Justin Sullivan
Advanced Micro Devices, Inc. (NASDAQ:AMD) has grown revenues 85% CAGR the past 2 years despite being supply constrained. The company is gaining market share and increasing margins, and has a lot of room to grow. The company is likely to be earning at a $10/share rate in a year. So why is the stock trading at less than 7x my earnings outlook? The market is failing to see what is really going on.
Market Mistakes Semi Inventory Correction for End of Cycle
For a couple years now, we’ve seen people calling for the end of the semiconductor market cycle. Those calls have intensified now as the industry has entered an inventory correction. What happened is that a lot of demand for PCs was pulled forward due to COVID and work/learn from home needs increased. And as supply chains were disrupted, manufacturers increased inventories. This has been a well-known phenomenon, and widely talked about the past couple years. The correction is playing out as expected.
Meanwhile, while there are pockets of weakness in PCs (COVID reset), and gaming (Ethereum demand ending), overall demand is still running along. The economy is still running hot, and unemployment is still close to record lows at 3.7%. There is little reason to expect the end of the semiconductor cycle to be at hand.
Nevertheless, some market participants have been indiscriminately hammering semis, and sending AMD down without respect to its strong fundamentals.
Market Mistakes Semi Weakness for AMD Weakness
COVID-related PC demand and Ethereum mining demand are the two major forces driving the inventory correction. As a result, a select set of semiconductor firms heavily dependent on those sources of demand have missed earnings and had to bring their guidance down dramatically. Intel (INTC) and Nvidia (NVDA) are firms with much more concentrated exposure to these two demand drop-offs, and they sold off as a result.
But other firms, like AMD, are not as susceptible to these sources of demand weakness. During their Q2 earnings, AMD noted that they were seeing weakness in PCs and Gaming and guided with that data in mind. And around a week ago, with only 2 weeks left in Q3, AMD management implicitly reaffirmed their Q3 guide and forward looking growth prospects, in large part thanks to strength in the high-margin datacenter.
It makes sense to sell INTC and NVDA on their weakness, especially as they should continue to lose market share to AMD. It does not make sense to sell AMD based on the weakness of competitors that AMD is taking share from.
Market Mistakenly Thinks AMD/NVDA Should Trade Together
Having followed AMD and NVDA daily for years, I can attest that the two stocks are often traded in tandem. The market got used to this kind of activity and appears to be continuing it even when it no longer makes sense.
While AMD used to be heavily exposed to gaming GPU sales, now that is a small minority of their sales. AMD doesn’t disclose gaming GPU sales revenues on their own, but a reasonable estimate is they comprised 15% of revenues in Q1. Since AMD went to meaningful lengths to focus on supplying GPUs to OEMs and gamers, likely half or more of those sales were to actual gamers (instead of for crypto mining).
In contrast, GPUs in the Gaming and Pro Viz segments comprised the majority of Nvidia’s revenues in Q1. And, Nvidia was shrewdly taking advantage of the crypto mining demand, and made a ton of money selling to miners. It now appears that the majority of their Gaming and Pro Viz segments was actually for crypto mining, and not actual gamers and designers. Now that the market is starting to realize that Nvidia’s sales were substantially inflated and unsustainable, the stock is down 65% from its highs. This is reasonable given how obscene NVDA’s valuation was, at over 100x TTM earnings at the peak—and much of those earnings was from crypto mining demand that is gone permanently.
But selling off AMD to this degree is nonsensical. AMD does not have close to the same level of crypto exposure that Nvidia did. For Q2, AMD beat earnings and guided higher. For Q2, NVDA had horrendous earnings, guided lower, and then guided lower again due to Chinese sales restrictions (which AMD disclosed has no material impact on their business).
The market has mistakenly sold AMD down as NVDA has fallen, despite widely divergent fundamentals. This error should be reversed in time.
Market Mistakenly Thinks AMD is High PE Stock
AMD has historically been a high P/E stock due to its strong growth prospects. The market appears to mistakenly believe that it is still a high P/E stock and should be traded like one, sending it down 5% in a day on a whim. I also see traders including AMD in their list of high P/E stocks to sell on weakness in the indexes. The market appears to be slow to catch on that AMD is now trading at about 15 times current earnings—below the average P/E ratio for the S&P 500, and Nasdaq.
