Falling PC stocks
We've have recently bought into personal computer chipmaker stocks because they appeared cheap. But they kept going down so we are at a loss. This called into question our PC bias and contrarian position. With slightly growing long-term PC demand, why is there no bottom? This trend can be seen especially for companies like Intel (INTC), AMD (AMD) and even Microsoft (MSFT). Can ARM instruction (ARMH) set tablets from Apple (AAPL) and others supply future demand and potentially lock out the old "Wintel" (Microsoft and Intel) players? Other articles handle these questions, but we will be looking at it from a risk management perspective. The chart below demonstrates that if you bought three months ago, say due to a Seeking Alpha article, you could have a paper loss of 46% if you invested in AMD, with no risk management. Or if you invested in Intel, you could have a paper loss of >16% while a random pick of stocks from the S&P would have given us a 9% return. The question is how low can they go? And should we cut our losses given the returns are below market average?
The Quantitative Growth Simulation
Simulations can help answer those questions, particularly when combined with investor acumen, it can help us manage risks. Using assumptions about growth, we created a spreadsheet model: a discounted cash-flow, with valuation using dividend-growth models. With a projected dividend, based on actual dividend for Intel and hypothetical dividend for AMD based on profit per share, we were able to create a chart (see below) of share value versus potential revenue growth rate (measured in fractions of 1, so 0.04 = 4%).
This simple model is known as the Gordon growth model. The range simulated is based on IDC growth projections. Return on equity for each company is measured using standard Capital Asset Pricing Model, known as CAPM. The related Weighted Average Cost of Capital, or WACC, is used for discounting. WACC is the minimum return that a company must earn on an existing asset to satisfy its creditors, owners, and other providers of capital. WACC includes the cost of debt and can be lower than the CAPM rate of return due to tax benefits of debt. For some companies like AMD, WACC is lowered through high debt. According to my calculations, AMD WACC falls around 6.06%, while Intel is at 4.05% based on assuming an expected average sector return at 5%. Other ways of calculating (e.g. thatswacc.com) use different assumptions resulting in different absolute values. However, they show the same 50% higher WACC for AMD when compared to Intel. In essence, AMD investors expect more return than Intel investors. Thus, they could leave in droves if AMD is expected not to provide that return. At the same time, investors could earn more if they hold AMD in the long run and it continues to thrive.
For our model, it was important to factor in revenue size as a hindrance to growth; the larger you are, a smaller portion, as a percentage of revenue is sustainable. Thus, we added growth caps to the chart. Take a close look at the forecast chart. It tells the upside and downside risk story by itself. (see above)
Intel can go down $7 more? AMD another $1.40?
According to this model, with a drop in annual demand, AMD is hit the hardest with a projected value as low as $1.85. If AMD manages to capture growth past the 6% annual rate, it also has the higher upside potential of more than four times higher (up to $20). The downside for Intel would be around $15, a drop of $6 to $7 from today's price, if it sustains a 3% annual drop in revenue. The upside being more than double today's price. Do not think this validates the scenario where ARM will take over Intel x86. It just shows that if Intel and AMD would lose 5 to 10% of the PC CPU market total (thus sustain a -3% revenue growth when the market is growing) per year, they can start to get really cheap. That scenario can only roll out if Windows 8 is a total disaster. Unlikely at this point, but we have to wait and see.
Rule of thumb - risk management
Let me start with a rule of thumb for stop-loss risk management when value investing on a falling share. Based on experience, it is suggested that you use a 30% stop loss rule for medium to long-term investments. Why? Because calling the bottom is difficult, but calling it close to the 30% range is easier. When prices drop in volatile markets, it's possible that the stock will drop 10% but then rebound almost immediately. Yet, once an item drops below 30% of the "value" price, the chances of a quick rebound are slim. The value price can be calculated based on chart, momentum or quantitative fundamentals; this is largely dependent on the individual investor. You could average down but then stop if it continues to fall past the 30% mark. With our predictions for Intel, it is unlikely that it will fall another 30%; Intel would need to fall $6.5, nearly as much as our worst case projection. Thus Intel is a hold if you went long recently. At today's price, AMD needs to fall just $1 to satisfy hitting our stop. For example, if you bought (or were long) Intel at $22.65 based on a 4% dividend yield, it is a hold. Alternatively, if you were long and saw value at $31, we would rate it a sell and not an average down.
