 100” data center GPU. This accordingly makes it an attractive solution for “complex computing tasks needed to power AI applications”, such as large language models (“LLMs”) backing the recent ChatGPT frenzy:
More notably, the AMD Instinct accelerators are known for their integration alongside the EPYC server processors in more than 100 supercomputers in the most recent Top500 list, up from 73 in the prior year. The achievement underscores its performance competency in supporting complex workloads spanning “climate, biology, and medicine, new energies and materials”, inclusive of application-specific LLMs, to accelerate discovery.
And in the latest development, AMD has introduced the next-generation MI300 series accelerators – its “first integrated data center CPU and GPU” that combines the power of 4th Gen EPYC server processors and CDNA 3 architecture-based GPUs – which is scheduled to start shipping by the second of the current year to further take advantage of opportunities in the “exascale AI era.” Continued integration of Xilinx’s “differentiated AI engine” across AMD’s CPU product portfolio is also expected to further the company’s “industry-leading inference capabilities,” with one of the first joint products – the XDNA Architecture – optimized for complex workloads such as AI and signal processing scheduled for launch in 2023. Meanwhile, AMD recently introduced Versal SoCs to deliver AI-enabled “high-quality, low-latency imaging” capabilities in healthcare, as well as aerospace application.
Based on preliminary Wall Street projections over the nascent AI opportunity pertaining to the use of LLM applications like OpenAI’s ChatGPT alone could drive annual sales of $3 billion to $11 billion in the current year. Although AMD’s presence in the market for data center GPUs remains in miniscule when compared to industry leader Nvidia, which commands more than a 90% share, its technological competence continues to bolster its presence in the ongoing AI arms race still, despite the transitory macro-driven setback expected over the coming months.
Gaming and Client (Consumer-Centric)
Just when market thought the adverse impact from slowing consumer end-markets have already been priced into the broader semiconductor sector following an aggressive valuation correction in 2022 that outpaced the broader market, results from the recent earnings season proved otherwise. Memory chipmakers – the corner of the broader semiconductor sector that is most prone to consumer weakness – is expected to incur a collective loss of more than $5 billion this year. The extent of turmoil is further echoed by Samsung (OTCPK: SSNLF / OTCPK: SSNNF) – the global leader in contracted memory chip manufacturing – which suffered from its worst profit decline in more than 10 years, while foundry giant TSMC also warned of a potential drop in current quarter sales for the first time in four years.
Even chipmakers counting AMD and rival Intel have not been spared, given the accelerating decline in PC shipments. Specifically, global PC shipments accelerated a decline to -28.5% in the fourth quarter from -19.5% in the third quarter, -12.6% in the second quarter, and -6.8% in the first quarter; full year 2022 PC shipments dropped 16.2% y/y, the steepest level from data dating back to the mid-1990s, underscoring the weight of worsening consumer weakness. The slowdown is further corroborated by AMD’s fourth quarter revenue declines observed in its consumer-centric gaming (-7% y/y; +0.8% q/q) and client segments (-51% y/y; -12% q/q).
And the challenge remains prevalent, which is consistent with management’s conservative sentiment over its consumer-facing segments. The U.S. economy is expected to contract in the second and third quarter by an annualized rate of 0.6% and 0.3%, respectively, despite its latest outperformance. Much of the anticipated downturn will likely be driven by buckling consumer resilience. Specifically, American household savings have come down significantly from the pandemic era, when bank accounts were buffered with unspent stimulus cheques due to COVID-related mobility restrictions, hovering near record-low levels in the low 2% range. Meanwhile, credit card debt has steadily racked up beyond pre-pandemic levels, with accumulated revolving credit outstanding reaching $1.2 trillion.
The threatening combination of declining savings and increasing consumer debt underscores the growing burden of inflation and surging interest rates. With the Fed likely laying its eyes on slowing the relatively resilient labor market next – unemployment remains at a historical low rate of 3.5% – to impede a wage-induced price spiral, related uncertainty is likely to persuade consumers to tighten their belts further within the near-term. Taken together with broader weakness in the global economy, which the World Bank expects to “decelerate sharply” with an added warning that “the world faces a tough year [that will be] tougher than the year we leave behind,” AMD management’s continued caution over its consumer-facing segments’ near-term vulnerability to weakness almost seems prudent, rather than conservative. Tempering any premature optimism among investors, while taking into serious consideration the near-term uncertainties over the gaming and client segments’ demand environment and broader market climate, could help AMD mitigate risks of a bigger setback ahead of a potential continuation in underperformance.
Embedded
AMD’s acquisition of Xilinx – a market leader in FPGAs and SoCs – last year marked a critical step in expanding its foray in applications across aerospace, automotive, cloud-computing and communications infrastructure. AMD’s embedded segment sales during the fourth quarter underscores continued resilience against the broader industry slowdown. Embedded sales grew 7% q/q to $1.4 billion in the fourth quarter, which is in line with resilience observed at rival Intel’s “programmable solutions group” that houses its sales of similar programmable semiconductor products.