Perhaps traders need to update their P/E ratio calculations. Or perhaps the market is confused because GAAP earnings are significantly lower than non-GAAP earnings (and Free Cash Flow) due to amortizing the charges related to the Xilinx acquisition. Regardless of the reason, I expect that this error will be corrected over time as earnings keep growing and the company buys back more and more of the stock float.
Reality: AMD Grew Earnings, Guided Higher, and Implicitly Reaffirmed Guide Recently
The reality is that while certain other firms were seeing sharp falloffs in their business, AMD was still growing. During their Q2 earnings call, AMD management noted that they were seeing weakness in PCs and Gaming, and incorporated that into their Q3 guide. And in an investor Q&A discussion with 2 weeks left in Q3, AMD management implicitly reaffirmed their Q3 guide, citing weakness in PCs and Gaming offset by strength in Datacenter and Consoles.
AMD should not have been sold because of weakness in other firms, when AMD was beating earnings and raising their guide, and reaffirming their guide, and had credible reasons for doing so.
Growth Prospects for the Next Year
Datacenter
Due to the high margins and AMD’s dominant product offerings, which are only going to get better in the coming months as 4th generation EPYC Genoa is launched, Datacenter is where AMD’s profit growth is most likely to continue to explode upwards. AMD should soon start taking majority market share in the hyperscaler segment, and make serious inroads in enterprise.
While all indications are that the datacenter market remains strong, AMD sales should grow significantly even if the overall datacenter market demand drops off meaningfully, as AMD takes greater market share from Intel.
AMD is also poised to launch CDNA 3, the next generation of their Instinct datacenter GPU product. While their earlier versions of Instinct showed significant progress, and started taking market share, the 3rd generation is described by AMD management as their first leadership product in the space. While I am not modeling significant share gains in the next year, there is a strong chance that AMD does well in this area, and that provides upside to my model.
Xilinx and Pensando are also poised to grow nicely in the datacenter their own right, as well as increase sales of EPYC and Instinct with AMD datacenter bundling. I am not modeling significant gains here, so contributions here would provide upside to my model.
Overall, even with potentially difficult economic conditions developing over the next year, it’s hard to see AMD Datacenter not growing significantly over the next year.
This segment grew 83% in the last year, despite significant supply constraints in EPYC. Management is reporting that they expect supply to catch up soon and that EPYC has now hit an inflection point in growth.
I am modeling 65% growth in this segment, and I view this as conservative. I would not be surprised to see the segment double or better over the next year, which would provide upside to my model. Given the high margins in this segment, much of the revenue growth drops right to the bottom line.
Client
AMD management confirmed at the Q&A recently that they have continued to be supply constrained in the Client segment, even during the recent PC downturn. AMD has had to focus mostly on the high end products and let Intel capture the low end. But management also confirmed that they have secured enough supply to start competing across the stack starting in Q4.
So the residual inventory drawdown occurring in the PC market will limit overall growth in the segment, but this will mostly come out of Intel’s business. AMD sales should continue to grow as they take share aggressively at the high end as well as competing again across the stack. And in 2023, Intel will still not be able to compete with AMD’s performance at the high end.
Even with the inventory reset in progress, this segment should grow 20% over the next year.
Gaming
Gaming is a bit of a question mark for the next year.
On the one hand, AMD is set to introduce their updated RNDA 3 GPUs, which for a variety of reasons (outside the scope of the article) are set up to be cheaper to produce than Nvidia’s new Ada Lovelace cards, and have very strong competitive performance compared to Nvidia’s new cards (with the possible exception of ray tracing). And for a variety of reasons, AMD should take a lot of share in the laptop segment.
On the other hand, since a huge portion of “gaming” GPU demand in recent years has gone to crypto miners, it’s unclear how much demand there actually will be from real gamers. The segment should recede from the mining highs.
