News Jensen Huang says even free AI chips from his competitors can't beat Nvidia's GPUs

[Admin]

Jensen Huang has taken to the stage to claim that Nvidia's GPUs are "so good that even when the competitor's chips are free it's not cheap enough." Huang explains that Nvidia GPU pricing isn't really significant in terms of the all-important data center TCO.

Jensen Huang says even free AI chips from his competitors can't beat Nvidia's GPUs : Read more

 

[CmdrShepard]

Jensen Huang has always been a veshenaan.

 

[hotaru251]

too bad companies favor profit over respect else they'd all refuse to buy nvidia for being talked down to by their CEO.

 

[Alvar "Miles" Udell]

The bigger they are, the harder they fall, though nVidia's P/E is only 77, while AMD's is 403, so the market has huge expectations from AMD, they're the ones that need to deliver, and that's going to need to start with aggressive pricing, something AMD has essentially abandoned.

 

[atomicWAR]

To Jensen I say...

I look forward to the day someone knocks Jensen off his high horse. We need competition and his antics only prove as much.

 

[DavidLejdar]

As it is, he sure may have a point. But then again, with all that hype and market movements etc., already tomorrow several corporations may announce a syndicate to accumulate as many shares as possible of TSMC. And then they might want to use production capacity in the near future for themselves mostly. Which would mean for Nvidia, that they would have a real production issue quite quickly, while getting their own fab up and running may easily take years. And with barely any revenue in sight, Nvidia's share price may drop so quickly, that someone else may consider incorporating them.

I mean, not saying that it will sure be at least as exciting. After all, the large tech investors seem to prefer companies, which don't have capital tied up in material assets. Just saying that it might not be all a smooth sail for years to come.

 

[bit_user]

The bigger they are, the harder they fall, though nVidia's P/E is only 77, while AMD's is 403,

I was going to comment on Jensen feeling the need to try and justify Nvidia's P/E, but... yikes!

so the market has huge expectations from AMD, they're the ones that need to deliver, and that's going to need to start with aggressive pricing, something AMD has essentially abandoned.

Um, you know the "E" stands for Earnings, right? The only way to boost earnings by price-reductions is if you think you can drive enough additional volume to make up for the reduced margins. I see neither adequate pent-up demand, nor enough price sensitivity to expect the market would be very responsive to further price-cutting.

AMD needs to do more to swing their P/E than that. The main thing would be acquisitions. That would potentially let them trade some of that share price for additional revenue streams, so long as it's an established company that doesn't overlap much with their existing portfolio. Of course, acquisitions take time and are limited by a company's internal resources. So, you can only turn that crank so fast. Plus, the revenue doesn't get recognized as yours until the acquisition formally completes, which can take a couple years (or longer, if regulatory approval gets drawn out).

 

[ekio]

Why are you making an article every time this guy speaks?
The last time he said something that was:
Kids, stop learning coding because my expensive gpus can do it for you (using human code as a source).
He will say anything that serves himself.

 

[Nyara]

Why are you making an article every time this guy speaks?
The last time he said something that was:
Kids, stop learning coding because my expensive gpus can do it for you (using human code as a source).
He will say anything that serves himself.

Nvidia (inflated bubbled up) stock market cap makes everything he says have a couple of hundred billion dollars domino effect, specially relevant as Nvidia alone accounts for most of the GDP growth in US for the last few quarters. We live in a sad world.

 

[craigss]

Once CUDA is taken out of the equation then AMD has the better solution, its quicker and uses less power, Nvidia has been very clever in initially making software thats propriety which then fans the flames of demand, its the oldest trick in the book create the need then supply that need, CUDA is the key once thats sidelined Nvidia will crash

 

[bit_user]

Nvidia alone accounts for most of the GDP growth in US for the last few quarters.

Source?

 

[Nyara]

Source?

US GDP (in USD) https://fred.stlouisfed.org/series/GDP
Q4 2023 27.944 trillion (gain of 1.536 trillions from Q4 2022)
Q3 2023 27.610 trillion
Q2 2023 27.063 trillion
Q1 2023 26.813 trillion
Q4 2022 26.408 trillion

Nvidia Market Cap (in USD) https://ycharts.com/companies/NVDA/market_cap
Q4 2023 1.221 trillion (gain of 0.856 trillions from Q4 2022)
Q3 2023 1.049 trillion
Q2 2023 1.034 trillion
Q1 2023 0.686 trillion
Q4 2022 0.365 trillion

Yes, it is getting absurd.

 

[Bikki]

Misleading title, "free chip" should be put into the context of TCO, which includes the cost to run them is still higher than NVDIA.

 

[bit_user]

US GDP (in USD) https://fred.stlouisfed.org/series/GDP
Q4 2023 27.944 trillion (gain of 1.536 trillions from Q4 2022)
Q3 2023 27.610 trillion
Q2 2023 27.063 trillion
Q1 2023 26.813 trillion
Q4 2022 26.408 trillion

Nvidia Market Cap (in USD) https://ycharts.com/companies/NVDA/market_cap
Q4 2023 1.221 trillion (gain of 0.856 trillions from Q4 2022)
Q3 2023 1.049 trillion
Q2 2023 1.034 trillion
Q1 2023 0.686 trillion
Q4 2022 0.365 trillion

Do stock trades even contribute towards GDP? If they do, it would be the value of the trades and not the value of the company that would matter. I guess you'd have to integrate the product of the daily trading volume multiplied by the daily average market price, to see how much it actually contributed towards GDP.

