Where does that come in?

So the memory is also an important part of this whole build-out, right? So the memory chips are used initially in the training, and they're also used in inference processes.

But the content of memory has continued to increase a lot more than I think people expected.

But what's also happened is that because AI has taken up such a huge part of demand, it's made traditional commodity memory super tight.

And that's where you've seen this massive price inflation in the last few years.

It's not a situation that seems like it will resolve soon because starting next year, you know, the memory makers, they primarily only raise pricing of non-AI chips so far.

And now the profitability of the non-AI chip on memory is higher than it was in AI, which Sounds strange because AI was supposed to be a higher margin chip.

And does that explain the divergence? No.

Where does that come in?

So the memory is also an important part of this whole build-out, right? So the memory chips are used initially in the training, and they're also used in inference processes.

But the content of memory has continued to increase a lot more than I think people expected.

But what's also happened is that because AI has taken up such a huge part of demand, it's made traditional commodity memory super tight.

And that's where you've seen this massive price inflation in the last few years.

It's not a situation that seems like it will resolve soon because starting next year, the memory makers, they primarily only raise pricing of non-AI chips so far.

And now the profitability of the non-AI chip on memory is higher than it was in AI, which Sounds strange because AI was supposed to be a higher margin chip.

And does that explain the divergence? No.

So does this mean that companies like Samsung and Hynix are going to be out of business by the end of the year?

I think that's a little bit premature.

I think what we're seeing here is just that it's a perfect storm for CXMT's listing.

You're coming off basically one of the biggest memory super cycles in recent history because of AI.

And I think, you know, if you think about why are people so excited and talking about this, because...

Chinese investors have been seeing global investors participate in this memory trade for the last nine months, whether it's the Korean, you know, obviously the Korean market, you know, Micron in the U.S.

So finally you have, you know, their own version of a memory company that they can participate in, which was what I think drove this huge interest in the IPO.

Phil?

It was also discovered that the brakes on the bus were tampered with so then there might have been foul play.

yeah that's that's not surprising because so i've been looking into gabriella santana francisco herrera so you know her whole alibi was really fucked up because like she was like this tap dancer and there there was at one point a famous tap dancing instructor in goon river um and then i guess she wanted to go to la to continue her career once her training was done But then she had all these accidents herself and then went on a serial killer rampage.

It's really crazy.

So there was a secretary at Goon River Hospital who wanted to be a movie star.

Her name was Vanessa McCleary.

And she died somehow.

Somehow.

And instead, the Dollar General manager died for him.

Is that enough to move the needle here, or what do you think it actually takes for the stock to move even higher? Because it didn't really move on the results last night.

Yeah, because if you look at the split right now, the data center revenue split is now almost 50-50 between the hyperscalers and the non-hyperscalers.

So I think if we see the non-hyperscaler portion continue to rise and exceed 50%, That would be a pretty positive indicator.

The other thing I think as a second narrative is this focus on the server CPU, especially with the attention being paid this year to Gentech AI.

This has been basically an area that NVIDIA is now pushing more aggressively.

They talked about potentially 20 billion in revenues from the server CPU itself.

Now, I think the, I guess the pushback though would be $20 billion for NVIDIA is not a huge number.

What do you think is driving the increase in those rental prices? Is it just the GPU increase? crunch broadly? Is it customers saying that, hey, maybe we don't need the latest and greatest we can make do with the Hopper family or the A100s? What do you think was behind that?

So the question becomes what makes the difference?

The stereotype that people have games might be some young person alone in the basement or in their room with their controllers who are playing, right? It feels like a very antisocial, very solitary thing.

But if you actually go back to the history of games, especially I kind of grew up with the game industry, the arcade where the big tall Pac-Man machines and the old arcade machines were, you would play the game, but there would be 15, 20 people around you, surrounding you, going up and down with your emotions as you are playing and so on, or possibly dissing you [laughs] between.

I wanted to basically showcase how social game can be, and that really was in one sense the motivation of a lot of work that I do.

What Frank Lee and Erin Truesdell are building sits right at that intersection where play becomes practice, where technology becomes connection, where games become care.

People might not realize this, but my career and Erin's career just starting is a very unusual path, a hard path for academic.

Academic disciplines are very disciplinary, right? CS people doing CS stuff, digital media people doing digital media stuff.

You don't go from medicine to, to alt control to outreach and so on.

Now, what do you think of that? What do you make of the valuation broadly?

Yeah, I mean, that's a good question.

I mean, I think the thing with Intel is that it's not a simple story now because you have, besides the core chip business, you have this geopolitical aspect of trying to be an independent foundry and trying to compete with TSMC at some point.

And this has been one of the areas we've been less confident on.

And I think one of the reasons, you know, shares have reacted as negatively as it has been is that beyond just the fact that they're not able to, they still have capacity constraints, It's the foundry outlook, the independent foundry outlook.

I think the market's gotten a little bit bullish about more customer engagements, which we acknowledge.

But customer engagements don't mean actual orders yet.

Or on the flip side, do you think it's better as part of a bigger company?

Does that mean you're not concerned about that at all?

Yeah, so that's a good question.

I think the reason people start to see some concern is that if you look at the TPU project, next year, MediaTek in Taiwan will have a TPU project with Google as well.

