Objectively, AMD had a mixed report. There are no serious red flags, per se, and the report was a bit better than Microsoft as management gave investors something to look forward to. Management sounded confident (or even adamant) there would be a H2 rebound in the data center and a Q2 rebound on PCs.
Certainly, the report wasn’t a disaster like Intel, and that comparative performance is likely helping AMD with flat price action despite there being some puts and takes.
Where the report was mixed is the weaker-than-expected data center sequential growth and lower gross margin. Investors will need to trust management has enough visibility into H2 to deliver. It’s helpful TSM echoed the same, which is a H2 rebound. However, the current information provided for next quarter is double digit deceleration sequentially in the data center due to high inventory levels.
We remain positive on AMD but want to be a good messenger on the nuances of the earnings report.
Financials:
AMD reported inline with expectations for Q4. There was a slight beat on revenue at $5.52 billion expected compared to $5.6 billion actual. This represented growth of 16% compared to growth of 14.3% expected.
The March quarter guide missed, and this is where analysts on the earnings call were primarily focused. Guidance of $5.3 billion missed by expectations of $5.52 billion, which equals growth of (10%) compared to (6.3%) expected. Please read the Earnings Call Notes below, as it required further discussion to put the pieces together. Prior to Xilinx, GAAP tracked about 1 point within the Non-GAAP margins.
The inline Q4 resulted in the FY2022 also being in line for growth of 44% and revenue of $23.6B.
GAAP Margins are not the best to focus on right now as they include the amortization of Xilinx assets and intangibles. Rather, for YoY comparison, the adjusted margins are a better indication of AMD’s business operations.
Q4 Margins:
- Adjusted Gross Margin of 51% compares to Adjusted GM of 50% in the year ago quarter and 50% in Q3
- Adjusted operating margin of 23% compares to 27% a year ago, analysts were concerned about this
- Adjusted net margin of 19.6% compares to 23% a year ago, analysts were concerned about this
FY 2022 Margins:
Same as above, the GAAP margins aren’t reflective of the business operations due to the Xilinx acquisition.
- Adjusted Gross Margin of 52% for gross profit of $12.273B
- Adjusted operating margin of 27% for operating profit of $6.345B
- Adjusted net margin of 23.3% for profit of $5.5 billion
Cash Flow:
AMD reported lower than usual cash flow in Q4:
- Operating cash flow of $567 million, for a margin of 10% compared to a margin of 17% a year ago and 17% last quarter
- Free cash flow of $443 million, for a margin of 7.90% compared to FCF of 15% a year ago and FCF of 15% last quarter
For Fiscal Year 2022, cash flow:
- Operating cash flow of $3.6 billion for a margin of 15.2%
- Free cash flow of $3.1 billion for a margin of 13%
The company repurchased $250M shares this quarter. There is $2.3 billion in cash on the balance sheet and $330 million in debt.
Revenue Segments:
The data center was up 42% YoY for $1.7B in revenue. This is $1 billion higher than the previous quarter for 6% QoQ growth. As stated above, the issue is the miss on Q1. It was vague in the quarterly report yet was indicated on the call that data center would be down “double digits” sequentially due to high inventory levels. However, management had positive things to say about H2 and the data center.
Client revenue of $903 million was down (51%) YoY and down (9.7%) sequentially. It was indicated on the call that client revenue would be down single digits in Q1 and this would mark the bottom, according to management. There was a reported operating income loss of ($152) million.
Gaming revenue of $1.6 billion was down (7%) YoY and flat sequentially.
Embedded revenue of $1.4 billion was up 1,868% YoY due to the Xilinx acquisition and was up 7.6% QoQ.
Earnings Call:
The first hint of some data center weakness in Q1 was in this information provided by the new CFO, Jean Hu:
“Year-over-year Data Center and Embedded segment revenue are expected to grow, offset by lower Client and Gaming segment revenue. Sequentially, Embedded segment revenue is expected to increase. Client and Gaming segment revenue are expected to decline largely consistent with seasonality. Data Center segment revenue is expected to decline due to elevated levels of inventory with some cloud customers.”
Despite not guiding for full year 2023, the CFO added:
“Directionally, we expect Embedded and Data Center annual revenue to grow from 2022 based on the strength of our product portfolio and expected share gains. In addition, we expect Client and the Gaming segment revenue to decline based on the current demand environment.”
This was followed by a question later on by Ross Seymore:
“So just trying to get the magnitude of just how much Data Center has to drop to make that outcome on the mix side be true.”
Lisa Su:
“Sure, Ross. So let's see. We said the Client and Gaming segments would be seasonal. So you would expect that the Data Center would be more than seasonal. So maybe to help you size that, think about the Data Center sequential drop as double digit, whereas the Client and the Gaming segments are more like single digit, if that helps.”
The bulk of the call was dedicated to dissecting the data center segment with some additional questions on gross margin and the Client segment/PCs.
Right out the gate, an analyst asked what is on everyone’s mind:
“But I've gotten about a zillion versions of the same question tonight, which was do you think the company can grow for the year 2023 overall? And if you could just kind of walk us through the drivers of the business as we work through the year? Thanks.”
The CEO answered with the following, indicating that data center would be a growth driver with emebedded:
“As we mentioned in the prepared remarks, coming off of a very strong 2022, there is some inventory at some of the cloud customers. And so, we are expecting a softer first half and then a stronger second half, but we feel very good about our market share position and opportunity to grow with Data Center.
