Discussion AMD 2022-Q4 Financial Results

[Vattila]

After Intel released their dire earnings and forecast this week, AMD again passed Intel in market capitalisation (currently $122B vs $116B). It will be interesting to see if AMD can retain their higher valuation this time around, which very much depends on their own financial results in the current market turmoil. AMD will report quarterly earnings after market close on Tuesday 2023-01-31.

AMD has forcasted revenue to hit $5.50B ± 0.30B. Zacks Estimates on average expects $5.51B, with $0.66 EPS, with a revenue forecast for next quarter slightly up at $5.57B, presumably based on AMD's competitive position and strong growth in the data centre segment, in particular, outweighing the weakness in the client segments.

Do you think AMD will miss, hit or beat their revenue target? Will they forecast revenue up, flat or down for the current quarter?

PS. While we all wait for AMD's results, here is a nice video by CNBC on AMD's history and comeback:

 

[DAPUNISHER]

I would like to see AMD making deliberate attempt to grow market share, even at the cost of margin. Profit will follow naturally and more easily. They're doing it on server, they should do it on desktop, too.

Now that they are adding iGPU to the full line up, OEMs will be able to use them in more configs. Previously only offering a weak APU stack, lacking the threads, cache, clock speeds, and even PCIe version, of the best Ryzen CPUs, while needing a dGPU, was a serious weakness v. Intel.

 

[Mopetar]

No, there are plenty of existing leases that haven't been drilled yet - some may NEVER be drilled. Some because the owner of the lease can't afford to, some because the owner doesn't want to spend the money (i.e. the shareholders want returns, not re-investing every dollar they make) some because the location of the lease isn't so great (i.e. they would have to run too much pipe to get it to a main line to take it market or use tankers both of which increase the break-even price) or some wells may have more gas yield than they are allowed to flare.

Your "point" is wrong, oil production dropping had EVERYTHING to do with covid.

It's a bit of both. The reason for new leases is to drill and produce oil that could theoretically be done at lower cost because no sane company wants to produce anything that has to be sold at a loss.

The rapid price spike naturally led to more expensive sources being turned off so that supply could meet demand. The problem was that on the rebound the supply couldn't ramp up quickly enough to deal with the demand and no one wanting to gamble on the prices staying high enough to justify the capital expenditure of developing some of those leases.

Allowing new leases would see production shift to the most efficiently produced oil and lower costs to some degree. The U.S. government could have theoretically tried to stabilize prices by buying up and holding oil to release to smooth out the downward and upward swings, but I don't think they have enough free storage to be capable of something like this given the amount of oil that we use. Though they'd be damn foolish not to take as much as they could for the privilege of being paid to do so.

 

[Doug S]

It's a bit of both. The reason for new leases is to drill and produce oil that could theoretically be done at lower cost because no sane company wants to produce anything that has to be sold at a loss.

The rapid price spike naturally led to more expensive sources being turned off so that supply could meet demand. The problem was that on the rebound the supply couldn't ramp up quickly enough to deal with the demand and no one wanting to gamble on the prices staying high enough to justify the capital expenditure of developing some of those leases.

Allowing new leases would see production shift to the most efficiently produced oil and lower costs to some degree. The U.S. government could have theoretically tried to stabilize prices by buying up and holding oil to release to smooth out the downward and upward swings, but I don't think they have enough free storage to be capable of something like this given the amount of oil that we use. Though they'd be damn foolish not to take as much as they could for the privilege of being paid to do so.

That's true, there is a chance newer leases might be cheaper to develop but that doesn't change the fact there are a lot of leases being sat on and not developed. Maybe there should be some sort of time limit for a "use it or lose it" so if company X won't develop it maybe company Y will. Maybe someone else might be able to drill a given lease more economically (based on different capabilities, or already having infrastructure in the area for other wells) but they won't get the chance so long as one company can basically sit on it forever.

The government does stabilize prices like you suggest, after a fashion, using the strategic petroleum reserve. It would have been nice to purchase oil at -$38/bbl during covid if there had been a lot of slack capacity in the reserve at the time, but I think they only do fills/draws on contract not at spot. There was a lot of political fighting about drawing down the reserve last summer when prices are high, but I believe they began adding back to it in December. Given the futures prices at the time they began the drawdown and when they started filling it back up, they will make a tidy profit though obviously that's not the goal.

