At some point one of these Nvidia doomers will be right but there is a long line of them who failed miserably.
The article explains that Nvidia's biggest customers (50% of datacenter revenue) are switching to their own hardware.
NVIDIA is very strategic about building product to avoid commodification -- both by building out network effects where software is tied to their proprietary sdk libraries, and by always focusing on being at the cutting edge of product.
Both these things can be true: a large company should try to build their own hardware to reduce supplier risk, and a large company should be open to suppliers that have better product that delivers business value.
So far, these large companies' internal hardware has been useful internally but never a complete replacement for NVIDIA, which keeps staying at the cutting edge of new capabilities.
NVIDIA already faced existential risk when Intel was commodifying all the dedicated motherboard components in the late 90s, 2000s, (like sound cards etc), so they're hyper-aware of this.
The article seems focused more on stock price and where to bet, than the market for GPUs or generic hardware vendors.
Fair Disclosure: I am very neutral when it comes to FLOPS/W/$ and the generality of those FLOPS. Given inference and training, the advantage is slipping.
Edit:
- I wonder what's stopping Nvida from releasing an AI phone
- A LLM competitor service (Hey, how about you guys make your own chips?)
- They are already releasing an AI PC
- Their own self driving cars
- Their own robots
If you mess with them, why won't they just compete with you?
Just wanted to say one more thing, that Warren Buffet famously said he regretted not investing in both Google and Apple. I think something like this is happening again, especially as there are lulls that the mainstream public perceives, but enthusiasts don't. To maintain the hyperbole, if you are not a full believer as a developer, then you are simply out of your mind.
It's a low margin business and would hurt the balance sheet more than the completely irrelevant revenue from a project like that.
I've been investing in semi for decades and what strikes me about this recent cycle is that so many don't seem to understand that semi is a highly cyclical business that is prone to commoditization waves and inventory/capacity overbuild.
And speaking as a trader, instead of reinforcing your firmly held base case, I'd strongly consider painting out the bear cases. Look at the roadmaps of the hyperscalers that are designing their own chips for internal use, etc. And never use the word faith when it comes to markets.
You could easily see NVIDIAs margins get chopped down, and see the multiple re-rate lower from here. Actually, I'd argue the name is well on the way down this path already.
It's almost guaranteed to happen sooner or later. Semi down cycles are usually brutal for semi equities.
That's not to say it isn't a great company. It's certainly not a Buffett name though.
The cyclical stuff was the argument made for semis during the 2010s when no one gave a shit about semis really. I think the game changed, but again, I do operate on faith, or in investor terms, conviction. The main evidence for why the game has changed to me (well, other than AI being the most incredible piece of tech we ever built) is mostly that there are companies that have no business making chip hardware now interested in making chip hardware. That's not usually part of the cycle.
One is often unrealistic and later one is lot more common. And one really should consider later one in long term investments.
B2C is a hellish headache that has marginal returns if you are not B2C first, and the amount of investment needed to be B2C competent just isn't worth it when there are alternative options to invest in
> A LLM competitor service (Hey, how about you guys make your own chips
Already exists. AI Foundary
> They are already releasing an AI PC
It's just an OEM rehash
> Their own self driving cars
Not worth the headache and also losing customers like Google or Amazon due to competitive pressure
------
Cannot reply: releasing their own "Nvidia Car" means they will lose their existing automotive partners becuase they will not spend on a competitor. Same reason Walmart stipulates EVERY tech vendor must decouple from AWS when selling to them.
I'm curious to know more about this if you (or anyone else) can elaborate on it.
What constitutes a tech vendor? Are you talking about Walmart buying PCs from Dell, from buying/renting a SaaS from someone, from IT-consulting coming in to do a one-time service for them (even if that service takes years)?
You're not talking about stuff like "Apple wants to sell iPhones to Walmart customers", I assume - yes?
I've tried explaining that one or two AI data center clients for Nvidia dwarfs the entire gaming GPU market, but he just doesn't get it.
Did you know that Nvidia has a gaming cloud running which might become the largest in the world at some time?
In 10-20 years, Nvidia might make more revenue from gaming cloud than they do today with gaming HW.
Anyway, maybe marvell should focus more on the the consumer side since hobbyists seem to be building crazy ai rigs and likely needs drives, at the very least, for models. It sort of seems like hobbyists are devouring any worthy gpu that gets produced.
WRT your edit: The answer to all of this is that it's very hard and requires a huge amount of investment to produce good vertical solutions in each of these spaces. You cannot build a good AI phone without first building a good phone. You cannot build a self-driving car without starting with a good car, etc. For robots, I'll point you to someone using Nvidia chips: the Matic is a complete ground-up rethink of how robot vacuums should work. It's taken them 7 years to get to early adopter phase.
More like you cannot build a self-driving car without starting with a good phone. See Huawei.
Agreed, and for all the "price crash" I still can't just whip out my credit card and purchase an hour or two on an H100/B100.
