I had written about Unity in a previous post where I discussed the “picks and shovels of the Metaverse”. Another company, that falls firmly into that bucket is Nvidia - the world’s largest (and best) AI company.
At its core, Nvidia makes the chips that are used to power the graphics cards for video games but today those chips also power the brains of AI computers and data centres, globally. In the past, semiconductor stocks would exhibit extremely vicious cyclical behaviour. Boom and bust cycles in the manufacturing of wafers due to very short demand cycles and very elongated supply cycles would make allocating capital in this sector like trying to guess the plot of a movie by looking at the end scene. Over the last couple of decades the growth of the fabless industry (chip designers that don’t own underlying factories) along with the rise of TSMC, the world’s largest chip manufacturer, has meant that companies have been able to specialise in doing, a few things very well, at scale.
Both gaming graphics processing and AI need the ability to parallel process large amounts of data. Nvidia is the master at producing chips that lead the world in this task. They have kept this lead that despite dozens of startups attempting to compete, it is proving to be an impossible task as the company keeps cannibalizing and pushing the envelope of its own business.
I would urge you to take 15 mins out of your day to watch their recent keynote presentation to investors
The important thing about Nvidia is that it is no longer just a chip designer. It is a chip designer with one of the world’s most powerful AI platforms, operating system and now its own apps.
NVIDIA see a world where they “million-X” data computation speeds (they have already done that in the last 10 years). Where AI is at the centre of everything we do, in fact, they want to build AI factories much like we build factories that churn out cars, they are building factories that churn out insights.
If I was to deep dive into NVIDIA this would be a 100-page blog post and nobody wants that. So let’s focus on the GTC (investor day) announcements for now and the insanity of the tech they are announcing.
🏎 They just announced a new data centre GPU called Nvidia H100 which has the largest leap in generational performance ever.
This looks innocent enough right?
This chip has an input-output rate (think of this as the width of the road) that is so big that just 20 of these chips can sustain ALL of the world’s internet traffic. Just stop and think about that for a second.
This is the type of hardware that isn’t just an incremental improvement for the world’s data centres it’s a giant leap. That is just one of many innovations that they announced at GTC.
But where things get really interesting is their software
Autonomous Driving
Nvidia is trying to dominate all areas of AI including autonomous driving. But rather than having a closed-ended platform like many others, they are inviting others to leverage their technology to build fully autonomous driving systems. Their first fully autonomous car will ship with Mercedes Benz in 2024, then Jaguar in 2025. This isn’t some pipe dream AI, this is real and we will see it on our roads soon, all powered by NVIDIA
Healthcare
Just like with the driving platform. Nvidia also has a world-beating platform for healthcare where they are recreating the human body in VR, training doctors to not only operate but do so collaboratively. Because their software works seamlessly with their hardware this is possible only with NVIDIA.
Autonomous Mobile Robots
Think of these as driverless cars for indoors. The use cases are enormous. Amazon is already using the tech to power robots inside warehouses, factories, shopping malls etc Anywhere that you are in a confined space and have heavy/dangerous/repetitive/sensitive tasks to do robots like these will save time and money. With labour shortages & costs going up in most of the world, we will only see these devices proliferate further.
Digital Twins
I wrote about digital twins in my post on Unity. Nvidia is another leader in the market in helping create a platform for Digital Twins. They are approaching it from the pure AI side, whereas Unity is coming at it from the gaming side with highly realistic lighting and a physics engine. Either way, you need Nvidia hardware to better run Unity. For me, digital twins is a very exciting use case for platforms like Nvidia’s. Whether it's replicating a city to figure out where to put telecoms towers, to recreating an interior of a building for space and design planning. Nvidia’s system allows you to do all of that together with real-time collaboration. Platforms like this will be interesting because of their stickiness, once the twin is up and running it is very hard to move to another platform. The challenge is going to be finding enough people that can actually create the twins in the first place. The world needs to train a lot of designers in new tools and fast.
Maxine web conference AI
Now we have some real-life use cases that will work today. Maxine is AI that can be used for video conferencing. The first picture is genuinely useful. It corrects the speaker’s eye contact so you can be reading a presentation or teleprompt but it looks like you are speaking directly to the camera. Also useful for when you are reading email during boring zooms. The second shot is real-time translation. I will withhold my views on this until I use it but we are definitely getting closer and closer to the point where direct translation AI is within reach. This will open up a whole new world of business opportunities.
During their investor day, Nvidia talked about having a $1trn TAM. This isn’t some number that an intern slid into a startup pitch deck, this is real because AI will touch almost every aspect of lives going forward. They laid out their TAM as follows
Gaming ($100bln) With 3bln gamers, 25% of whom will spend $100 per year on gaming in high-performance GPUs
Enterprise software ($150bln) - Nvidia provides AI tools and SDKs to 50mln servers at $2000 per CPU
Omniverse ($150bln) - Nvidia’s version of Metaverse, their collaboration platform. 45mln designers and creators and tens of millions of factories and warehouses using digital twins. Revenue per is thousands to millions of dollars per twin (discussed later)
Chips and Systems ($300bln) - core to their business
Automotive ($300bln) - software opportunity far higher for the software than the hardware
Valuation - has never been cheap for this company. At a $662bln market cap and 43x24 PE, they will do $41bln of revenue in 2024. BUT I would argue that the quality of the revenue is getting better moving forward as it’s more software than hardware-based so they deserve a premium.
I have exciting news to share: You can now read Billy’s Newsletter in the new Substack app for iPhone.
With the app, you’ll have a dedicated Inbox for my Substack and any others you subscribe to. New posts will never get lost in your email filters, or stuck in spam. Longer posts will never cut-off by your email app. Comments and rich media will all work seamlessly. Overall, it’s a big upgrade to the reading experience.
The Substack app is currently available for iOS. If you don’t have an Apple device, you can join the Android waitlist here.