Looking at AMD, it has performed even worse, falling 10% in the last 2 weeks. Based on the above rule, we should be selling and closing before the 18 October 2012 results relative to our position at ($4.10). However, there is a high chance of a rebound when short interest is large and with a relatively good uptake of the AMD promotions. According to most finance sites, more than 20% of the AMD stock is on loan, still shorted. Thus implying speculators are more active than investors. Shorters are not investing; rather, they are borrowing other people's stocks, selling them in hopes of buying them back with cheaper stock. That gives the stock price a strong potential rebound. Unfortunately, in the short-term, we can see them trading at a lower rate like Hewlett-Packard (HPQ), which broke 9-year-lows recently. Let us look at AMD, in a little more depth.
Revisiting AMD - Alternative Growth Avenues
As an investor and a software developer, I also have faith in AMD for the long-term despite our dire short-term predictions. This is because of AMD's Fusion APUs and GPGPUs, as well as its lead over Intel in discrete graphics. Furthermore, the AMD lead, Heterogeneous System Architecture (or HSA) Foundation, is gathering momentum. AMD is a leader in discrete graphics, such as the Radeon 7970 on a FX-8150 8 core desktop we use. The performance of this platform is impressive compared to Intel. For instance, a time consuming Monte Carlo simulation for a bank using the above setup produced a 20x higher performance when compared to the top end Intel Xeon server with latest AVX data parallelism. This performance was achieved using OpenCL and C++ AMP, languages which form the core of the technologies being worked on by the HSA foundation. Similar or higher performance can be achieved with Nvidia (NVDA) pro cards (e.g. Quadro), but they are not being integrated on the CPU. The OpenCL friendly Radeon 7970 architecture, known as Tahiti, will be integrated in the next APU core that is due in 2013. That will be a data center competitive advantage.
Don't just take our word for it, though. Industry heavyweights like QualComm (QCOM) also believe that the future is widespread use of integrated general purpose graphics processing. Last week, QualComm joined as lead member of the HSA foundation, adding a significant amount of funding. They joined to provide software developers, like us, better and more standardized tools for their future SOC mobile chips. The tools and standards that are being created, will help all CPU makers. A likely exception will be Nvidia, which has its own tools. This can improve performance per watt for data centers. For AMD, this can result in capturing high performance computing market share from Nvidia. A great way of realizing the upside growth scenario, which we modeled above. AMD has not yet benefited from this growth, since Nvidia has a lead with the proprietary Cuda language for GPGPU and HPC solutions.
Last but not least, this scenario is not a long way off. C++ AMP, is already supported by Microsoft Visual Studio 2012. It was first supported on the hardware side by AMD, in collaboration with Microsoft. You can find additional details on these topics by searching Google (GOOG). In any case, do not be surprised that some data centers could start moving to APUs to get better performance for numerical tasks in the mid-term future. In this regard, Intel won't stay still as they support OpenCL already. However, their GPGPU performance significantly trails AMD; they have less to gain from HSA foundation initiatives.
So what does all this mean?
In the short-term, my suggestion is to sell - we probably have not reached the lowest low yet. In the mid-term, we would hold. And for the long-term, buy. Meanwhile, Intel at least pays dividends, paying for the "bread on the table" so to speak, as we await their mobile growth results.
Personally, I do not expect to be profitable in the short-term, investing in these companies that provide discretionary goods. Energy commodities will be where the short- term profits are. If you had a choice would you put off refueling your car, or delay upgrading your laptop? Energy tends not to be discretionary. Nonetheless, Intel and AMD have far more plus sides than down in the long-term, albeit in different ways.