Nvidia and AMD: AI Updates from Taiwan Semiconductor’s (NASDAQ:NVDA) Q2 Earnings

Taiwan Semiconductor Manufacturing Company Limited (TSM) (“TSMC”) reported yesterday (July 20, 2023) its second quarter 2023 earnings. TSMC’s revenue reduction guidance for the year sent to Nvidia Corporation (NASDAQ:NVDA) and Advanced Micro Devices, Inc. (Nasdaq:AMD) the shares collapse. However, TSMC’s commentary on the future of artificial intelligence (“AI”) chips has not received the same attention. When thorough, TSMC’s commentary on AI is good for long-term investors in Nvidia and AMD. I explain the reasons for this assessment below.

TSMC agrees with Nvidia and AMD’s AI growth forecasts

In a recent article titled “Nvidia and AMD Paint the Same Growth Picture,” I noted:

[AMD CEO Lisa] Su’s claim that the AI ​​chip market will surpass $150 billion by 2027 is in line with Nvidia Corporation CEO Jensen Huang’s claims during Nvidia’s latest earnings of nearly trillion dollars of data center infrastructure. built or replaced in the next few years.

It seems that TSMC broadly agrees with the predictions of AMD and Nvidia. During the second quarter earnings call, CEO CC Wei said:

Today, demand for server AI processors, which we define as AI CPUs, GPUs, and accelerators that perform training and inference functions, accounts for approximately 6 percent of TSMC’s total revenue. We anticipated this to grow nearly 50% CAGR over the next 5 years and grow to a low teen percentage of our revenue.

While these are different metrics, TSMC’s prediction of a 50% CAGR over the next 5 years aligns well with Lisa Su’s prediction of the data center AI accelerator’s total addressable market (TAM) growing at “a compound annual growth rate of more than 50% to over $150 billion in 2027.”

TSMC’s comment further strengthens the bullish case for AMD and Nvidia. The kind of growth talked about here would go a long way toward justifying their high valuations (especially Nvidia’s) and leave room for upside beyond that.

AI chips will remain in short supply

Here’s what Wei had to say about the available capacity for AI chips:

For AI, right now, we see very strong demand, yes. For the front end part, we have no problem to support. But for the backend, the advanced packaging side, especially for CoWoS [Chip on Wafer on Substrate], we have a very small capacity to — very difficult to meet 100% of what the customer needs. So we’re working with customers on a short-term basis to help them meet demand, but we’re ramping up our capacity as quickly as possible. And we expect those tightening to be released in the next year, probably towards the end of next year. But in the meantime, we’re still working closely with our customers to support their growth.

He further noted that the increase in CoWoS capacity would be approximately 2x.

Furthermore, there is some possibility that TSMC could still be short on CoWoS capabilities beyond 2024, with Wei stating:

We are planning our CoWoS capacity, although probably not enough yet, but we are working very hard to increase it. Overcapacity, today’s not a concern, today’s concern is not enough capacity to support all of the very strong demand.

As readers can see, TSMC expects that some of the advanced bundles (CoWoS) involved with AI chips will likely be in limited supply until the end of 2024. Therefore, demand is likely to continue to outstrip supply for quite some time because AMD and Nvidia won’t be able to expand production beyond a certain point even if some N4 and N5 capacity is available.

While Nvidia and AMD may not be able to sell as many chips as they would like, they may both be able to sell as many as they can produce for different quarters. That’s already true for Nvidia, and it may prove to be true for AMD’s upcoming MI300 chips as well, simply because Nvidia won’t be able to expand beyond a certain point. AMD’s chips could therefore do very well by virtue of sheer availability.

The tight supply should also mean that the extremely high margins charged for AI chips are likely to continue for more quarters.

Overall, while Nvidia and AMD investors may be disappointed by capacity constraints, they should also be pleased with TSMC’s confirmation of so much demand for AI chips that it has overwhelmed its advanced packaging capacity for at least a few quarters.

The world of computing is evolving

Finally, the revelation that AI chips currently account for 6% of TSMC’s total revenue is useful enough to gain perspective on the speed and scale of the AI ​​revolution.

6% of TSMC’s second-quarter revenue was approximately $1 billion. The vast majority of this should be attributable to Nvidia. Nvidia led a sequential increase of just under $4 billion in AI chip sales this quarter and previously sold some AI chips as well. Given the exorbitant margins on these AI chips, it makes sense that Nvidia is paying around $1 billion to produce them.

Now, while 6% of TSMC’s revenue may not sound like a lot, it’s actually quite a lot. First, we should note that advanced nodes (N7 and below) account for only about half of TSMC’s revenues. This means that AI chips have already become about 12% one-eighth of all TSMC’s advanced node revenues (which include both smartphones and high-performance computing). And this happened despite the limitations of capabilities and, of course, still very early in the AI ​​revolution.

At least in my eyes, the fact that one-eighth of TSMC’s advanced node capacity is now distributed towards AI chips is a strong sign that the computing world is evolving very rapidly towards AI. A strong bullish indicator for Nvidia and AMD.

Conclusion

While investors may be a little disappointed by capacity constraints that are likely to limit growth for Nvidia and AMD in the very near term, Taiwan Semiconductor has painted a very positive picture regarding the future of AI. Long-term investors should eventually see TSMC’s AI updates as another credible data point signaling an extremely strong future outlook for Nvidia and AMD.