Parp’s NVIDIA Short Report Part 2: The clock is Ticking Fast for NVIDIA.

About 4 months ago I wrote and shared my personal short thesis on NVIDIA detailing why I believe its valuation is likely to collapse within the coming year or so. For those who have not read it, I will link it here. https://www.moomoo.com/community/feed/115760633217429?share_code=0yPiM2

In recent months there have been some new developments. I had also learned a couple things that I believe are worth going over here. I have also slightly decreased the time horizon for when I believe NVIDIA’s valuation could potentially drop to just 8 - 16 months as of writing this due to unprecedented developments in the macroeconomic climate pertaining to the value of physical crude oil. I left a positive perspective I believe the bulls might like to hear at the end, so keep reading. Anyway, let’s begin.

* Cancelations, delays, and bottlenecks:

Recent reports suggest roughly half of all US data center builds have been either delayed or cancelled. This is primarily due to bottlenecks such as shortages of power infrastructure, and lacking vital imports from China. This is a reason why I highlighted in the first part of my short thesis that China is ahead in the AI race in some ways. Power hungry NVIDIA chips simply aren’t the way forward unless we drastically upgrade our national power grid immediately. This trend of data center cancellations likely puts a theoretical ceiling on NVIDIA’s valuation as more GPUs likely will not be ordered if their big tech customers know it’s unlikely they can even power them in the near future. This could eventually lead to hyperscalers being left with stranded GPUs that sit on the shelf, just to collect dust and depreciate over time as they cannot be used, nor resold.

* Model Optimization and Algorithmic Efficiency:

If AI is truly as revolutionary and powerful as it is made out to be by big tech and its investors, things like model optimization and algorithmic efficiency should be almost inevitable. What is model optimization and algorithmic efficiency you might ask? In simple terms, over time we should be able to optimize AI models to run faster and analyze more data all while requiring less compute. In fact, AI should in theory be capable of optimizing itself eventually. If AI models can run just as efficiently, if not better, all while requiring less compute. Consequently, that could drastically decrease demand for the hardware to run and train AI models. This could also magnify the already major risk of overbuilt infrastructure and massive overspending. We have already seen a great example of this happen recently with Google’s TurboQuant, which led to a decline in memory stocks closely tied to the AI infrastructure buildout due to it’s ability to reduce LLM memory usage by 6x. So that means in theory it could take as little as one piece of innovative software to potentially collapse NVIDIA’s valuation overnight under the right circumstances. People attempt to counter this point by citing Jevon’s Paradox, however that is more theoretical than a proven outcome for many technologies, it’s very hit or miss. I personally believe it’s far more likely demand for AI compute will eventually level off as time goes on. Jevon’s Paradox also cannot be applied when these companies are generating little to no returns on AI, running into many bottlenecks, and so far mostly just losing money on these data centers. This may surprise some of you but I am actually quite optimistic about AI’s potential over the long run and do believe it is an important technology, though it needs to be much more heavily regulated. However, in the near term I view it as unlikely AI can generate sufficient returns to justify many of the current valuations, hype, and enormous CapEx going towards AI infrastructure, mainly fueled by corporate FOMO.

* An unprecedented energy shock. The potential impact on both data centers, and my time horizon:

As almost everyone is aware by now, we are currently experiencing the largest energy shock in history. The Strait of Hormuz is effectively closed, cutting off around 20% of the global oil supply. Physical crude oil is sitting around $140 a barrel as of writing this. The longer this goes on, the worse the damage will be for the global economy, obviously. Inflation is already ticking back up rapidly and consumer sentiment has just reached an all time record low. Ironically the market pumped on these two reports, which I believe gives even more credence to the bubble hypothesis. The market has so far largely tried to shrug this off, but I don’t believe that can be sustained for much longer. I believe AI companies actually stand to lose the most in the face of this energy shock, especially if it persists much longer. It’s no secret AI data centers consume a massive amount of energy just to operate. Energy makes up roughly 50% of the cost to operate these data centers currently, data centers that still generate little to no returns on AI. Rising energy prices will of course increase these costs drastically if sustained which will turn these data centers into even bigger money pits than before. This is likely terrible news for hardware companies closely tied to AI infrastructure investment as rising energy costs could eventually begin to eat away at hyperscalers initial AI CapEx projections. I believe that the longer this goes on and the higher energy prices rise, the faster this bubble could burst. One extremely interesting data point here is that oil spikes of 30% or more that are sustained for about a few months, or longer, historically has preceded almost all recessions and bubbles popping post WW2. So if history is any indicator here, things are looking somewhat dire as we are now about 6 and a half weeks into this already, and I believe the damage is being drastically underestimated in many ways. Because of these developments I have decided to reduce my initial time horizon of when I believe NVIDIA’s value will fall to just 8 - 16 months. The longer this oil shock drags on, the shorter my time horizon will become.

