Now time on his industry. Giant fox Con is holding its annual tech Day in Taipei. The company, known formally as hot High, is the largest contract electronics maker in the world, and now it's pushing to diversify and build more future proof income streams. That's the hope at least our tech report. Annabel Dreolism joins us from the event. Annabel, thanks Tom.
Yeah, this is a really big day for Ho Hi. It's the fiftime I've ever held its tech day. It's also the fiftieth anniversary this year of the company. And is you said, it's that focus on what are the future growth drivers for the business, and there's no question today, yes there's the focus on AIS but rather evs, but AI service is really what's grabbing a.
Lot of investor attention as well.
And I'm very pleased to say that we're joined by the chairman, Young Lou this afternoon, and thank you very much for joining us. And as we said, is really so much attention is on the AI service and the market.
Share that you're gaining in this area.
In particular, you just said earlier on stage that sales of AI service made with Blackwell ships are doing better than expected.
Can you share more details about what you meant by that?
Yeah, the black Well demand very very strong, as Jameson College is crazy. The demand is crazy, and we're building up the capacity, and the capacity, as we mentioned, is probably very very big, much bigger than we thought so so far. The shipping day will be sometime by the end of the year.
So you're planning to start the shipments of servers and the demand is better than you expected. What were you expecting and what are you actually seeing? Can you quantify it for us?
Okay, because there were a lot of saying about, you know, the compute. The need for the compute may be saturated pretty soon, but it looks like that the demand is still growing. So that's you know, that's beyond our expectation.
Again, if it's beyond your expectation, what was your expectation? How many, for instance, were you planning to ship by the end of the year. How many you actually looking to ship instead?
Yeah, if you talk about our shipping plan, I would say we planned to ship in the beginning of Q four, but due to certain peacup with the super chip, so it's moving later to the later part of the Q four from that angle, it's worse than what we expected, but the demand is much bigger than our expect.
So the shipman you expected to come earlier in the fourth quarter, it's been delayed to the fourth quarter. The demand that you're seeing is actually better than expected. Again, can you share more quantity, like how much of the demand are you trying to capture here?
How much of the market?
But that I really cannot give you the quantity, exact quantity because we just started to build it, Okay, but we see the demand is over what we you know, we can shift.
So if you can't meet the demand right now, how long do.
You see that extending?
For?
What sort of sales do you have.
Booked going into next year as well?
We see in yeah, maybe somewhere around quanity thousand, that kind of a number.
You're looking to ship twenty thousand AI service maybe with black Well ships over the course of next year, is that right right?
Okay?
And in any particular quarter that you're looking at in particular for that or do you see.
It being that's a year per quarter?
Are you expecting a bulk of those to go on one particular quarter or do you see it spread evenly across the course of this year or next year.
Rather, it really depends on once we start shipping the demand, whether the demand will will change or not. If the demand doesn't change after we start shipping and the demand or maybe it will be even higher than we.
Expected, well, what's what sort of capacity?
Then? How much can you ramp up production?
Do you think how many of these service could you be providing to the market?
Do you think, Hm, okay, but that that.
Really cannot give you the number because the facility is there, but the equipment is being put there, and also the power needed to build those servers. You know, they require a lot of power to do that, so that's also been prepared. So the capacity of that, I really can't give you that number from my head. I had to go and check it out, right, I don't want to give you the Ryan inspectation.
The demand that we see from hyperscalers, there has been that big question sort of hanging over AI for the course of the year. Would we start to see any concerns around that fading? Do you see any of the customers that you speak with them.
I'm speaking here.
About the hyperscalers that are looking like they maybe need to pause on investments for a period to wait and see if they if they actually do see ROI.
We haven't seen any of that. We haven't seen any of that concern. But you've probably heard some analysts talking about that concern, but from the market, we don't see. We haven't heard any concern like that.
And video has very tight control over the designs of its products. How closely do you work with the company to add your own value to the designs as well your own expertise?
I'm thinking energy for instance, is one area in the calling.
Technology mentioned in the past that fast Come is most integrated vertically integrated the supplier to the server industry. If you don't account TPU, CPU and some minor ic components, the component the percentage that we are able to cover by ourselves is anywhere between eighty to ninety percent. That tells you how vertically integrated that we are capable of.
When you add that sort of value to the company or to Nvidio's designs, how much value do you extract as a business and how much can you differentiate yourself as opposed to other AI server resemblance and.
The differentiation in terms of the vertical integration capability, Yes, we are one of the best. But in terms of what added value to the overall.
The over.
Server, I think it really depends on how much uh Mmediate is pricing their CPUs and gp us. That definitely will impact our gross popular margin.
In terms of proper margin.
Then what I understand for the AI service, it can be in the load teens. Do you see a way to expand that and too increases.
I think in the low teams you're probably right, but that will have to be based on, you know, how much vertical integrate, how much components that you're able to provide. So with that it varies to a certain degree. But because the CPU and GPU is very high priced, so the rest of the server, you know, has only some limited impact in terms of a cord overall.
Right, Chairman, stay with us for a moment.
Tom will hand back to you now for this portion of the interview.
But Chairman, if I cannot continue, to ask you some further.
Questions, and this is around the shift to the smartphone and space in particular.
Yes, okay, I want to talk more about the Thank you very much,