Nvidia dethrones Apple as the most valuable company
markets.businessinsider.com
markets.businessinsider.com
The PE ratio calculated by current market cap of $3.5T divided by that annualized earnings is 52. Not unusual for a growing tech company.
I remember the arguments I hear here on HN these days "No company can be that big and still grow. Also competition will catch up!" from the early 2000's when Apple's market cap crossed $100B.
And from 2017 when Tesla's market cap crossed that of BMW.
It's not that easy. Companies are moving targets. Their future will be determined by revenue factors that are not generating any revenue today. With Apple it was the iPhone. With Tesla it was self-driving. With Nvidia it will be ... ?
Judging by the history of this industry that seems inevitable.
They have a nice money cushion nowadays, but mankind isn't going to keep rebuying iPhone and iPads forever.
Afaiu, Nvidia is a software company. It chose a headstart of a decade.
GPUs are just a platform for running it, in a sense. It’s Nvidia who could run on a competitors GPUs if it wanted, but not vice versa. With own GPUs it has a full lock-in mechanism.
Judging by the history of this industry that seems inevitable.
You mean games?
They have been above Apple's since back then. And Nvidia had grown them to 65% even before the AI craze. They main reason why it becomes so crazy is that demand exploded and with it unit shipments which have grown gross margins and net profit insanely.
Nvidia's management has the best track record in the industry concerning margins. Jensen himself said that he says no to business opportunities if he thinks the margin there is too low. That's why Nvidia has left consoles and left it to AMD almost entirely. Nintendo Switch is an exception because Nvidia could there simply deliver a developed shelf product and skipped the huge effort of developing a console specific chip.
Early 2000s the iPhone hadn’t been invented.
Apple’s market cap would be a tiny fraction of what it is today if they had remained a Mac + iPod + software company.
Apple’s genius has been their ability to enter and dominate (and sometimes create) completely new markets.
Since the early 2000s, the following new markets have all eclipsed the Mac or the iPod in annual revenues for Apple:
iPhone iPad Services Accessories
Basically, whatever company Apple was in the early to late 2000s, is the least financially significant bit for Apple today.
That was my point.
Companies - especially tech companies - go through phases. Apple is now a phone company, Microsoft a cloud computing company, Tesla an AI company ...
The value of a tech company is not in their current product line. It is in the product lines they will build in the future.
The growth of the Appstore business is another example of what I am trying to convey here. That companies are moving targets. When the iPhone came out, people were estimating how fast the competition could catch up with the hard- and software. AFAIR appstore fees were not a part of the discussion back then.
Compute power from Nvidia's ecosystem can benefit everything and Nvidia is working on growing the ecosystem with HW + Networking + SW. Nvidia's data centers as ecosystem will become as flexible in doing whatever you need as the PC and Smartphone became that way. And that is their real moat. They will have competition in many specific areas but no one will compete with Nvidia solutions on a global scale of applications.
If AI isn't this productive then they are still the safest bubble company in the bubble.
This situation, to me, says that most people are unsure of how productive AI will be so they don't want to directly invest in it, but following dot com logic companies partaking in tech bubbles will still beat the market as long as their P/E isn't in the multi hundreds year on year. Nvidias' P/E is similar to intel back then iirc
The modern ai flavor has been there for almost two years now yet nothing hints to a productivity boost, nothing even remotely close to a scale that would be relevant.
Crypto mining kept demand for Nvidia processors artificially high throughout the pandemic and now then they transitioned immediately into AI hysteria. They, of course, have a healthy business without those things, but it is not entirely clear to me if such accelerated growth is sustainable.
Imagine that you can build a complete factory virtually first and then test run it. Planning is the most effort in building a factory so you have to do it in virtual or real world equally. But in virtual world you can allow mistakes and correct them while in the real world you have 1 shot and a simple mistake can cost you millions later on.
The same applies to many analog industries which aren't digital yet. Digitalization for IT and Tech is computers which are used in development and administration but analog processes in manufacturing are still far from being digital. Digial worlds is what will be a huge change in traditional industries and Nvidia is creating a global solution for it called Omniverse. And of course with a headstart, since the release was several years ago and now Nvidia is slowly distributing it into the world.
Nobody sees it, nobody gets it but just like Nvidia started DGX 8 years ago and today everyone is surprised about Nvidia's dominance in AI compute, the same will happen with Nvidia and digital worlds.
Also digital worlds will uplift 3rd world countries because they can skipp Industry revolution 2 and 3 and go straight to virtual solutions which then apply to real world. Just like many 3rd world countries in the past simply skipped cable TV and hard wired phones and jumped directly to mobile phone technology where they dominate now (see South Korea and others).
I wonder how many factories / processes need this much simulation before they are build.
Sure, the software may not be as good as NVIDIA's at the moment, but would it take AMD $3 billion to match it, or $3 trillion? It didn't take NVIDIA trillions to develop CUDA!