It's still multi-year contracts and "Contact Sales".
See it that way, if you have an OS/SW for all the industries you mention then who is your competitor? Not the participants in that industries. Nvidia can partner with any automotive company but won't compete with any of them as long as they don't build cars. But imagine the potential of every self driving car being build using Nvidia AI?
Think about the potential of every robot build using Nvidia AI?
Think about the potential of any AI Service using Nvidia AI?
See, Nvidia isn't directly competing in the enduser market but instead focuses on the B2B. Nvidia can also create many different revenue streams from 1 customer.
For example an automotive customer: - Nvidia HW in car for AI - Nvidia data center on-prem/cloud for DriveSim in car - Nvidia Omniverse for car design and manufacturing simulation - Nvidia Isaac for robotics/logistics in manufacturing plant - Nvidia Cosmos+GR00T for robots inside the plant - Nvidia edge devices inside any robot in the plant - Nvidia NeMo data center on-prem/cloud for AI models / LLMs for internal use
And what will be the advantage? Nvidia can actually make it more and more seamless to operate between all Nvidia solutions. For example, you can do an update to your robots in Cosmos, simulate it in Omniverse and with 1 click update your Nvidia driven real robots. The alternative is that you have 3 solutions from 3 different vendors with no interface between them.
People have no idea, what Nvidia is actually creating. Nvidia has more SW engineers and even Nvidia employees call Nvidia an AI SW company. They publish so many libs and lots of other SW stuff that it's sometimes hard to keep up. Just look at all the RTX goodies for gamers which Nvidia is developing. And they are all free, well except that you need Nvidia HW for it. The same model, Nvidia will apply to ALL industries in the world. And here they discuss about CSPs being an issue for Nvidia while Jensen focuses to build a Mega Corp. which potential TAM is in every industry in the world :)
This is what will help protect Nvidia now that DC and cluster spend is cooling.
They own the ecosystem thanks to CUDA, Infiniband, NGC, NVLink, and other key tools. Now they should add additional applications (the AI Foundry is a good way to do that), or forays into adjacent spaces like white-labeled cluster management.
Working on building custom designs and consulting on custom GPU projects would be helpful as well by helping monetize their existing design practice during slower markets.
Of course, Nvidia is starting to do both, with Nvidia AI Foundry for the former and is working on the latter by starting a GPU architecture and design consulting as announced at GTC and under McKinney
No they do not. The article explains that Google, Amazon, Microsoft, and Meta are developing their own hardware and software for AI/HPC.
Google Gemini was not trained using CUDA or Nvidia hardware.
Chinese CSPs are the only ones can develop their own hardware / software for AI / HPC.
Apart from Nintendo, who has successfully partnered with Nvidia? Apple, Microsoft and Sony have all been burnt in the past.
Nvidia has started formalizing that last year [0], but it's a new muscle for them.
[0] - https://www.reuters.com/technology/nvidia-chases-30-billion-...
There’s maybe some wiggle room, in that these AI distributed systems might not (?) look like HPC/scientific computing systems—maybe they don’t need Infiniband style low latency. So these other funky networks might work.
But like, Nvidia has the good nodes and the good network. That’s a rough combination to compete against.
That's not a trend yet. We're about to enter an era where most media is generated. Demand is only going to go up, and margins may not matter if volume goes up.
> The open question is long-term (>6yrs) durability1. Hyperscalers (Google, Microsoft, Amazon, and Meta) are aggressively consolidating AI demand to become the dominant consumers of AI accelerators; while developing competitive, highly-credible chip efforts.
Hyperscalers aren't the only players building large GPU farms. There are large foundation model companies doing it too, and there are also new clouds that offer compute outside of the hyperscaler offerings (CoreWeave, Lambda, and dozens of others). Granted, these may be a drop in the bucket and hyperscalers may still win this trend.
But to your point, Disney is using GenAI in their new live action Moana film. Presumably that'll do lots of sales.
Others talk about chips when Nvidia thought about interconnects 8 years ago. Today, competitors try to catch up on this while Nvidia talks about One Giant GPU.
The next step will be scale up and then scale out.
Nvidia is always ahead because what the article fails to see is that where CSPs are today is where Nvidia was in the last decade. Nvidia has a working ecosystem for everyone which they can now fine tune with actual customers.
Not ideal for them but hardly a death blow
They are losing their biggest customers to custom in-house silicon, and smaller orders are going to compete with a market being flooded by superfluous hardware from companies which either went bust due the the AI bubble shrinking, or went bust because they weren't able to compete with the big fish.
Gaming is only 7% of Nvidia's revenue.
I figure regardless of tariffs and competition and other fluctuations the demand for computing power will endlessly trend upwards and as much as we can produce will be consumed.
Good luck to you.
Personally, I have found NVIDIA to be one of the most hyped stocks I have ever seen and it feels weird to take a position that it is under-hyped. That said, people aren't capable of beating the market consistently so I would never invest based on this, or any intuition or information I had.
