(not financial advice)
I believe we might be approaching a black swan event.
Nvidiaβs will likely break $1,000 soon which begs for a stock split.
After that NVIDIA will cross $3T in market cap and will become the biggest company in the world.
The latest was in 2021, when they did a 4-for-1 split, the day before it traded for around $744 β cheaper than it is today.
TL;DR: I predict NVIDIA will be the #1 by market cap this year.
A stock split does nothing to the company's market capitalization. This figure remains the same. The thing is β people often act irrationally after a stock split and it is possible that the price will skyrocket after a split.
Now there are 2 things important things to discuss:
Nvidia is announcing amazing things happening for them left and right. Literally every day. And they are, basically, monopolizing the AI market that will be worth trillions. If Nvidia manages to keep their monopoly until we have common humanoid robots, I wouldnβt be surprised if itβs the first company to cross 10T market cap. Take a look at this: The race for AI robots just got real
Thereβs a lot of doomers saying that Nvidia is doomed because their GPUs are thermally inefficient. Thereβs a ton of startups trying to design GPUs that work specifically with AI or Large Language Models. They are trying to get a chunk of the cake for years, with no success. These companies are essentially vaporware. Bait for investors. Money is money. Examples: https://www.graphcore.ai/products/ipu https://wow.groq.com/lpu-inference-engine/
I went out and researched the stuff in-depth.
Turns out that, at datacenter scale, Nvidiaβs GPUs are incredibly thermally efficient and designing new architectures doesnβt make sense from any point of view (except the founders cashing in on investorβs money). There could be some improvement in hardware, but Nvidia is also making improvements with every release.
Emad Mostaque (Stability AI): βPower efficiency for Nvidia products for hyper scaling is very good."
And Nvidia has a huge advantage over everyone else β all the software is designed to support their architecture (mainly CUDA). And this isnβt something that a team of dozen senior software engineers can fix. Weβre talking YEARS of development of different software libraries for different programming languages. Done by thousands of people.
Itβs a complete monopoly.
Nvidia is 100% sold out with backorders ranging to up to a year.
The only remote competitor at hardware is Google with their TPU v5 but, in recent years, Google has been very bad at releasing good products when it comes to software and I donβt see them becoming a GPU hardware company.
Let me end with this very credible source (former Google, Y Combinator):
Iβve heard rumours of Microsoft wanting to build a nuclear power plant for OpenAI.
And when it comes to me β the moment they split the stock, Iβm making a big bet on Nvidia.
keep it real,
Robin Hood
Good strategy... $NVDA is aiming for a 4 for 1 split and can get approval in June
So, I am not sure about the tweet on the power consumption taking down the grid of a state. Tomβs Hardware analysis shows 3.5 million H100s would consume ~13 TWH power annually, which is about the same amount of power consumed annually by the city of Chicago. That is a ton of power. That is in the context of the US using 4,000 TWH compared to Chinaβs 6,800 TWH. This is great context - thanks for setting off this lightbulb!