In sum, AMD should take a good chunk of market share from Nvidia, from a segment that will be down YoY. As a result, I expect AMD’s Gaming numbers to be roughly flat.
Embedded
I model this segment at 20% growth over the next year. There are some products that could drive growth significantly higher, but I will be conservative and look beyond them.
$10 EPS Rate Likely Next Year
In sum, I model that the growth in these segments should boost quarterly EPS to around $2.50 exiting 2023, a $10 per share run rate, with the growth in EPS coming heavily from the Datacenter segment.
$200+ is Likely as the Market Wakes up
AMD’s earnings grew 67% YoY in Q2, despite the inventory and crypto demand issues that caused Intel and Nvidia’s earnings to fall off a cliff. I am projecting AMD’s earnings to more than double by the end of next year. What’s an appropriate multiple for a tech company growing earnings around 60% to 100% per year?
Often that kind of growth attracts a multiple of over 50. But let’s be more conservative. Management has guided for 20% long term revenue growth, and for higher margins. Current management has a long history of sandbagging guidance to the point that it is a joke.
But we’re being conservative, so let’s assume a 20 multiple. A 20 multiple on $10 earnings run rate would be $200, putting the stock price at triple the current market price. With a history of earnings approximately doubling annually at that point, it’s not hard to imagine the market putting a higher multiple on the stock than 20x.
Setup is Strong
I’ve been following AMD closely since entering at $9 in 2017, and it’s been my largest position most of that time. I have navigated through a lot of volatility in the stock as the market shifted back and forth between people who saw the increasing prospects and people with mostly unfounded fears. Just to name a few of the roller coaster turns, the stock went from $15 to $9, $34 to $17 (2018 crypto dive), $60 to $38 (March 2020 COVID), $100 to $74, and now $164 to $67 over the last 10 months.
Not only is this the steepest drawdown, but I believe it’s the one with the strongest buy at the bottom. In most of the prior dips, AMD’s prospects were still more speculative in the sense that the earnings that would show up would require some time to materialize. During a lot of that time the company was printing only pennies or dimes of profit each quarter, but setup to do a lot more in time.
But today, not only is the company printing profits of over $1/share each quarter, but the earnings have been exploding upwards. People can buy AMD not only on the realization that future profits are coming, but also because of how mind-bogglingly cheap it has gotten today based on current actual earnings.
And the deal gets better by the day. With AMD trading at something like 15x current earnings, management is buying back a lot of very cheap stock each quarter. Probably about 1% of the shares outstanding will have been repurchased just during Q3 from the company’s free cash flow. The longer the share price stays so depressed, the more shares are taken out of circulation, and the faster EPS grows.
What if I’m Wrong?
Even if there is a significant recession, and AMD takes a lot of market share in a declining market, and margins don’t grow, and earnings stay steady, the stock would still be a solid investment.
Expectations are so low, that AMD would have to fail pretty hard to only meet them. In fact, some fundamentals-blind analysts are only forecasting flat returns for the next 18 months. I don’t know how any finance professional can look at the product strength, and how quickly the company is gaining market share, and the guidance for growth from the conservative management outlook, and see flat sales. But analysts like to pick price targets near where the market already is. And it’s impossible to put a price target anywhere near the current market price without assuming some very unlikely scenarios next year.
With expectations so low, there is an enormous amount of upside to the stock, even if my modeling overshot the target. On the flip side, with expectations so low, it’s hard to see many plausible scenarios where the company doesn’t exceed expectations.
If my modeling is close to accurate, AMD is currently trading at a forward P/E lower than Intel’s, even though AMD’s earnings are growing quickly and Intel’s are falling off a cliff. AMD’s gross margins and non-GAAP net income exceeded Intel’s in Q2.
Similarly, NVDA is trading at 63 times Q2 annualized non-GAAP earnings, despite management guiding for Q3 to be significantly down from Q2 results, and with earnings falling off a cliff. AMD’s gross margins and non-GAAP net income exceeded Nvidia’s in Q2. If AMD were valued like NVDA, the stock would need to more than triple based on current earnings, let alone the growth I expect next year.
I view AMD as a rock solid strong buy.
Be the first to comment