The results corroborate consistent positive progress in AMD’s integration of Xilinx following the transaction’s completion last year, as evidenced by the quick introduction and roll-out of several products in recent months that rely on the latter’s technologies. These include the Versal SoCs as discussed in the earlier section, as well as the AMD Xilinx Automotive (XA) Zync UltraScale+ MPSoC platform designed for autonomous mobility applications. The development follows AMD’s confirmation of extended support for “all [Xilinx] 7 series FPGAs and adaptive SoCs through at least 2035.” And previously expected synergies are already materializing, as observed by AMD’s recent extension of collaborations with 5G mobile network provider Viettel High Tech on the deployment and application of the Zynq UltraScale+ MPSoC devices, as well as with mobility supplier DENSO Corporation (OTCPK: DNZOF / OTCPK: DNZOY) on the application of XA Zync UltraScale+ MPSoCs into its “next-generation LiDAR platform.”
Nonetheless, the segment is not immune to near-term macro headwinds. In addition, we also view AMD’s anticipated ramp-up of Xilinx integration risking exposure to a partial cyclical downturn as yearslong 5G network infrastructure investment cycles draw close to an end. Specifically, American telco giants including Verizon (VZ), AT&T (T), and T-Mobile (TMUS) are gearing up for a transition away from its 5G-heavy investment cycle in past years towards greater focus on monetization, as they wrap up the last of their respective network build-out efforts later this year. But demand for 5G solutions provided by Xilinx remain resilient among network servers providers, which is consistent with its recently extended collaboration with Viettel, as they remain critical to supporting the next-generation network infrastructure’s functions. And on the auto front, although easing supply chain constraints and key component shortages are buoying a recovery in sector, near-term consumer weakness as explained in the earlier section also risks a swift reversal. This could pose a potential near-term risk to AMD’s extended foray into related growth opportunities enabled by Xilinx on auto-specific products such as the recently introduced XA Zync UltraScale+ MPSoCs (see more here for our analysis on the near-term autonomous mobility outlook).
Meanwhile, AMD’s adaptive compute acceleration platform (“ACAP”) products enabled by Xilinx will likely remain a bright spot, considering burgeoning AI momentum as discussed in the earlier section. Specifically, Xilinx’s proprietary “AI Engine” (“AIE”) technology, which is built on its programmable Versal ACAP architecture to accommodate a wide range of use cases and workloads, will be key to driving the next-generation XDNA Architecture. The XDNA Architecture will integrate Xilinx’s AI expertise across AMD’s existing processor line-up, starting with the Ryzen desktop CPUs this year. Specifically, the recently introduced AMD Alveo V70 AI Accelerator is built based on the newest XDNA architecture, enabling “industry-leading performance and energy efficiency for multiple AI inference workloads.”
We view this development to have come at an opportune time, as AI-enabled applications gain momentum following the recent release of ChatGPT, which has garnered attention to advances made in HPC in recent years and applicability of said developments in day-to-day settings. With close to a third of recently surveyed American professionals indicating they have used “ChatGPT or another artificial intelligence program in their work,” continued integration of Xilinx’s AI technology – such as the latest Alveo V70 AI Accelerator, which can facilitate wide-ranging applications including natural language processing and LLMs – into AMD’s product portfolio will likely grow into a key competitive advantage over the longer-term. The hardware development will be further complemented by Xilinx’s AI software expertise, with planned future integration across the EYPC line-up expected to further embolden AMD’s position in HPC.
The Bottom Line
The semiconductor sector is likely not yet entirely out of the woods, despite what the recent rally might suggest. Consumer end-markets in particular are likely to deteriorate further, with spill over risks into enterprise segments that is already taking shape. This will continue to elevate vulnerability of previously more resilient corners of the industry – like data center demand – to near-term macro-driven deterioration.
Although persistent fragility in market sentiment in response to mounting macroeconomic uncertainties on the horizon still will likely to continue to dictate the Advanced Micro Devices, Inc. stock’s near-term performance, its longer-term growth trajectory remains well sustained and intact. Its recent engagement in key industry consolidations have also been met with positive integration progress. In addition to Xilinx, continued progress made in materializing on synergies with the Pensando acquisition is likely also taking shape, considering robust performance in the data center segment which houses sales of its data processing units (“DPUs”), likely driven by resilience from the cybersecurity vertical.
For now, Advanced Micro Devices, Inc. stock likely remains victim to a deteriorating market backdrop. A recent pulse survey shows that most investors expect further market declines to come as they anticipate a greater impact to corporate earnings ahead of tightening financial conditions. Most do not expect a structural market recovery until at least later in the year, implying that the stock’s recent rally likely faces greater hurdles to come. This continues to create an opportunity for potential pullbacks in AMD shares’ performance in the near-term, in tandem with anticipated market response to the dire macro backdrop. AMD currently trades at 24x estimated earnings, outperforming the Nasdaq 100’s 21x – and reasonably so given the underlying business’ relative resilience to peers, sustainable growth outlook, and robust free cash flows.
With the tech-heavy index currently outperforming its 10-year average P/E ratio of 20.5x, and the continuation of near-term earnings corrections discussed earlier likely to drive the multiple higher, thus increasing its vulnerability to a further correction ahead of the looming downturn, the ensuing drag could potentially setback the AMD stock’s performance over the coming months as well. This would create a compelling opportunity for investors to better partake in the underlying AMD business’ longer-term upside potential.
Editor’s Note: This article discusses one or more securities that do not trade on a major U.S. exchange. Please be aware of the risks associated with these stocks.