But I think if you look at the scale, MediaTek's project is probably only going to be a couple hundred thousand units, whereas Broadcom's going to ship well over 3 million next year.

So the scale is not even similar.

But I think what's interesting to take note is that If you look at Amazon, Amazon has been doing a lot of the front-end portion of the chip themselves.

So it's a different model than what Google so far has chosen.

It seemed to be the call that changed the sentiment.

How frustrating is it to get the fundamental story right and the price action wrong?

wrong?Yeah.

No, that's a good question.

I, I think when you think about this though, it was just really about the conviction that the management needed to give people.

Uh, we had a really strong run-up into results, so I think going into these results, people were looking not only for a good number, but something tangible for analysts to be able to raise their numbers significantly after these results, and w- I don't think we got that.

Um, so I think that was why the, uh, you know, the shares didn't react so well.

And I think the other point was China as well.

Do you have a sense of how this is coming together, of what this proposal looks like?

Right.

...

the importance of AI right now, uh, and, and why it's so important to invest it in.

You know, for example, if you look at sovereign AI right now, um, you know, the announcements earlier in the year-

Right.

... about, uh, you know, Saudi Arabia are going to invest this huge amount of this.

But there's also the question of have, are they ready yet?

Would that change at all within the geopolitics? Do we just still see a dominance of international investment in Pacific Rim semiconductor and technology?

Yeah, I mean, and, I mean TSMC is 75% owned by foreigners, uh, foreign investors.

But also, and I think this past year, there's a lot of changes, because they're now spending more money and production in the US.

So they've talked about 30% of their most advanced 2-nanometer production will be in the US over the next few years.

So I think this investment process has been surprising, uh, frankly to most of us, in terms of how quickly it's moving and the pace.

Who's the next Morris Chang that HSBC sees in the Pacific Rim?

I think you're gonna start to see, um......

some sort of...

How do you guys account for that?

I think like most industries, everyone's dealing with the uncertainty around tariffs.

But I think what's different is the importance of semiconductors right now in the global economy, especially with AI.

It is such an important engine of all the AI infrastructure investment.

so i think there is going to be you know there there will be ongoing uncertainty when it comes to tariffs but i think the market has started to shrug this off right and i think you know especially you know if you talk about ai infrastructure you know a lot of it is uh you know it's been driven by enterprise so this the tariff impact is not going to be as significant because i think if you talked about consumer they'll become a bigger impact because you're going to consumers are going to feel that immediate impact of tariffs

How do you think about that?

Yeah, it's a very... complex one because you do see with the latest change for the Nvidia chip, they were banned earlier in the year and now they're actually allowed to sell in.

So it shows you, I think there's a lot of, there's still a lot of unknowns and I think both sides are still working on in terms of trying to figure out what's the right balance.

What led to a call that bold?

sure i i think the main thing is that you know compared to where we were at the beginning of the year where we got a lot more conservative um in terms of the outlook of amd's ai revenue we see some changes in the last couple months that's definitely much more constructive and i think at the beginning of the year there was very little visibility the mi325 chip that they launched in the uh in in the first quarter wasn't getting quite the feedback we were expecting but now as we see the mi355 the feedback's been a lot more positive And also, we see a big uplift in pricing, much more than we previously expected.

And we do see the hyperscalers seem to be interested.

So hence, we think there could be a big reversal in terms of the AI revenue for this year.

So we now have more than $8 billion in revenues, which is considerably higher than the consensus around $6 billion.

This reversal of fortunes when it comes to AMD, if you will, is it a read to the broader chip sector? Are there other stocks that you think also are going to be beneficiaries of this increased demand for AI outside of NVIDIA, of course?

Sure.

So you bring up a really interesting point, Frank.

I just want to build on some of the conversation we've already had so far this morning and understand from your perspective why this won't be the beat and raise quarter that maybe some people got used to in years gone by.

Yeah, I think if you look at the beaten race, we haven't really seen that for the last couple of quarters, NVIDIA, since the second half last year.

So I think what you're seeing is that, and one of the reasons I think we see is that there's still some ongoing challenges they're facing this year, which they didn't face last year.

You know, number one, you had, you know, the supply chain is still having some ongoing issues, especially on the downstream side and the way they ramp their NVR rack architecture up.

We've had the China ban come through earlier as well.

Also, I think the GTC roadmap that was launched, it wasn't as exciting as people had hoped.

We're not seeing a huge spec migration that will drive this incredible pricing power going forward.

Which side of the equation do you think is going to be the more important one this earnings season?

The rest of the street's at $175.

Yeah, so I think if you look at the initial fear of DeepSeq was that it would change the AI CapEx trajectory because of significantly lower costs.

But within a few weeks, what you've seen is the hyperscalers have not only kept their CapEx, they've actually raised it.

And on large part, a lot of them are raised by more than 20%.

So I think that settled some of the initial fears that people had about DeepSeq having an immediate impact to earnings or demand, at least through this year.

Now, in terms of why we think it might be a near-term tailwind is that We've seen in the last month or so a big change in body language, especially coming out of China.

Chinese internet companies like Alibaba have raised their CapEx quite a bit.

What channel checks are you seeing that are giving you the indication that everything seems to be on track?