Description

Nvidia Corporation is a liquid, large-cap technology company focused on the visual computing market, primarily through its GPU products. The company has been rocked by the overall slowdown in PCs, and particularly by huge declines in associated OEM GPU shipments. Nvidia now trades at 10% EBIT/TEV yield, which makes it cheaper than approximately 90% of our investable market (which we view as firms with > $2 billion market cap).

In order to understand Nvidia, you need to first understand its core product – the GPU. We’ll provide a little history that should help orient VIC readers.

Back in the 90s, the increasing processing requirements of graphics hardware, and new applications, such as 3D graphics and spreadsheets, began to bog down the CPU. What was needed was a high performance chip that could handle these new intensive processing loads, and free up the CPU.

Although there were intermediate technological steps, in 1999, Nvidia created the first Graphics Processing Unit (“GPU”), which was a chip designed to manage graphical output from a computer, rendering images and animation, but which could also be used for other applications requiring high performance.

Discrete GPUs

At this point, PCs typically had two separate systems: a CPU and then a separate graphics card (or “add-in board,” with a GPU chip on it) to handle graphics. Since graphics displays were not part of the motherboard, early GPUs were so-called “discrete” GPUs, meaning they acted as a separate processor for graphics. Unlike CPUs, which were optimized for sequential processing, GPUs used parallel architecture, which could handle many tasks simultaneously. Additionally, GPUs included dedicated graphics memory, with dedicated RAM for the card to use, but which was not shared with the system.

And this technology dichotomy persists today, with CPUs and GPUs both included on an OEM basis on shipped PCS, or a consumer might buy a separate video card if he wants extra performance. GPUs are still regarded by many (especially gamers and other high-end users) as a kind of high performance option.

Additionally, while GPUs tend to offer higher resolutions, and have higher processing speed, they are energy hogs, and tend to generate a lot of heat, which makes them hard to integrate with mobile devices. Thus, discrete GPUs are more commonly found in desktops (and notebooks), and it is the discrete GPU, which is essentially a high-end video card deployed in a PC, which represents Nvidia’s core business.

Before moving on, a quick note on how we handle dates. Nvidia’s 2015 fiscal year ended on January 25, 2015, which can make for some difficult mental gymnastics when skipping back and forth between fiscal and calendar, so for simplicity in the body of this report we will refer only to calendar years. Thus, a calendar 2014 reference would refer generally to Nvidia’s Fiscal 2015, while calendar Q214 would refer to FYQ215, since these are roughly equivalent. It just makes life a little easier.

The Bear Case for Nvidia

The main concern people have about Nvidia is that, as discussed above, its GPU technology is found mostly in PCs, and it is thus heavily dependent on the PC industry, and as pretty much everyone at this point knows, PCs are in terminal decline. Gartner and IDC recently released alarming worldwide PC shipment figures for Q215, which showed the steepest declines since 2013. According to Gartner, vendor shipments for Q215 were 68.4mm, versus 75.6mm for Q214, for a 9.5% YoY decline rate. The IDC figures (which do not include tablets) were even worse, at 11.8%.

The news is even more grave for the GPU market, where Q1 declines were the worst since the financial crisis in 2008. According to Mercury Research, the overall PC graphics market declined by 15.8% in Q115 versus the quarter a year earlier. Ouch.

Consistent with these terrible dynamics, in May, Nvidia announced an earnings miss -- with Q115 EPS (again, this refers to fiscal Q116) flat from a year earlier at $0.24 -- which was their first quarterly miss in 5 years. The culprit? The company’s OEM and IP platform, which made up almost a third of Nvidia’s revenues in 2014. Quarterly OEM and IP platform revenues were down a stunning 38% YoY. The company also guided revenues lower for Q215, below expectations. Complicating the picture, the company also announced a winding-down of its Icera Modem operations, which means the company will no longer compete for the broader mobile computing market, and will also result in a $100-$125 million restructuring charge. The company is also embroiled in an expensive IP law suit with Qualcomm and Samsung, and has spent upwards of $70 million on it thus far. In short, there’s a lot to complain about here.

None of this recent history should come as much of a surprise to seasoned tech investors. The term “creative destruction” applies especially well to the brutally competitive market for semiconductors, which is defined by rapid innovation, and the relentless destruction of previously dominant technologies.

So the bear case seems clear. This company is a mess, with its core GPU business getting eaten alive by declining PC sales, as it gets shut out of the broader mobile market, fights IP battles, and tries to compete against some of the largest, best-capitalized technology companies on the planet, including Intel, AMD, and Qualcomm.