Similarly, for users of NVIDIA hardware, are they willing to fork over $3 trillion of their own money instead of, say, a "mere" $2T to AMD and work out the kinks in the software compatibility with the remaining one trillion dollars!?
These valuations are not just insane, they are certifiably nuts.
I'm still waiting for Apple Intelligence to ship, Windows Copilot is looking more like Clippy every day, and ChatGPT might partially replace Google Search for me, but not the rest of Google's products.
I just don't see that $3.5T materialising as revenue before the bubble bursts, and if it doesn't, not before AMD starts taking a larger slice of the trillion-dollar-pie.
Compare ICE auto industry profits since 2003 when Tesla was founded to money invested in Tesla till 2019 when Model Y was finished. Tesla had very less money to work with.
Multiple iterations of Volkswagen's car software sucked so bad and the company culture and leadership just couldn't develop good software so Volkswagen created a separate startup style subsidiary just to make software. Guess what? That startup also failed and they recently bought a $5 billion stake in Rivian so that they could use Rivian's car software. Meanwhile their sales are tanking.
Also look at SpaceX vs. ESA or Boeing/ULA with reusable boosters. Or China with EUV and chips. Even nation state level with essentially unlimited funding may not work.
Indeed, though that sword cuts both ways, if a hard problem gets fixed with lots of money being thrown at it then a competitor may be able to fix it for less if they have the right culture and leadership.
> look at SpaceX vs. ESA or Boeing/ULA with reusable boosters.
The way I understood it was that SpaceX was granted $3B of tax payer money to set up a moon mission and all they achieved was a single booster catch. That project is gonna go so far over budget both in terms of time and money that it's giving me a stomach ache just thinking about it.
There's simply no comparison when it comes to cost efficiency.
AMD doesn't need to duplicate everything NVIDIA provides, they just need to duplicate the parts relevant to most of the $3T spend the market seems to be expecting.
Just make llama.cpp work robustly on AMD accelerators, and that might unlock a $500B slice of the pie by itself.
Sorry, got something in my throat.
That is OpenCL 3.0 is OpenCL 1.0 without the OpenCL 2.x stuff that no one ever adopted.
In effect, they already have. Both AMD, Apple and a number of smaller OEMs all wrote GPU compute shaders to do "the AI inference stuff" and shipped it upstream. That's about as much as they can do without redesigning their hardware, and they've already done it.
Nvidia wins not because they have everything sorted out in software. They win because CUDA is baked into the design of every Nvidia GPU made in the past decade. The software helps, but it's so bloated at this point that only a small subset of it's functionality is ever used in production at any one point. What makes Nvidia fast and flexible is integration of compute in hardware at the SM-level. This is an architecture AMD and Apple both had the opportunity to reprise, even working together if they wanted, but chose not to. Now we're here.
I tend to steelman the idea that it was AMD and especially Apple's mistake for eschewing this vision and abandoning OpenCL. But apparently a lot of people tend to think that AMD and Apple were right despite being less efficient at both raster and compute operations.
With OpenCL, C99, some C++ support, printf debugging, and that is about it.
For good C++ experience, one needs to reach out to Intel's Data Parallel C++, which has Intel's special sauce on top of SYCL efforts, which only became a reality since a British company specialised in compilers for game consoles decided to pivot their target market and produce ComputeCpp, followed by being acquired by Intel.
A real threat is Cerberas. Their approach avoids the memory bottleneck of loading the whole model DRAM->SRAM for each token batch. I hope their IPO goes well before the AI bubble pops.
For example, arrays of identical and simple chips with in-chip memory could have similar performance as the monolithic Cerebras wafers-scale chips. Not for all workloads, but some, such as inference.
The problem is that AMD seems to be allergic to writing a software shim that allows users to keep using their existing Tensorflow and PyTorch code at a 10% performance penalty.
The reason Nvidia dominates is that all the AI research is done on their GPUs for the past decade. That's why any new approach in AI will be on Nvidia's GPU and Nvidia will always fine tune their ecosystem based on research results.
That's why the 6 months of A100s used to traing GPT-3.5 become 10 days with Blackwell. That's why Nvidia dominates since nobody even considers alternatives in AI research validation than Nvidia. And one major reason is that CUDA is dominant and AI researchers have known it for a decade.
You see, Nvidia is talking about "all AI", so training, inferencing, networking, applications, everything. HW competitors only mean inferencing when they talk about AI and wonder why traction is so slow. Nobody on this planet would buy a Nvidia data center for research and training and then a competitor's data center for inferencing next to it. You simply extend your Nvidia data center seamlessly (thanks to Nvidia tools) and use it dynamically at full utilization for training, inferencing and whatever you need otherwise. To beat Nvidia, you have to beat them at everything, not just inferencing and even there on specific fields lol.