Hacker News Reader

Comment by ein0p | Hacker News Reader

As much as I’d like to say it’s overpriced, its only potential competitor, AMD, is even more overpriced with an eye watering P/E of 324 as of right now, three times that of NVIDIA. Such numbers are stark raving insanity for hardware companies. For comparison, Apple, fully in possession of the world’s most valuable brand, and a gigantic money printer powered by a loyal, billion+ user strong customer base, and a deep pipeline of products, has P/E one third that of NVIDIA, and 1/10th of AMD. Make this make sense.

View on HN

The forward PE is 31 for AMD, and Nvidia is at 33. Apple's forward PE is 26. The market prices are based on future expectations. Nvidia is expected to increase earnings by something like 3x, and AMD by 10X

Apparently, AMD can remain irrational longer than retail investors can remain solvent.

Right, I'm not saying they will hit those projections, that's what they are though

To be fair, it's much easier to increase your profits when your profit margins are 5% (AMD) vs 50% (Nvidia)

How does one find information on forward P/E?

Are there historical forward P/E charts to get some context on whether the current prices are expensive / in-line with how tech is valued over the past ten.. twenty years?

Thanks!

Generally, any forward metric is the consensus (mean) of analyst estimates, unless you have your own model for what you think it will be (in which case you probably are still callibrating it around the street's view)

Since you aren't going to have an account with every broker to get their research, instead you'd sign up for one of Capital IQ, FactSet or Bloomberg which aggrgate the info for you and are listed here in increasing order of price and quality

Banks would have the chart you're looking for but they're fairly easy to recreate if you have the data. (Source: I worked in M&A for nearly a decade in such banks)

Intel is jsut as much a competitor to Nvdia as AMD. I suspect they will work together on creating a CUDA alternative which is the biggest problem both of them face. If they can figure it out I dont see how nvdia isnt fucked as the money is in the big cloud purchasers who have every incentive to switch as soon as a legitimate alternative presents itself. Even if amd/intel dont beat Nvdia coming close with a product that "just works" for half the price is certainly on the table and will be compelling for the hyperscalers.

Intel can’t even make a competitive CPU anymore, let alone a competitive compute-oriented GPU. It’s way behind on lithography, with no end in sight. Their GPUs also aren’t properly integrated with any popular DL framework. I honestly have no idea what they’re thinking. If I were them, at least PyTorch would work flawlessly with their current Arc offerings. It doesn’t - it’s a decidedly “enthusiast” affair for people with too much time on their hands. You don’t even need a huge team for this. 25-30 competent C++ folks for 9 months to a year is all it takes.

I firmly believe intel is incompetent. I also firmly believe that with enough time and money they can come up with a decent offering. Their GPU is still very young, and theyve got plenty of time to figure it out. Theres just no way nvdia keeps its datacenter GPU stranglehold for a decade plus which is what theyd need to do for this market cap to make sense.

> CUDA alternative

The should name it OpenCL.

"Nowhere to go but up!"

Up? Idk, I certainly don't have the nerve to buy into a position at this kind of P/E. But they don't need to run faster than the bear, and their next best competitor has the agility of a stoned sloth when it comes to the software side of things.

Feel free to short sell both?

Is this the only reply you can come up with in this post? Maybe some people want to have legitimate discussions on valuations.

To expand: you can be an armchair speculator about companies being 'overvalued'. But the market consensus is what it is, and it's likely the best estimation for the expected value of the discounted future returns of the company.

(And if it's off, you can bet against that consensus.)

Reply on news.ycombinator.com