TSM Earnings Support AI Demand Accelerating, CapEx Weighs on Stock Outlook

Sam (0:00)

We're back on Morning Trade Live and focusing on TSMC's chart. This is the one year it really tells the story. I mean, obviously we saw a massive climb throughout the course of 2026, but we have come off the highs, particularly as semis have really borne the brunt of some of the heavy selling pressure we've seen, particularly in recent sessions. Still though, we have got this name up almost 40% year to date, over 40% for the year right now. We are trading down 3%, even off the back of record profits here. So let's dig into the numbers and go inside and out on TSMC and its earnings reports. So joining us now to discuss that is Derek Yan, Senior Investment Strategist over at KraneShares. Derek, good morning to you. Great numbers. Just walk us through some of the highlights here.

Derek Yan (0:47)

Yeah, thank you for having me. I mean, it's a clear beat. They have guided Q4 earnings, Q4 revenue growth, like nearly 60% year over year.

That's very strong numbers, very impressive. Q3 revenue growth as well. More importantly, they have a very strong confidence on the capex going to 2027 We're going to see that in January meeting.

One thing I think that's really a validation is they send people on the ground to check the progress of the data centers, the energy supply, and that give them a lot of confidence for that capex going forward. I think all the signals from the earnings call is really, AI is not slowing down. AI still is more demand than supply.

Sam (1:38)

Well, that's exactly right. We heard from ASML yesterday who counts TSMC is obviously a major customer. I mean, it's saying that it's getting requests for orders out into 2030 I mean, these companies have the visibility right now. We know the demand is there. Why the sell-off?

Derek Yan (1:54)

I mean, this is where macro signal and market sentiment is really deviated from the fundamentals.

Talking about TSMC, they're a very disciplined capital allocator. For them to adding new FAP, for them to adding new capacity, requires real contract and commitment from their clients. If we have this very strong fundamental signal, but on the bond market, on the credit market, I think investors are seeing a different picture where in a macro, you have potentially higher expectation on the rates, and also on the private credit market, I think investors increasingly worry about finding those AI data centers to build out. So that's more sentiment, that's more I think short-term compared to I think 2020-30. 2030 is come like visibility, that's more long-term build out.

Sam (2:53)

So they've got the demand, they've got the pricing power. I mean, with operating margins at 58.7 percent, do you think they could potentially charge more? Because it was interesting that CC Way came out earlier in the year and said that there probably wouldn't be appetite to do that, even though the memory folks are. But we have certainly seen obviously a bit of a change in tune as far as upping some of those prices from the reporting.

Derek Yan (3:20)

Yeah. In July, actually, the CC Way is really saying that they want to be a long-term partner. But I think in this running call, they also mentioned that the profitability is there's no sitting for that.

So we're going to see in January where they have seen margins is pretty good, or even they have two nanometer ramp up, and also they're adding more facilities in the US, those are going to eat margin. But with all that, the margin numbers come out very solid. So I think eventually, they're going to be very steady and they're going to work with like NVIDIA and everybody else to figure out the best pricing going forward in 2027

Sam (4:07)

What about a name like TSMC? I mean, if CapEx is raised again, moving into the new year, is there a concern for the accessibility or at least the cost of capital for a name like that?

Derek Yan (4:21)

Yeah, I think cost of capital is really focused on the Neo Cloud because those names are using leverage, right? Even though they're using leverage, those are backed by a contract. So in terms of the hyperscaler, they have a worry about their investment grade, right? So they have an abundant access to the capital as well. So I'm not going to worry a lot about overall capital access, I think at this stage is more marginal, worry on some of the Neo Clouds. But going back, I think currently, the capital has become like a bottleneck for AI buildout, but the market reaction is really oversold. The market oversold a lot of names that's really in the infrastructure place that's still in the supply constraint.