* Monopolistic behavior, and its detriment on innovation for AI and hardware:

Not too long ago it was reported that NVIDIA paid $150 million to block AMD from getting an Oracle build contract. This is monopolistic behavior and should not be acceptable. Those who are bullish on AI in particular should be calling this out more than anyone. Not only is the market supposed to be free and fair, this behavior artificially hinders competition which consequently negates the incentive to innovate at an accelerated rate. This is especially bad when you consider China is essentially neck and neck with the US in the AI race.

* Data center moratoriums, regulations, and dangers.

The US midterm elections are not too far off, and one of the key issues that will be in focus going into the midterms is AI data centers. AI data centers are wildly unpopular with the American public as they drastically increase both water and electricity bills in surrounding areas. They also create very unpleasant noises and are popping up everywhere and creating a nuisance. AI is still not very well regulated as it is still a new technology and could pose serious threats and problems further down the line as the technology progresses. We’ve already recently heard reports of an emergency meeting between Scott Bessent, Jerome Powell, and top bank CEOs over the supposed dangers Claude Mythos has over the system. Whether how much of it was just a clever marketing tactic or a true threat is up for debate. The point still stands it is wise for the safety of humanity, and the system, that we put a pause on this development and have regulators take a good look at this technology before any potential harm can be done. It’s also only fair that we ensure these companies absorb the costs of rising energy and water bills instead of unfairly passing them off onto people living in the area who never asked for any of this to begin with. (I am not indicating any political leanings here, just my thoughts on this particular issue)

* 5 brief, yet alarming side notes:

1. Currently by some metrics this is one of the most, if not, THE most overvalued market ever recorded. Even an investor like Warren Buffett with a strong buy and hold mentality has been selling off in recent years as most valuations are no longer attractive to him. Recently he has been selling positions like Apple and Amazon, right before stepping down as CEO of Berkshire. Also in a recent interview he stated he is waiting for a much better opportunity, implying he is waiting for a huge dip buying opportunity, and that the recent sell off was “nothing”.

2. The Yen is weakening against the dollar quite drastically and is getting very close to levels the BOJ takes seriously. A further decline in the Yen could force the BOJ to intervene in the currency market. The last time his happened it led to an unwinding of the Yen carry trade, which lead to a global market crash back in August 2024. The Yen carry trade still remains massive and in play and is at serious risk of further unwinding if the BOJ intervenes once again. It’s also an issue that is very underreported on and goes under most people’s radars.

3. Inflation is quickly rising again thanks to the current energy shock and as a result it’s expected consumer spending will eventually slow if this persists. Rate cuts are also highly unlikely to come until inflation is back under control which could take quite some time. This all comes at a time of a weakening labor market, putting the fed in a very tough position, with the risk of stagflation increasing.

4. US household wealth tied to the stock market has reached an all time high since the reporting began back in the 1940s. This is now surpassing even the peak of the dot-com bubble.

5. Hedge funds are now reportedly almost the most bearish they have been in about 5 years, with many of them selling off or going short. This is likely due to the worsening macroeconomic climate and uncertainty surrounding geopolitics, all paired with high valuations.

* Why you should be excited for a downturn:

One final, and optimistic note depending on your perspective. I believe if you are truly bullish on AI companies, a major market downturn could lead to once in a generation buying opportunities. If you are a true long term investor, this is what you should be constantly hoping for daily. This is how investors like Buffett compounded enormous amounts of wealth over time. It has much less to do with intelligence, and more to do with temperament and perspective. (Of course I do also believe Buffett is brilliant) Hopefully some of you find this prospect at least somewhat encouraging and exciting.

Sources will be listed in the comments.

$NVIDIA (NVDA.US)$ $Alphabet-C (GOOG.US)$ $Advanced Micro Devices (AMD.US)$ $Berkshire Hathaway-B (BRK.B.US)$ $Oracle (ORCL.US)$

101K Views