However, all is not lost. As is often the case in technology, here we have a case here where a company has a core legacy business that is shrinking, but has new growth areas to which it is attempting to pivot, in an effort to reinvent itself. The question is: can this new growth offset the effects of the shrinking legacy business?

You can make a good case that value investing is often about front-running investor expectations. And when expectations are very low, a company doesn’t necessarily have to show dramatic success, but merely has to exceed these diminished expectations, which can result in the market being surprised.

We argue this is could be what happens for Nvidia, since there are reasons to believe all is not as bad as it seems in Nvidia’s OEM and IP business, due to unique market conditions, and it also has some spectacular growth prospects embedded in several lines of its business, notably gaming.

OEM and IP Business

First, let’s review Nvidia’s exposure to PCs, and put recent results into context.

Nvidia’s OEM business relates to GPUs that are included with PCs that are not “gaming” PCs or other PCs with highly specialized uses. These are low margin commodity-like GPUs, where Nvidia has no pricing power. This is the company’s pain point.

Intel is the market leader in integrated GPUs and other chips with graphics, since it offers its own proprietary GPUs. “Integrated” generally means the processor is integrated with the CPU package, and shares RAM with the system. It’s tough to compete with Intel, although this is not really the market share fight that Nvidia is waging, since Intel is focused on these generic, lower performance integrated GPUs for the mass market. Remember, Nvidia is more focused on discrete GPUs, which are much higher performance GPUs. Still, we need to consider this pain point.

We think this particular juncture in the PC markets offers an overly bearish view on what’s happening within Nvidia. One important fact to consider when evaluating the overall PC market is the upcoming release of Windows 10 in Q315. This had led to a lull in the refresh cycle, which has had no recent catalysts to drive PC sales since Microsoft ended support for Windows XP in early 2014. IDC expects PC sales to recover in 2016, and grow slightly through 2019. Additionally, an appreciating US dollar has caused a significant recent currency impact, especially with regard to Europe. So you can make a decent case we are in a PC trough, which is creating terrible short-term optics for the business. “Availability bias” is a well-known behavioral bias afflicting investors, who focus on items that trigger emotions and are easily recalled. Perhaps the dire news coming from the PC world is creating such an availability bias for investors.

Change in Business Mix

Also worth noting is that Nvidia’s business mix is undergoing a fairly dramatic change. Consider the contribution of Nvidia’s OEM and IP business to overall revenues, which had declined from 48% of the business in 2012, to 22% for Q414. That’s a pretty big shift underway. Also, Nvidia’s IP revenue, which derives from a licensing deal with Intel, today accounts for on the order of 5% of overall revenues, meaning that by now, with Q2 2015 unfolding, the total contribution of OEM is a fairly small portion of overall revenues. So sure, you can take another whack out of OEM, but that’s becoming easier to overcome, especially with the growth drivers discussed below. And yes, it is true that quarterly OEM and IP sales declined by 38% YoY for Q414/Q415, and this is a scary number, but we argue this is largely due to the unique PC market conditions described above, and is further mitigated by OEM’s reduced importance to the revenue line. We think it’s increasingly difficult to argue that OEM weakness will somehow dominate the firm’s overall revenue growth rates going forward, and we also believe that future OEM weakness may not be as bad as many expect.

Gaming

In contrast to Nvidia’s OEM woes, the company’s gaming platform is growing at explosive rates, with 36% YoY revenue growth from 2013 to 2014, and 25% growth for Q115 versus Q114. And with $2bn of revenues for 2014, or approximately 44% of Nvidia’s total revenues, it’s a huge portion of the business. So is it unreasonable to think gaming alone could more than offset future OEM declines? We don’t think so. Furthermore, this is a profitable platform. Gaming margins are strong, with gross margins approximately equal to the company’s average gross margin of 56%.

The company estimates there are 1.7 billion gamers worldwide, and in 2015, global household spending on gaming is projected at > $70 billion. This is a significant number, more than triple global music industry, at $20 bn, and it even dwarfs global box office revenues of ~$40 bn. Nvidia is well-positioned to participate in this huge market.

PWC estimates PC gaming will grow at an approximately 5.7% CAGR through 2019. Freemiums, social gaming, and hypergrowth areas such as eSports (competitive/professional gaming) or the upcoming Oculus Rift and other virtual reality offerings, will contribute to this trend, as will Steam, currently the dominant internet gaming platform, which has revolutionized the distribution of games. At the end of 2014, Steam had ~125 million active users, up from ~50 million in 2012. Screen Digest estimated that 75% of games bought online for the PC were through Steam. That’s a nice distribution channel for the company. And this is a young and passionate consumer, for whom gaming is an important part of how they spend their free time. We like this consumer dynamic.

