Discussion AMD Earnings Q3 2025

[mikegg]

AMD: $1.20 x 4 = $4.8

NVDA: $1.05 x 4 = $4.2

Chips sales are seasonal. Q3 and Q4 usually the best quarters.

Part of NVidia "Gaming" is also AI. Could be easily 1/3 of the gaming, and real gaming being sub $2 billion.

Let's not make numbers up.

 

[mikegg]

I'm sure even Jensen himself knows the AI bubble will burst eventually. Just have to keep selling the hype until then.

He's probally long onto the next thing to hype.

And do you think we're in 1999 or 1995 of the dotcom bubble for this AI bubble? The difference could mean sell everything now or buy as much as you can now.

 

[Thibsie]

Is all this really about AMD Q3 earnings?

 

[Joe NYC]

Chips sales are seasonal. Q3 and Q4 usually the best quarters.

I picked the most recent results.

Let's not make numbers up.

We know that similar to Crypto boom in "gaming", there is an AI boom in "gaming".

We will never find out from Jensen what percentage of "gaming" was Crypto. Similarly, we will never find out from Jensen what percentage of "gaming" now is AI. All we can do is estimate.

Gaming, in aggregate, has not gone up by 1/3 in last 4 quarters as NVidia reporting shows. Does not seem real to me.

Some people think the opposite, that gaming is in fact going down.

https://twitter.com/x/status/1986851382876426273

 

[Joe NYC]

Is all this really about AMD Q3 earnings?

It is, I think.

 

[inquiss]

No. The PE ratio of AMD will tank in that scenario.

What do people not get here? AMD has a 115.42 P/E ratio compared to Nvidia's 53.

Both company's stocks are equally exposed to AI. AMD because of its AI promise. Nvidia because of both AI promise and complete dominance.

Let's suppose both AMD and Nvidia drop to 15 P/E ratio. AMD would drop by 7.6x. Nvidia would only drop 3.5x.

Nvidia's non-AI business is also bigger than AMD's non-AI business.

How do you get to that 115 PE?

 

[johnsonwax]

For those who think that AI is definitely in a bubble, put your money where your mouth is and short the market/AMD/Nvidia/Google/Meta/Microsoft. They're all running on AI-fueled runs.

By saying that there is an AI bubble, you're saying that the market will be lower in the future than today. So short it. You're almost guaranteed to make money if you think you're right.

If you ask me, I would tell you that we're probably in 1995 of the dotcom and not 2000. We got a few more years of insane growth to go. And even when it pops, market valuations will still be bigger after it pops than in 2025 - like how markets were still higher in 2001 than in 1995. My opinion is actually the majority opinion because the markets are still at or near all time highs. It's the loud minority that keeps talking about the AI bubble right now.

Horace Dediu haș a great saying: "People who can predict the future we call futurists, people who can predict when it will happen we call billionaires".

You say 'just short the stock' as some idiot who learned everything they know about investing from television dramas. There's a cost to shorting a stock. You short when you think you know when something will happen, or as a hedge. I'll offer up another useful quote: "The market can remain irrational longer than you can remain solvent".

At the end of the days, real dollars need to flow in the door. They don't exist. Everyone knows they don't exist. But in every bubble the first half of the investors that bail make real money and the last half of the investors get screwed, and the public is left to pick up the pieces. A LOT of investors made bank in 2008. Almost none of them were amateurs.

 

[johnsonwax]

AMD's market valuation is driven by their revenue outlook.

You can't say that. In fact, that's never true because there is no natural P/E baseline for equities. All stocks rise and fall to some degree on broader market sentiment, sector sentiment, and so on. That doesn't mean fundamentals don't matter - they do quite a lot in most situations (looking at you GameStop) but fundamentals are never the whole story. I don't think any of the tech compute companies aren't carrying at least some AI tailwinds here - including Apple who has announced zero plans to generate revenue off of AI.

 

[johnsonwax]

Not sure why that matters.

1. I’m talking about AMD stock. Not where their revenue comes from. Their sky high PE ratio is due to Wallstreet believing that they can get a piece of the AI pie.

2. If there is a bubble and it pops, it isn’t going to reduce AI hardware demand to 0. Companies will just buy less GPUs for a while. That’s worse for AMD because the only reason people buy AMD GPUs is because they can’t get Nvidia ones due to insane demand.

Because AMD still has a relatively strong revenue story to tell out of their CPUs and other products. That puts a fundamentals driven floor under their stock price that is quite a bit higher than Nvidia's if GPU/NPU spending goes to zero. Diversification is beneficial in recovery situations and AMD has more than Nvidia.