If they created a much larger reserve (say 10x larger or about two months' worth of worldwide usage / year's worth of US usage) intended solely for more 'economic' strategy like how it was used last year they could market arbitrage and smooth out the seasonal pricing swings plus have the capability to be the "buyer of last resort" on the spot market to avoid scenarios where oil prices go so low (let alone negative) where a lot of domestic producers feel they have to shut their wells. Of course it wouldn't be very politically popular for the US government either domestically or abroad to be propping up oil prices even if that's part of a trade that also limits how high they can get! It would have to be run independent of the administration in power at the time like the Fed, not that the Fed is immune to political conspiracy theories...

 

[scineram]

Oh look, Sony PS was 16% of AMD revenue in 2022.

https://twitter.com/x/status/1630600116339081217

 

[Markfw]

Semi-custom may decline in 2023, but server will stay on the rise indefinitely, to more than make up for it.

 

[KompuKare]

So given that 16% figure and Sony's boastful way of quoting sales numbers, do we have enough info there to infer how much Sony pay per PS5?

Hm,

Sony’s PS5 just had its biggest quarter of sales since release

It’s supply shortages might finally be over.

www.theverge.com

Sony will have to continue selling PlayStation 5 consoles at a similar rate if it wants to hit its target of 18 million consoles sold by the end of its 2022 financial year,

quoting this Reuters article:

Sony aims to sell 18 million PS5 consoles this year amid China lockdown risk

Sony Group Corp on Tuesday said it planned to sell 18 million PlayStation 5 consoles this business year as robust game sales helped it more than double fourth-quarter operating profit.

www.reuters.com

But financial year is a calendar year.
AMD had $23.6 billion revenue in 2022 (again their year not a calendar year).
So 16% of $23.6 billion is about $3.78 billion.
Divided by 18,000,000 is $210 per console?
Seems far higher than any previous figures I tried to get.
PS5's chip is about 308mm² which at $6,000 per wafer (7nm must be cheaper by now, right?) comes to $45. Not bad if they are actually getting for $200 it.

Yes, I'd rather the dGPU market was flooded with millions Navi22 or Navi21 but those are far better margins than I've ever heard for consoles.

 

[gdansk]

Sony planned to make about 1 million PS4 (Pro?) in 2022 as well. So need to deduct some for that.
And I expect AMD charged them a many million $ fee in 2022 for the 6nm shrink so it makes calculation difficult.
All told I think it should be cheaper than $200.

But in any case it is good business for AMD. And I do not suspect Sony has a problem with the arrangement except lack of capacity which would impact everyone except Intel.

 

[Ajay]

Semi-custom may decline in 2023, but server will stay on the rise indefinitely, to more than make up for it.

Well, there is no way to know that. There is no law that prevents AMD from stumbling in 4 or 5 years; or for some new disruptive technology to take away sales.
I think the growth rate of server sales by AMD still appears to be slower than I would have expected. Tough biz.

 

[Tigerick]

Well, there is no way to know that. There is no law that prevents AMD from stumbling in 4 or 5 years; or for some new disruptive technology to take away sales.
I think the growth rate of server sales by AMD still appears to be slower than I would have expected. Tough biz.

I expect AMD's data center revenue will surpass Intel by 2025 if AMD would launch Turin@192C and Bergamo 5c@256C next year.

 

[scineram]

They won't.

 

[jpiniero]

I looked at AMD's 2021 financial statement and there is this:

Two customers, A and B, accounted for 14% and 11%, respectively, of our consolidated net revenue for the year ended December 25, 2021. Sales to Customer
A consisted of products from our Enterprise, Embedded and Semi-Custom segment, and sales to Customer B consisted of products from our Computing and
Graphics segment. A loss of either of these customers would have a material adverse effect on our business.

Customer A is probably Sony. So it's not exactly new that Sony is a big chunk of their revenue. Dunno why they don't call them out by name. B is probably HP.

 

[scineram]

Well in 2022 they also had Xilinx and Epyc growth. So Sony increasing is quite significant.

 

[moinmoin]

Dunno why they don't call them out by name.

Same reason why console revenue always got obfuscated with some other unrelated businesses: To make it harder for competitors in that area to deduce exact numbers of each other.