Indeed, the growth in gaming is in turn driving growth in the hardware required to play games. Jon Peddie research estimates that PC Gaming Hardware, with sales of $21.6 bn in 2014, will grow at 0.6% in 2015, with growth accelerating in 2016 and 2017, at 3.7% and 2.7% rates, respectively.

While PCs today are the dominant platform, and Consoles are not far behind, the real growth story for games is in mobile (phones/tablets), for which gaming revenues are projected to overtake those for consoles within the next few years. According to Newzoo, mobile games revenue is expected to double from $17.5 bn in 2013 to $35.4 bn by 2017.

GeForce GPUs

Building on its historical reputation for PC graphics cards, Nvidia has positioned itself as the dominant PC gaming platform, with its GeForce GPUs. The microarchitecture goes by the codename, “Maxwell.”

To date, Nvidia has shipped 30 million GeForce 8 series cards, and estimates there are 180 million active GeForce users today, which is a larger installed base than PS3 or Xbox 360. That's interesting. The company also recently launched the Titan X, which is the fastest single-GPU graphics card in the world. Nvidia also offers GeForce Experience, which is a driver updater and allows easy customization and optimization of graphics settings, and brings aspects of console play to the PC. The company claims there are now more than 50 million PCs with GeForce Experience.

What this may mean for the future is that if you are a hard-core gamer, and you want to play on high-powered desktop, you will have an Intel CPU and an Nvidia video card. Period. Full stop.

In Q314, Nvidia launched the GeForce GTX 970 and GTX 980 graphics cards. These have been enthusiastically received within the gaming community, to say the least, and have allowed the company to increase its market share.

The Nvidia / AMD Discrete GPU Horse Race

Now that we’ve covered the OEM market (non-gaming) GPU and GeForce (gaming) desktop GPU, markets, it’s time to discuss market share. The discrete GPU market consists of two players: Nvidia and AMD. That's really it. Ten years ago these two firms shared the market roughly equally, but that has changed over time, and has changed very dramatically recently.

Nvidia has simply destroyed AMD in recent quarters. Back in Q413, Nvidia has a 65% share, while AMD had a 35% share, but just one year later, by Q414, Nvidia’s advantage had grown to 76%, versus AMD’s 24%. Part of the problem for AMD is that it is fighting a war on two fronts: it is competing not only in GPUs, but also in CPUs, where is must contend with Intel. And today, Nvidia spends more on R&D than AMD does across its entire business. Meanwhile Intel spends approximately 10 times what AMD does on R&D. Can AMD bounce back and regain market share while battling on two fronts? It’s unclear, but I wouldn’t want to place a big bet on it.

In the meantime, Nvidia may benefit from near-monopoly positioning in the discrete GPU market, where it can exploit economies of scale, exert control over prices, enjoy abnormal profits, and invest more in R&D to maintain its dominant position.

Fabless Manufacturing

Like Intel, AMD is a semiconductor foundry. Foundries manufacture the wafers used for processors, and are often located in China and Taiwan due to low labor costs. Foundries are hugely capital intensive.

Nvidia separates the design of chips from their manufacture. Unlike AMD, Nvidia does not own its own chip fabrication processes, but instead outsources the fabrication of its chips to a foundry, the Taiwan Semiconductor Manufacturing Company Limited. In general, fabless developers can benefit from lower capital costs and concentrate on R&D. Interestingly, Nvidia recently entered a new supply agreement with Samsung, which will begin producing chips for Nvidia this year. Although not much is known about this new supply arrangement, it is rumored that Samsung may be offering wafers at as much as a 10% discount to the previous supply agreement. That could add a nice bump to margins over time.

Mobile

Mobile refers to devices you can carry (mostly smartphones), but also includes notebooks. A powerful development in mobile computing has been System on a Chip (SoC) technology.

Traditional CPUs execute instructions and perform basic operations, but the PC requires much more to function, including chips for memory, graphics, connectivity, and so forth.

Since mobile computing has restraints related to heat, and on power, in terms of battery life, it can’t really use a traditional CPU. Mobile devices must weigh performance tradeoffs against these system constraints, and a CPU is too large, hot and power hungry.

