AMD Q2 2023 Earnings Preview: Some Themes To Watch

Advanced Micro Devices (NASDAQ:AMD) reports Q2 2023 earnings today (August 1, 2023) after close. This is AMD's first earnings release since Nvidia Corporation (NVDA) provided its enormous guidance for the current quarter in May. In the meantime, there have been significant updates from Taiwan Semiconductor Manufacturing Company Limited (TSM, "TSMC") and Intel (INTC) about the expected impact of the AI revolution on their businesses.

Although AMD's main AI offering—the upcoming MI300 series—doesn't debut until Q4, AMD seems likely to provide updates about its outlook, as its peers have. AMD investors should therefore pay close attention to this earnings release, and particularly to the details of AMD's various businesses as the company makes AI its "#1 strategic priority" (as per AMD CEO Lisa Su during the last earnings call).

In this article, I outline some themes that readers should pay close attention to in order to get a better understanding of AMD's future outlook, particularly in response to the AI revolution.

Data Center Revenue Timeline

During the company's last earnings call, AMD had guided 50% data center revenue growth for H2 as compared to H1. This guidance was seemingly maintained by AMD's Senior VP & GM of Server Business Unit Dan McNamara at the Bank of America 2023 Global Technology Conference.

A couple of points bear attention with the data center CPU business. First, during its Q2 earnings call last week, Intel CEO Pat Gelsinger had suggested that AI chips may cannibalize data center CPU sales in Q3 (and perhaps beyond):

We do think that the next quarter, at least, will show some softness. There's some inventory burn that we're still working through. We do see that big cloud customers, in particular, have put a lot of energy into building out their high-end AI training environments. And that is putting more of their budgets focused or prioritized into the AI portion of their build-out.

TSMC CEO C. C. Wei had also provided similar commentary during the company's most recent earnings call.

It will be interesting to see if AMD concurs with their assessments, and if it maintains its guidance for H2. There is some chance that cannibalization from AI chips may prove to be an unexpected headwind for AMD's data center CPU business. However, it is worth noting that AMD's previous guidance for H2 should be based, in large part, on the ongoing ramp of its Zen 4 chips. It is possible that these may be less likely to be cannibalized than less performant, less efficient alternatives (including ones by Intel but also AMD's Zen 3 offerings). Investors should pay attention to how the cannibalization issue is playing out for AMD.

I have discussed the (potential) cannibalization issue in more detail elsewhere: "AMD And Intel: Data Center CPU Update From Taiwan Semiconductor's Q2 Earnings," and "Intel Q2 2023 Earnings: Key Updates About The Impact Of AI."

Second, investors should keep an eye on how the data center revenue ramp looks between Q3 and Q4. We know revenue is supposed to grow 50% half-over-half, but paying attention to the sequential growth and acceleration (or otherwise) between Q3 and Q4 should give investors a better sense of whether the rapid data center growth that AMD has guided for H2 is likely to continue into next year. Will data center CPU growth plateau as Zen 4 ramps, or is the growth likely to be sustained for multiple quarters? Looking at the revenue recognition between Q3 and Q4 may provide some hints, and perhaps we will get some commentary from management on the subject as well.

Advanced Packaging Capacity For The MI300 Series

AMD's MI300 series of AI chips is just around the corner. The chips are expected to launch in Q4 with the MI300A (which combines CPU, GPU, and memory on the same package). The MI300X (the GPU-only variant) is expected to launch in Q1 2024. Other variants have been rumored, but they have not been confirmed by AMD so far.

Investors should pay attention to AMD's ability to supply MI300 chips in significant volumes. During TSMC's recent earnings report, it was disclosed that advanced packaging capacity is the main supply constraint for AI accelerators. Specifically, there is a shortage of CoWoS (chip-on-wafer-on-substrate) capacity, and although TSMC is working to roughly double CoWoS capacity through late 2024, supply is likely to remain tight during this period (and possibly beyond). I have written in more detail about this in my article "Nvidia And AMD: AI Updates From Taiwan Semiconductor's Q2 Earnings."

It seems likely that the issue of CoWoS capacity for MI300 chips will be raised during the earnings call. How aggressive had AMD been when it booked advanced packaging capacity for these chips—was AMD expecting to test the waters or to sell these chips hand over fist? Should unexpectedly high demand materialize for the MI300 series, to what extent would AMD be able to expand production beyond their plans, perhaps by switching packaging capacity from other offerings? The answers to these questions are not clear, and they could determine how well AMD does in the AI accelerator space over the next few quarters. Investors should, therefore, pay very close attention.

Ryzen AI

Earlier this year, AMD launched its first Ryzen AI chips—which bring built-in AI acceleration to its Ryzen series of client CPUs. The launch of relatively quiet and there are only a few SKUs so far, as AMD seems to be testing the waters for demand for CPU-based AI acceleration.

Compared to this relatively measured stance, Intel came out with some extremely bullish commentary for AI acceleration in CPUs during its recent earnings call. CEO Pat Gelsinger called AI "a critical inflection point for the PC market over the coming years that will rival the importance of Centrino and Wi-Fi in the early 2000s." In other words, Intel expects client-side AI acceleration to be a killer feature that drives a new upgrade cycle in client computing.

It should be interesting to get more details about AMD's long-term forecast for the client market, and if it aligns with Intel's very bullish commentary. It is worth noting that AMD's Q1 client computing revenues of $0.7 billion were down 65% year-over-year so there is quite a lot of room for AMD's client revenues to increase over the coming quarters, particularly if Ryzen AI chips prove popular. An acceleration in demand, along with improvements in inventory levels, could boost AMD's outlook considerably.

It should also be of interest to see if AMD provides any new details about engagements with Microsoft and OEMs in terms of bringing Ryzen AI based hardware and software applications to market.

Conclusion

After Nvidia's bombshell earnings in May, investors have been wondering if the AI revolution will be similarly beneficial for AMD. AMD's position is a bit different from Nvidia's, so it will take some time for a clear picture (positive or negative) to emerge. In the meantime, paying close attention to the above themes should help investors gain better insight into the potential impact of the AI revolution on AMD's future prospects.