Also on the embedded side, I would say we have a very strong portfolio there. The Xilinx business has done very well in 2022. It's a diversified set of markets. We see strength in a number of the end markets. And so, we think that's also a grower for AMD.
On the other side, our Client and Gaming businesses, we believe, will decline. We have made good progress. When we look at the PC markets in the second half of the year of 2022, we were really trying to rebalance inventory.”
That was the first of many times management clarified that the data center would grow in H2. Here were a few other times:
“Our expectation is that sort of the first half softness for cloud and then second half strength as that's worked through. But like I said, it's different for each customer. And then in terms of overall growth, as I said, we're very bullish on the overall growth of our Data Center business and the opportunity to gain share as we go through the year.”
Discussion on Q1 Being the Bottom for PCs:
Vivek Arya:
“But when we look at the shipments, right, from you and your competitor, they could be down as much as 40% or 50%, right, year-on-year in Q1. So do you think there's a possibility that the TAM assumption of just down 10% could be an optimistic one?”
Lisa Su:
“I think second quarter – first quarter should be the bottom for us in PCs. We – and then grow from there into the second quarter and then into the second half. And I should note also, Vivek, I mean, we just launched our Ryzen 7000 Series with sort of our AI capabilities, both from a notebook and desktop standpoint.”
There was more reiteration later by Lisa Su but she was hesitant to say if the H2 strength will result in YoY growth.
“I think our Data Center grow – growth in the second half versus first half, we expect that to be significantly stronger. As it relates to clients, we would also expect it to be stronger. Again, depending a bit on macro and sort of how the TAM actually evolves.
I think for the Embedded businesses, I would say that we expect to grow over the full year 2023 versus 2022. What we see right now is a fairly strong backlog and good visibility into the first half of the year. I'm not ready to say that Embedded will grow in the second half versus the first half, though, because we're coming off very strong growth already. And so I think those are the puts and takes.”
It was interesting to hear that AMD took market share on PCs as the narrative has been that Intel did due to aggressive pricing
Lisa Su:
“I would say that in general, the PC market share numbers are probably a bit noisy right now, just given all of the sell-in, sell-through and the inventory dynamics that are being worked through. Actually, in the fourth quarter, we believe we gained a little bit of share in the PC market.”
There was (yet) another question on PCs:
Mark Lapacis:
“And Lisa, correct me if I am wrong, I thought I heard you say in an answer to an earlier question that you expect the PC client, but just to grow into second quarter. So is that suggest that 1Q, you think is the bottom on the PC? And then I had a follow-up? Thank you.”
Lisa Su:
“We do believe the first quarter is the bottom for our PC market – for our PC business, and we'll see some growth in the second quarter and then a seasonally higher second half.”
Discussion on Adjusted Gross Margin Guide:
Management provided guidance of 50% on the adjusted gross margin which is 3% lower than the year ago quarter. Due to data center being down sequentially, the Q1 margin is expected to be lower than usual.
Matt Ramsay:
“But I kind of wanted to focus on the drivers of the longer-term margin that's down, I guess, three or four points from where you were a few quarters ago despite more mix of the revenue coming from Embedded and Data Center?”
Lisa Su:
“In terms of the sequential question that you had from Q4 to Q1, that's just a product of the mix. So with Data Center being lower sequentially that – that's that. We are also working through our client inventory clearing [..] And so as Jean said in the prepared remarks, we would expect margin expansion as we go into the second half with the growth in Data Center, Embedded and some normalization of the client business as well.”
Stacy Rasgon:
“Can you give us any idea like first half to second half? Or I mean just for the full year, do you think gross margins grow year-over-year from the 52% that you printed in 2022?”
Jean Hu:
“The major headwind we are facing is really Client side, which if you think about the gross margin in the first half of 2022 versus the first half of 2023, the major impact is from the client revenue, inventory correction, which impact the gross margin in the Client segment […] But overall, we feel pretty good. Once we normalize the Client segment, our gross margin will continue to expand.”
Conclusion:
We plan to see it through with AMD. I had just written in the Tesla write-up that we won’t get a perfect ER from any company this quarter, so investors will need to subjectively determine what they are willing to hold through and what will cause them to move to the sidelines.
We don’t have to wait too long as the Q2 guide on PCs should potentially clear some of the cloudy skies. From there, we will know from not only AMD but also from TSM, NVDA and others if the much-anticipated H2 rebound will be on time. It makes sense because the comps were low in Q3 from the initial PC miss and same for NVDA’s crypto miss.
If a picture is worth a thousand words, then this is why we plan to see it through with AMD. Notably, other semi industry analysts believe AMD’s true market share is in the mid-20s. We covered this here.
Source: Reuters
Notably, I had stated on the forum in the pre-ER report for AMD that I’m not expecting much from the company for Q1. We are looking at Q2 and beyond.
Look for Meta’s comments tomorrow on capex to potentially hurt AMD and NVDA stock if the rumors are true about a $2B pullback on capex spending. We covered this on the forum here. We aren’t too concerned as there’s only one path to build out the AI economy – which is data centers and eventually edge microdata centers. If not Meta, then other Big Tech players will step up. Meta is simply trying to keep up as there are many hyperscaler customers to consider.