Enter the SoC. The SoC is a kind of highly functional, multi-purpose, low power CPU/GPU that addresses these issues for mobile. An SoC might include a CPU, GPU, RAM, a USB controller, power management, wireless functions, and more, all integrated. Essentially it is an entire PC on a single chip. You can argue that SoCs are in some ways the next step beyond the CPU. Regardless, today SoCs are the dominant technology deployed in mobile. And Nvidia today has a state-of-the-art SoC offering.

Tegra X1

The Tegra X1 is a “mobile super chip,” and includes a CPU, GPU, memory controller, and image signal processor. It is the most advanced SoC in the market for mobile applications, offering desktop-type performance in a mobile format. It is the first mobile chip to achieve a teraflop of computing power, which is equivalent to what the most advanced supercomputers were able to achieve in 2000. If you want to run windows or high-powered applications, you use a PC. But if you want a handheld for gaming, social media, youtube, movies, etc. you should use Tegra. Tegra is so important to Nvidia that the company has been breaking it out separately in the financials.

“Great!” you say. “So this means Nvidia should dominate the smartphone market!” Would that it were so. There are two problems with this.

First, the Tegra is so powerful that if offers in fact too much power – more than is required for today’s casual smartphone or tablet user. Second, Nvidia can’t compete with Qualcomm in modems.

Modems, Qualcomm, and the Mainstream Mobile Market

While SoCs are an important technology in mobile, another important aspect of mobile is cellular baseband modem, which allows mobile devices to connect to the internet. Qualcomm dominates this business, which has contributed to its lead in SoC, which is integrated with its modem.

Importantly, Nvidia threw in the towel on the modem market this year, when it decided to wind down its Icera Modem operation. In the future Nvidia will partner with a baseband 3^rd party supplier for modem capabilities. This basically hamstrings the company’s ability to compete in the generic Android smartphone business.

What the Icera wind-down appears to mean is that generic phones and tablets are not going to be a core focus for the company. That’s a tough pill to swallow, but it’s just too hard to compete with Qualcomm’s advantage in modems. And again, the Tegra has more power than the traditional mobile user really needs. This is a big deal since it signals the company will abandon the mass market, and focus instead on niche products. But that’s not all bad, since there are numerous other areas requiring high performance where they can compete, and they may have a lot of success with Tegra in mobile gaming.

Shield

For instance, Shield uses a Tegra chip and sits at the intersection of the PC and mobile. It is a device used as an Android set top box to connect with your TV, enabling PC gaming, and also includes a gaming console. Nvidia also offers Shield gaming handhelds and tablets.

High Performance Computing (HPC) and Cloud

Because GPUs have become so powerful, they are increasingly being used for processing that has traditionally been done by the CPU. So-called “GPU-accelerated computing,” can be used to accelerate calculation-heavy processes, such as complex simulations, by reducing load on the CPU. The GPU just makes computers run faster.

This has many applications in science, analytics, and engineering, including “deep learning” areas like facial recognition, speech processing, medical imaging, and seismic image interpretation. Accelerated computing also can reduce power loads at data centers. On the Q1 earnings call, Nvidia said that their accelerated data centers business was growing at over 50%. Nvidia’s Tesla is a market leader in accelerated computing, and HPC and Cloud made up 6% of Nvidia’s 2014 revenues, which were up 53% from 2013, and 57% YoY in the first quarter of 2015.

Auto

In 2012, Tesla Motors began shipping the Tegra, as part of its electric sedan, as a mobile cockpit computer, in order to provide beautiful high resolution graphics on the dashboard. The chip provides visual computing, an enhanced touchscreen, and an infotainment system. Nvidia has also been working with Audi and Bentley, and the Tegra is also being used to assist autos in “seeing” the environment around them. At ~4% of revenues in 2014, Auto is a small piece of Nvidia’s business, but it grew 85% from 2013. The first quarter of 2015 was a blockbuster, with revenues up 121% versus the first quarter in 2014.

Enterprise

Enterprise addresses the high performance computing market, and relates primarily to a couple of graphics platforms.

The first is Quadro for visualization. Quadro is used in workstations, and is for accelerating CAD processes and other design applications.

The second is GRID, which accelerate virtual desktops and applications. User can access critical applications remotely, including those requiring a lot of processing, such as 3D or multimedia. GRID customer count increased YoY in the first quarter of 2015 by about 8 times, to 250 customers.

