YWR: 10 NVIDIA Takeaways.
Nvidia makes me uncomfortable.
Uncomfortable because I don’t own it.
We’re all waiting for the AI stocks to crash… but, maybe we have it all wrong.
Maybe we are underestimating everything.
Maybe AI stocks like NVDA, AMZN, META and GOOG have never looked better.
On May 19th Jensen Huang gave a keynote speech at Computex in Taiwan. It was probably the most important speech in the entire financial services industry. It explained everything. It laid out the road map for where we are going, and yet no one mentioned it.
None of the 58 keen financial experts in my multiple chat groups said a word about it. Just the usual stuff about uranium, Trump, how the US is doomed, or AI in general. Nothing about Nvidia.
So if NVDA is such a crowded trade how come nobody is paying attention?
I think most funds and investors don’t own NVDA, or are underweight. So it’s not something they want to think about. They just hope it doesn’t go up too much and kill them. It’s too painful to spend almost 2 hours hearing how well they’re doing because they feel they missed it.
Easier to ignore it, say it’s expensive, and hope it crashes.
I get it. I missed it too. But I don’t feel like missing it for the next 10 years too. And strangely, the valuation doesn’t look to bad here.
So here are my top 10 takeaways from the Computex KeyNote.
#1 Jensen is one of the All Time Greats. Put him up there with Jobs, Bezos, Musk, Buffet and Zuckerberg.
The older I get the more I appreciate the expression ‘Back the Jockey not the Horse.’ It makes me love founder led companies. These guys are cut from a different cloth. They make bold bets, redefine entire industries, and create new paradigms while their corporate moron competitors are chasing ‘low hanging fruit’ and worrying all night what the board thinks of them and if their contract will be renewed. Always playing defence and CYA.
Meanwhile, Jensen is thinking years ahead. He’s building moats, creating industry standards (like with CUDA), and locking up the key manufacturing partners in Taiwan. He’s redefining Nvidia as we speak. We think Nvidia is building chips, but Jensen says he’s building AI Infrastructure. There is an entire AI Industrial Revolution happening and Jensen is moving at light speed to make Nvidia the center of everything.
Another thing. The guy is charismatic, super healthy and speaks well. Being charismatic and able to communicate well is an asset when you are building entire industrial movements and ecosystems (which he is). The keynote address was 1 hour 40 min and he went non-stop through the entire thing with no break, no sip of water, no notes, and no-teleprompter. That’s a physical and mental feat. You can’t help liking the guy and wanting to be part of what he’s creating. Or, owning his stock.
#2 We are still so early in the AI trend. Think about it. It’s only been 3 years since we started playing with chatbots. Now we’re addicted to them. And we still haven’t connected AI to our corporate processes to create the AI information flows of the future (The Containerisation of Thought). Companies don’t run on AI. Yet. But the tools for this are starting to come out, like the new releases from Retool, or the new software from VAST. So AI is still early. Then there is the whole robotics thing which hasn’t even started. It feels like waves piling on top of waves. We keep waiting for the 1999 playbook, but is this playbook something different?
#3. $50 billion AI Factories not ‘datacenters’
Jensen says the word ‘datacenter’ is outdated. These are AI Intelligence factories and they manufacture tokens. He says in the future companies will start to speak in terms of how many tokens they have manufactured. It’s an interesting concept and Jensen refers to it often in his talks. And it’s starting to happen. I caught Sindar Pichai saying exactly this at their I/O conference. In one year Google has gone from processing 9.7 trillion tokens/month to 480 trillion!
And Jensen’s right about the incredible scale of these Intelligence Factories. Can you imagine a datacenter in the scrublands of Abilene, Texas which costs $50 billion!!!! $50 billion? I mean how is that possible? But this is what Jensen is saying we need to understand. The future is intelligence factories and they are on a scale way beyond ‘data centers’.
Of the $50 billion cost for this ‘Intelligence Factory’, $40 billion is the electronics.
#5 Taiwan is the epicentre of AI and Robotics. Every month we run the Global Factor Model to track the top estimate trends in the world. And for the last year the YWR Global Factor model leaderboard has been littered with mid-cap Taiwanese electronics and semi-conductor companies. But these were names we didn’t know. Wistron, Pegatron, Wiwynn, Quanta, WT Microelectronics, ASUS. Who are these companies? What do they do? Why are they scoring so well? What’s the story?
In fact there are so many highly ranked Taiwanese companies in the factor model that Taiwan is consistently ranked as one of the top 5 countries in the world (along with China, Japan and Korea). The data screams to us that Taiwan is hot. But again. What’s the story?
Then Jensen lays it all out in two beautiful tribute videos.
I am so grateful. Now I get it.
At 43 minutes Jensen goes through the entire construction process for a new Blackwell AI server and all the Taiwanese companies which are part of the manufacturing process. The video is to thank all Nvidia’s Taiwanese manufacturing partners, but for us it explains who does what and how it all comes together.
Then at 1hr:32min there is another tribute video to show how Taiwan is leading in ‘software defined manufacturing’. TSMC, Pegatron, Foxconn, Wiwynn, Quanta are building digital twins of the physical factories so they can transition them entirely to robotics.
You realise Taiwan is the Silicon Valley of AI and Robotics and these companies you have never heard of are the leaders in AI manufacturing. There is a 1 of a kind ecosystem in Taiwan. And you realise the geopolitical ramifications too. Taiwan is priceless. The US will never move away from Taiwan. Not now.
#6 Reinforcement of our Contract Manufacturing Theme.
Foxconn is highlighted over and over as a key player in the Nvidia ecosystem. It affirms our view that Foxconn (Hon Hai Precision), FLEX and Jabil are right in the middle of building the AI intelligence factories of the future, right in the middle of building humanoid robots at scale, and benefit from the new world order of manufacturing with dynamic tariffs. The Secret Beneficiaries of Tariffs and Manufacturing 3.0.
#7 $100 billion opportunity for Samsung.