NVDA enjoyed a massive run up in 2021, a run-up which has already been totally erased in 2022 as the five-year performance converges toward the industry average. Unfortunately, there is still room to run here and I believe that NVDA has substantial downside remaining.
Commodity Prices / Macro
I’ve covered this topic extensively in prior articles, but I will rehash my case for a recession here (see A Recession Is Inevitable, Don’t Get Fooled By Bear Market Rallies).
The global economy is about to get hit with a double whammy of contractionary fiscal and monetary policy simultaneously. While rates have already repriced (although it’s not clear this trend is finished), the fiscal deficit is set to fall substantially in 2022 and likely continue to fall further in 2023. Furthermore, balance sheet reduction has just begun in earnest and is on pace for >$1T reduction over the next 10 months. This combined with a yield curve that is deeply inverted is basically a guarantee for a recession in the next 12-18 months (and likely sooner as demand already appears to be falling dramatically in a few key sectors such as housing, commodities, and other durable goods sectors such as autos).
The global semiconductor market is large and growing. The majority of growth is projected to be driven by computing and data storage, wireless communications, and automotive electronics. Unfortunately, none of these industries are recession proof and EV demand in particularly should be impacted in the next 12 months by recessionary headwinds. Computation and data storage demand growth will be driven by applications in AI and cloud computing, which are R&D driven and will also be sensitive to cost cutting measures as margins shrink and companies look to deleverage.
Semiconductor PPI still remains at an elevated level as inflationary pressures persist. Overall, semiconductors prices have trended downward over time as efficiency gains have enabled greater computational power from less resource. I would imagine that as global demand begins to fall, this trend will revert back to normal and the industry will return to a state of deflation.
This has already begun to play out over the past week as SSNLF and AMD reported disappointing results this past week. While TSM posted strong results, it is likely that the drop in demand has not fully materialized for them yet (also benefitting from currency tailwinds). This is on top of increasing restrictions from the US government on sales to Chinese companies. I also believe that PCs will continue to lose market share as well as gaming as smartphone demand will be relatively more resilient, hurting NVDA as they are more focused on these categories.
Zooming in on auto demand – the magnitude of car price increases both for new and used cars in 2021 was historic. Both of the indexes were roughly flat for 25 years from 1995 to 2020 and then increased over 20% (almost 50% for used cars) in 2021 driven by low rates and home nesting / work-from-home trends which encouraged car ownership particularly during the COVID lockdowns. Car sales (like other consumer durables) are very sensitive to rate increases (Powell said as much himself in his congressional testimony in June) and there is likely to be substantial mean reversion in these sales.
Already demand destruction is starting to get priced in with Ford lowering guidance, Carmax posting large drops in new and used car sales in Q2, etc. I believe growth investors underestimate the damage that falling growth expectations can have on valuations. While NVDA has been punished substantially in the past year, I believe a retest of pre-COVID highs is in store.
Zooming in on NVDA
NVDA kicked off their latest upcoming product cycle at GTC which focused on upgraded products for gaming, workstation and data center AI. Gaming run-rate was guided down dramatically to $1.3Bn to $2.5Bn – while management blamed this on efforts to clear inventory of prior generation high-end Ampere GPUs for enthusiasts, it’s clear that gaming demand is falling substantially as the category is highly discretionary and is likely to be impacted early by a broad-based recession. In fact, pretty much all of the categories that management is focused on expanding currently – AI w/Omniverse, Digital Twins, ADAS, and Robotics are dependent on R&D / innovation to drive growth and will likely see substantial demand destruction as R&D budgets are trimmed for 2023.
While I am no expert on the nuances of NVDA’s business in China, I think this headwind will weigh on the stock in the near term. Management claims that they are able to offer sufficient alternative choices within their product offering that are not restricted and that they have the ability to seek a license if it is restricted, but I believe that broader geopolitical concerns and investors seeking to limit China exposure will weigh even if fundamentals are relatively unaffected.
I do think NVDA’s vertical integration offers some competitive advantage. Management views accelerated computing as needing to be vertically integrated as a full stack computing problem to write the operating system or cloud/enterprise distributed operating system, run time engines, libraries application frameworks or develop the storage, networking and cybersecurity. Their view is that none of the customers are just buying the chip, they need the NVDA computing stack to drive speed and efficiency. NVDA has several vertical platforms such as its graphics stack, AI, physics and Ray tracing engines, scientific computing stack, NVIDIA AI, and NVIDIA Omniverse. At the same time, I do not think NVDA’s “flywheel” has fully solidified and they are still subject to competitive pressures in their core businesses. In the coming recession, I don’t think they will be able to gain substantial market share and the market appear to agree as their relative outperformance in the industry is eroding.
Obviously much has changed since the 2008 financial crisis, however it is interesting to note that at the depths of the ‘08 crash, NVDA’s EV/EBITDA multiple fell all the way to <3x. While NVDA’s multiple has fallen over 50 turns from its bubble high in 2021 (from 88x to 27x), it is still relatively in-line with its pre-covid valuation. In my opinion, the market is still far too optimistic on semi-conductor demand where growth is driven by highly cyclical inputs. I believe NVDA’s profitability and multiple have much further to fall.
Looking at the research model, analysts are far too optimistic on growth expectations (and they are roughly in line with management’s expectations). Analysts are projecting steady revenue growth in 2023 and greater margins than 2022, a goldilocks scenario (low prices, strong demand). I believe that both these expectations are flawed.
In 2009, operating profit turned negative and it was not until 2012 that profitability stabilized at pre-recession levels. I believe that a similar period of reduced profitability is likely over the next several years. I believe that as the market becomes more fearful of a recession this will be materialized in the form of further multiple contraction. While the prime time to sell obviously was at the beginning of 2022, we are still not through seeing pain in this space.
The NVDA chart objectively looks horrific and given the magnitude of the run-up in 2021, was not able to complete even a second test of the highs made in Q4.
There’s not another major level of support I can see between pre-covid highs and the current price level. I believe that once this support level is broken (which will coincide with a broader market sell-off), NVDA will likely fall 20%+ from current levels. To reach the pre-covid highs ~$73, NVDA will need to fall 37% further for reference.
On a tactical note, I do think NVDA is short-term oversold and will bounce in the near future. Implied vol spread is well above its 250-day average and I think in general, the market is oversold right now and will bounce somewhat before moving down further.
In conclusion, NVDA has enjoyed a substantial run-up over the last five, capped off by a monumental 2021. While the company is highly profitable (despite some cost headwinds), I believe that revenue growth and profitability are set to fall off substantially (and substantially more than both management and analysts are predicting). This coupled with a multiple that is still high relative to lows in the GFC, I believe that it is highly likely that NVDA retests pre-covid highs.