Enterprise represented 18% of Nvidia revenues in 2014, and margins are good, but growth has been unexciting, at 4% for 2014, and due to currency headwinds Q115 Enterprise was actually down 4% from the quarter a year earlier.

Patents

NVidia has over 7,300 patents, including 600 fundamental graphics patents, which is the richest portfolio of graphics patents in the world. No individual competitor, including Intel, SMD, Samsung, Pixar or Apple, has over 200 fundamental graphics patents. Nvidia is simply the world leader in graphic IP.

IP Law Suit

One wildcard in the mix is that Nvidia filed in Q414 the first IP law suit in its 22-year history, claiming infringement on 7 of its GPU patents, by Qualcomm and Samsung. In April, the International Trade Commission held a Markman hearing, which is a pretrial hearing in which a judge determines facts about the claims. The hearing went well for Nvidia, as 6 of the 7 patents at issue were given favorable claim construction. Nvidia has also asked for an “exclusion order,” which would essentially ban many Samsung devices from importation and sale. Samsung has something on the order of a 30% share of the U.S. mobile user base, so you can be sure this has their attention. Perhaps this will motivate them to settle via a licensing arrangement. Samsung countersued, and Nvidia has since asserted four new patents in a follow-on countersuit. And so the food fight begins.

So that’s an overview of the business, and below we include some financial highlights.

Strong and Growing Margins

Nvidia has exhibited strong margin growth over time. Back in 2007, the company’s gross margin was 45%, and today it stands at 56%, for a geometric average growth rate of 3% for the past 8 years. That’s better margin growth than 85% of the stocks we analyzed in our large and mid cap universe (> ~$2 bn market cap). As the OEM business shrinks away and creates less of a drag on gross margins, we think Nvidia has room to continue to grow margins since growth is coming from Enterprise, HPC & Cloud and Auto, which all feature higher margin products. Operating margins were also strong at 16.2%, placing the company in approximately the top third of our investable universe. We think this long-run margin growth demonstrates the company has had a sustainable competitive advantage in its markets, and one that will persist.

Return on Capital and Assets

Over the past 8 years, the company has earned a geometric average ROC and ROA of of 8.5%, and 6.6%, respectively. While these are not spectacular returns, they are above average, which demonstrates Nvidia’s ability to generate a return on the capital it has invested and its ability to utilize its productive assets effectively. The consistency of these results demonstrates the business has been a good investor over a long time frame.

Free Cash Flow

Nvidia has generated a cumulative total of $4.6 billion in Free Cash Flow over the past 8 years, which equates to approximately two thirds of its asset value of $7.2 billion. We believe cumulative free cash flows, when scaled by assets, are an especially useful metric for assessing the quality of a business. The company has been a steady generator of free cash flow for many years. The company’s FCF generation, on this basis, is better than 78% of the companies in our investable universe.

Financial Strength

The company is generating positive net income and free cash flow, and free cash flow > net income, which indicates it is not using accruals, which would be a red flag for us. The company is also paying down debt, which increases its overall financial flexibility, and the company’s current ratio is increasing, indicating greater liquidity and ability to meet short term cash needs.

The company is also a net repurchase of equity in the past year, and reduced the share count by 12% during 2013 and 2014. In the Q1 call, Nvidia announced its intention to increase capital return (repurchases and dividends) to shareholders in 2015 to $800 million, up from $600 in 2014. The Board also increased authorized share repurchases through 2018 of $2 billion. We think this is a shareholder friendly firm.

Positive Operational Momentum

Nvidia increased its return on assets, as well as its FCF returns on assets over the past 12 months. Positive trends in the underlying business appear to be in place. It also increased its asset turnover rate, indicating a more efficient use of assets. Gross margins also increased for the sixth year in a row. The firm does not appear to be greatly at risk in the event of some unanticipated adverse financial shock.

Cheap

The company also trades at a 10% EBIT/TEV yield, which places the company in the cheapest 10% of the overall market of all stocks over ~$2 billion. The company is priced for a poor outcome, which may be unjustified in light of the above factors.

Summary

We think if you believe in the emerging possibilities in visual computing, if you think the concerns around the OEM business are overblown, and if you believe the underlying financial dynamics for Nvidia demonstrate its ability to continue to chart a leadership path in this market, then the stock looks like a reasonable bet at this price.

Catalyst

- Release of Windows 10 offsets OEM losses

- Gaming growth trends remain in place

- Continued traction for Tegra X1

- Continued wins in HPC & Cloud, and Auto

- Negotiated IP license to settle law suit

- Supply of cheaper chips under new fab supply agreement