 

[johnsonwax]

Maybe part of it but I am a bit doubtful it is the primary cause. I think it’s more that Sony and MS had more inventory than they needed for a while and dropped their orders to near 0 for a while. That inventory is now cleared and they needed more for the coming holiday season.

Retail spending is falling. More notably, a growing share of consumers are being squeezed out so the accessible market for consoles is probably shrinking. Part of the problem is the consumer debt situation is becoming really untenable as the cost to service that debt is extremely high and people are increasing rather than decreasing their reliance on that debt. Delinquincies are climbing - see everyone from Carvana to Klarna. That may hold on through the holiday season or it could crash out beforehand. MS and Sony will still place their orders but if demand falls off they'll be left holding inventory and just put orders to 0 for the subsequent quarters. Eventually the bill will come due.

US consumer sentiment is extremely low. It's not great outside the US either.

 

[adroc_thurston]

Because AMD still has a relatively strong revenue story to tell out of their CPUs and other products

Very very strong especially if they maintain their commercial PC breach (they will).

MS and Sony will still place their orders but if demand falls off they'll be left holding inventory and just put orders to 0 for the subsequent quarters. Eventually the bill will come due.

Consoles sales still look healthy though.

 

[inquiss]

Very very strong especially if they maintain their commercial PC breach (they will).

Consoles sales still look healthy though.

Really hope they maintain that breach and I'd like to see them be able to open a crack into the OEM market too. Lack of competition in that area as an intel stronghold doesn't make sense based on the products.

 

[adroc_thurston]

I'd like to see them be able to open a crack into the OEM market too

You mean like OEM DT?
Meh slow cycle there.

 

[Doug S]

If you believe that the AI bubble will pop 100% and the market size will be smaller than in November 2025 after it pops, the borrowing cost to short is trivial to the potential drawdown.

I have, first by selling down my holdings of Nvidia and a few other high flyers (still hold some but not as much as I did) and quite recently via leveraged ETF "bear" funds against the semi industry. All the recent circle jerks with Nvidia investing in companies to buy more of their products was for me a leading indicator that we're in the terminal stages of the bubble.

I didn't short Nvidia because as Buffett says the market can remain irrational longer than you can remain solvent, so actual shorting is IMHO way too risky for any typical individual investor. Buying puts is a safer way to short but if you jump in too early (or choose a time horizon too far out) the theta will kill you.

If you believe AI growth is going to continue through the rest of the decade are you buying all the AI related stocks you can on margin? Are you loading up on far out of the money calls so you can cash in when they continue to grow and blow past your strike price? You want others to put their money where their mouth is, are you doing the same?

 

[inquiss]

You mean like OEM DT?
Meh slow cycle there.

OEM DT and more consumer laptop. Laptop market share is not going up very fast. Is it even going up?

 

[adroc_thurston]

Is it even going up?

Rev share is up.
Units are for poor people.

 

[Joe NYC]

OEM DT and more consumer laptop. Laptop market share is not going up very fast. Is it even going up?

Here are some reference points for Client revenue: AMD vs. Intel
2022 Q3: $1.0 vs $8.1
2023 Q3: $1.5 vs $7.9
2024 Q3: $1.9 vs $7.3
2025 Q3: $2.8 vs $8.5 *

* The $8.5 billion is not apples to apples. It contains ~$900m to $1 billion of NEX division, so real number is probably ~7.3 - ~7.4 billion. Intel called it flat YoY, so that matches.

So, client $ market share of AMD went from:
2022 Q3: 10%
2025 Q3: 27%

 

[johnsonwax]

Very very strong especially if they maintain their commercial PC breach (they will).

They likely won't in the short term. They'll maintain their share, but if we have a market panic odds are overall spending will drop as discretionary corporate spending is frozen and consumers hold onto their cash. Things will likely to back to normal after some time, but how long we don't really know. The 2008 recovery was relatively quick because the Feds recognized the problem was a lack of liquidity and could inject enough cash to solve that immediate problem. Things would have been MUCH worse had that not happened, and had it not been enough. Competency in government matters in these situations. It also assumes someone is willing to buy the debt the US would need to raise to inject that liquidity, and well, that's sort of a problem right now.

Consoles sales still look healthy though.

Yeah, but that's being led pretty strongly by Switch, which AMD isn't a supplier for. They're beating both Xbox and PS5 combined.

