Look like there is little debate whether NVIDIA (NASDAQ: NVDA) is a good company or not.
Within 12 hours after the article, “Is An NVIDIA Pullback Imminent?” was published on SeekingAlpha.com, close to 6000 page views and 81 comments were posted to discuss NVDA’s fair valuation. Most of the assertions, though, were made on an ad hoc basis. Some may argue that the prices have been driven by emotion, fundamental does not matter. We, the authors, do concede the arbitrary nature of using simple price multiples including the PEG ratio, especially for a tech company with such volatile revenue and earnings.
Therefore in this follow-up, I elect to use three of the more elegant and complex valuation models which call for specific forecasts of forward-looking growth rates in revenue, earnings, and cash flow of NVIDIA.
Franchise Value Model
When top-line growth is the focal point for shareholders, such as NVIDIA’s revenue has grown at high double-digit rates for the last five years, “franchise value” of the firm may be considered. Franchise value is the ability of a firm to replicate a proven successful business model on new lines of business at a higher return to stockholders. In the expression below, the construct of the stock valuation model based on the franchise value, is to add future contribution from the higher ROE (the second term) to the existing stock value from the existing business (the first term).
Justification: The use of the franchise value approach is appropriate for NVIDIA, since the company has shown a convincing track record and the ability to extend the success in Gaming GPUs into Artificial Intelligence, Virtual Reality, and Self-Driving technology. These three segments, contributed close to 40% of the current revenue, are on a path of explosive growth on the industry level, as NVIDIA is one of the only firms that are in the forefront of the four leading growth areas for next decade.
The key determinants of this approach are the estimate of the return on equity (ROE) of the new business. NVIDIA’s ROE on the existing business is 25%. With the increasing capital expenditure xx planned for 2018, we assume that the ROE on new business will surpass the current 25% on the existing GPUs between 25% and 30%. Given the current volatility in NVDA of a beta near 1.75, a 15-20% discount rate is reasonably assumed. The book value growth rate is in a range 14-18%, per growth rates in last few years. Standard Monte Carlo simulations have been performed, on 100,000 permutations, and produced a median estimate of NVDA fair value at $128 within the range of $109 and $147 for about 85% probability.
Growth Duration Model
Justification: As investors in the Tech industry love to use PE ratios to value tech stocks, however, comparing the absolute level of PE or PEG ratios seem meaningless. Tech companies, by design, are heterogeneous. Therefore, we elect to use a relative valuation approach, growth duration model, with the assumption that the relative valuation of two entities is determined by their relative growth rates. So the relative difference in PEs should reflect the earnings growth difference which prevails in the next n period, as shown in Equation x below. As AMD is often compared with NVDA, their relative market share in the discrete GPU space is often considered a zero-sum gain.
The street has estimated that NVDA’s earnings will grow around 32% and AMD 30% in 2018. As AMD is still at a loss, a negative PE will not make sense. Yet, the street predicts that AMD will turn profitable in 2017-2018. Thus, the AMD earnings estimate gives a forward PE of 50, compared with a forward PE of 40 for NVDA (Source: Yahoo Finance).
The most important determinant of this model, also the most controversial one, is the future earnings growth rate of the NVIDIA. The historical annual earnings growth rate ranges from the low of 23% for the last 5 years’ CAGR to the high of 88% for the last 1 year’s CAGR. Clearly, no one will realistically expect that the 2017 earnings growth rate will repeat the 140% of Q4 2016. Given that the analyst estimates, 20-35% for 2017, have been historically accurate and conservative, it seems sensible to assume that the 2017 earnings may be in the range between 18% and 25% (see Figure above).
In addition, we pick the fight we can win by NOT picking on AMD’s estimates. It seems that the street is ready to give AMD’s Vega the benefit of the doubt so it is expected to gain some GPU market share back from NVIDIA’s Volta launch, albeit not significant. Thus, analysts’ estimate 30% of AMD’s 2017 earnings growth rate will stand. Again, Monte Carlo simulations have performed and produced a fair value distribution suggesting that NVDA has a fair value around $110 with an 85% probability falling in the range between $95 and $125.
Two-Stage Free Cash Flow Model
Traditionally, the street has been overly fascinated by the quarterly earnings growth rates. However, when earnings volatility becomes so extreme, say to triple digits, even the most die-hard fans of the company should realize it is not sustainable. When the most recent earnings growths are not expected to continue, shareholders should always fall back to the basics, i.e., future cash flow of the company.
Justification: For a company like NVIDIA, which has had a significant capital expenditure and more in the pipeline to invest in three of tomorrow’s high growth areas, Artificial Intelligence (AI), Virtual Reality (VR), and Self-Driving technology, the long-term outlook on stock values may be looked upon based on the free cash flow (cash flow net capital expenditure) to the equity holders. To them, the stock valuation is “the discounted value of future free cash flow.” As NVIDIA is investing in the next-generation segments that will produce exponentially high future cash flow, it is reasonable to assume, at a minimum, a two-stage free cash flow growth pattern. In the model below, the premise is to discount next two years’ free cash flow at one first-stage growth rate and the rest of the future cash flow at a second-stage, constant growth rate perpetually.
For the last 2 years, NVIDIA has had an annual 15-20% free cash flow growth rates. The street has estimated around 40% growth for Q1 and Q2 of 2017, compared Q1 and Q2 of 2016. We use the standard investment rationale that the initial capital expenditure should generate a lower free cash flow in the R&D first stage but a higher rate in the perpetual growth second stage. The higher growth rate in the second stage, just 2% higher than the first stage, is based on a conservative stance since the CapEx invested should produce an explosive growth from the emerging next-generation technology in next generations (see Figures above).
Therefore, NVIDIA’s first two years’ free cash flow is assumed to be around 10-15%, considerably lower than both the historical and analysts’ forecasts of 40%. The perpetual cash flow after is estimated around 12-17%, per long-term industry average. With these inputs, simulations have been performed and generate a range of fair value estimates between $90 and $111 with a median estimate of $101.
In short, the three distinctively different valuation approaches and the street analysts produced the following fair value estimates for NVDIA:
If you don’t agree with my input of the estimates, feel free to plug in yours in any of the models to come up your own fair values.
NVDA investors! Pick your poison!