As far as I'm aware, what would count toward GDP is their revenues. For the past 4 quarters, in reverse order, those are:

• $0.0221T
• $0.01812T
• $0.01351T
• $0.00719T

 

[Nyara]

Do stock trades even contribute towards GDP? If they do, it would be the value of the trades and not the value of the company that would matter. I guess you'd have to integrate the product of the daily trading volume multiplied by the daily average to see how much it actually contributed towards GDP.

GDP is the full change of assets in the last 12 months in a country, plus exports, minus imports. A stock market for 1 trillion for a company, if it does not change, produces 0 GDP, but if it moved from 0.1 trillion to 1 trillion in the last 12 months, GDP and GPD quarters will account to it as a 0.9 trillion gain in GDP.

After 12 months have spanned and assuming the company kept a market cap of 1 trillion, it will no longer contribute to future GDP, except on the revenues it generates (which is of about 0.060 trillion per year right now for Nvidia, or 60 billions.)

So, as of December 31 of 2023:

0.856 (Nvidia stock raise in last 12 months)/1.536 (GDP increase, aka last 12 months asset increase/revenue increase)
=
0.5572916667, or Nvidia stock's contributed 55% the GDP growth of US's 2023.

Well, at least potentially, that money must be spent first yes.

 

[kjfatl]

The likely winner in this is Intel.
They might make their own solution in addition to being the major supplier for high-end AI chips sold to Nvidia, AMD, Meta, Microsoft, Google and others.
TSMC can make the mid-range chips w/ China making the low end ones.
All info puts them at least 1 full generation ahead of TSMC in 4 years, maintaining that lead.

 

[P1nky]

US GDP (in USD) https://fred.stlouisfed.org/series/GDP
Q4 2023 27.944 trillion (gain of 1.536 trillions from Q4 2022)
Q3 2023 27.610 trillion
Q2 2023 27.063 trillion
Q1 2023 26.813 trillion
Q4 2022 26.408 trillion

Nvidia Market Cap (in USD) https://ycharts.com/companies/NVDA/market_cap
Q4 2023 1.221 trillion (gain of 0.856 trillions from Q4 2022)
Q3 2023 1.049 trillion
Q2 2023 1.034 trillion
Q1 2023 0.686 trillion
Q4 2022 0.365 trillion

Yes, it is getting absurd.

This shows how little understanding of economics you have.

 

[peterf28]

I hate reading this news, I owned 80 NVDA stock, and sold them all in 2010 for cca ~1000USD together. Now worth like 60000. ((

 

[Pierce2623]

Yet people are buying the mi300? Imagine that.

 

[Pierce2623]

GDP is the full change of assets in the last 12 months in a country, plus exports, minus imports. A stock market for 1 trillion for a company, if it does not change, produces 0 GDP, but if it moved from 0.1 trillion to 1 trillion in the last 12 months, GDP and GPD quarters will account to it as a 0.9 trillion gain in GDP.

After 12 months have spanned and assuming the company kept a market cap of 1 trillion, it will no longer contribute to future GDP, except on the revenues it generates (which is of about 0.060 trillion per year right now for Nvidia, or 60 billions.)

So, as of December 31 of 2023:

0.856 (Nvidia stock raise in last 12 months)/1.536 (GDP increase, aka last 12 months asset increase/revenue increase)
=
0.5572916667, or Nvidia stock's contributed 55% the GDP growth of US's 2023.

Well, at least potentially, that money must be spent first yes.

Warning kids. Thinking you understand things you don’t makes you look stupid.

The likely winner in this is Intel.
They might make their own solution in addition to being the major supplier for high-end AI chips sold to Nvidia, AMD, Meta, Microsoft, Google and others.
TSMC can make the mid-range chips w/ China making the low end ones.
All info puts them at least 1 full generation ahead of TSMC in 4 years, maintaining that lead.

Realistically, Intel would have to do more than just gain a slight advantage to start gaining major TSMC customers. AMD and Apple are two of TSMC’s 5 largest customers (with Apple being the largest) and they’re extremely unlikely to ever use Intel Foundry Services. Qualcomm and Mediatek would be the easiest big money customers to get. Maybe Nvidia will come but AI is too big for any company to maintain an unassailable lead. There’s too many startups willing to take technological risks for big gains.

 

[vanadiel007]

He forgot to add "right now" to that statement. You can bet there's some serious R&D going on at the traditional big players in the market to produce AI chips.

It's just a matter of time before another Company takes the lead on either pricing, performance, power consumption or other key metrics.

 

[Avro Arrow]

Let me guess... "Everything just works!" right?

 

[Samlebon2306]

Anything that rises too fast, falls much faster.

 

[The Historical Fidelity]

The bigger they are, the harder they fall, though nVidia's P/E is only 77, while AMD's is 403, so the market has huge expectations from AMD, they're the ones that need to deliver, and that's going to need to start with aggressive pricing, something AMD has essentially abandoned.

Wow that’s the highest P/E ratio I’ve ever seen. Definitely not a company to invest in if you like dividend based ROI.

 

[bit_user]

The likely winner in this is Intel.
...
All info puts them at least 1 full generation ahead of TSMC in 4 years, maintaining that lead.

You're assuming everything goes to plan. Recent history would cast doubt on such assumptions.

We need look no further than their decision to cancel desktop Meteor Lake, in order to see that not all is right with Intel's manufacturing division. If you look at how the Meteor Lake-powered laptops compare against the top-end Raptor Lake models, you can clearly see that Intel 4 was another underperforming node - just like their early 10 nm iterations.