 

[johnsonwax]

Are you loading up on far out of the money calls so you can cash in when they continue to grow and blow past your strike price?

I made so much money doing this with Apple. Downside - you aren't going to sleep for that year.

 

[johnsonwax]

Rev share is up.
Units are for poor people.

Yeah that last typewriter maker really cleaned up with 100% revenue share. Don't lose sight of what the measuring machine is telling you and what it's not.

 

[Fjodor2001]

Regarding who may bring nVidia down in the AI race, it may not be AMD but Google with their Ironwood TPU:

Hot Take: The True AI Chip Challenge for NVIDIA Isn’t from AMD or Intel — It’s Google’s TPUs Heating Up the Race

Hot Take: The True AI Chip Challenge for NVIDIA Isn’t from AMD or Intel — It’s Google’s TPUs Heating Up the Race

wccftech.com

If we go from a de facto monopoly to price competition things may change fast.

 

[inquiss]

Here are some reference points for Client revenue: AMD vs. Intel
2022 Q3: $1.0 vs $8.1
2023 Q3: $1.5 vs $7.9
2024 Q3: $1.9 vs $7.3
2025 Q3: $2.8 vs $8.5 *

The $8.5 billion is not apples to apples. It contains ~$900m to $1 billion of NEX division, so real number is probably ~7.3 - ~7.4 billion. Intel called it flat YoY, so that matches.

So, client $ market share of AMD went from:
2022 Q3: 10%
2025 Q3: 27%

Yes, agreed, and thank you. Rev share is up, but that's without them making any inroads into OEM desktop and making less progress in laptop than desktop. There are some walls to be broken down, as has started to be made in corporate laptop. Move volume in these areas means better relationships leading to better designs imo

 

[adroc_thurston]

Yeah that last typewriter maker really cleaned up with 100% revenue share. Don't lose sight of what the measuring machine is telling you and what it's not.

PC's ain't typewriters, fortunately enough.
It's an eternal, forever market, where you wanna high ASPs at moderate units.

Regarding who may bring nVidia down in the AI race, it may not be AMD but Google with their Ironwood TPU:

Hot Take: The True AI Chip Challenge for NVIDIA Isn’t from AMD or Intel — It’s Google’s TPUs Heating Up the Race

Hot Take: The True AI Chip Challenge for NVIDIA Isn’t from AMD or Intel — It’s Google’s TPUs Heating Up the Race

wccftech.com

If we go from a de facto monopoly to price competition things may change fast.

Meh

 

[DrMrLordX]

AMD PE ratio based on current quarter is 233 / 4.8 = 58.25
Only about 17.5% of those earnings are at risk (to AI bubble bursting)

NVidia PE ratio based on the last quarter 188 / 4.20 = 44.78
But ~90% of the earnings are at risk.

BTW, you are continuing to ignore the Earnings. Suppose both companies' stocks do drop to 15 P/E ratio, but also, their earnings drop as profits from AI disappear.

AMD earnings drop 17.5% from $4.80 annualized = $3.96. Multiplied by 15 = $59.40
NVDA earnings drop of 90% from $4.20 annualized = $0.42. Multiplied by 15 = $6.30

^^^^^^

If the AI bubble pops and AMD gets caught up in the crash in an equivalent fashion, AMD stock becomes a bargain. It will recover much sooner as traders realize AMD wasn't that exposed. I suspect though that traders mostly know this and this discussion is just a chip on a shoulder.

Thank you!

It doesn't matter what their market cap is.
The relevant stuff is on the balance sheet and there you have growth across every BU (sans embedded).

^^^^^^^^^

Is all this really about AMD Q3 earnings?

Only barely, at this point.

You can't say that. In fact, that's never true because there is no natural P/E baseline for equities. All stocks rise and fall to some degree on broader market sentiment, sector sentiment, and so on. That doesn't mean fundamentals don't matter - they do quite a lot in most situations (looking at you GameStop) but fundamentals are never the whole story. I don't think any of the tech compute companies aren't carrying at least some AI tailwinds here - including Apple who has announced zero plans to generate revenue off of AI.

The above posters summed it up quite well (overall).

 

[dr1337]

If this kind of hardware exists, it also means enterprise hardware will be magnitudes better than home hardware which means cloud AI models will be vastly more capable than what we can run at home.

True, but at the same time there is going to be diminishing returns for a lot of workloads. The current quality of cloud AI is already good enough for millions of customers as is. I don't think the cloud providers will ever vanish in the AI market, but their monopoly will go away as more memory and bandwidth becomes available to end users.