Semiconductor Sector

Topic 122 news 4399 Subscribers

Why AMD Is Buying Server Maker ZT Systems

Show Transcript
Bloomberg 20:39 · 10.8k Views Trending

Advanced Micro Devices agreed to buy server maker ZT Systems in a cash and stock transaction valued at $4.9 billion, adding data center technology that will bolster its efforts to challenge Nvidia.

Disclaimer: This content is for informational and educational purposes only and does not constitute a recommendation or endorsement of any specific investment or investment strategy. Read more

Transcript

  • 00:00 This is basically data centers, data centers, data centers, which the last time I checked
  • 00:04 appears to be the hottest game in town.
  • 00:05 What is AMD going for here?
  • 00:07 Yeah, you bet.
  • 00:08 So this is actually the structures a little different than it.
  • 00:10 It's actually almost like an Aqua hire.
  • 00:13 So ZT Systems has two pieces of it.
  • 00:15 They've got a
  • 00:16 a a server manufacturing business
  • 00:18 and they've got like services and support business with effectively like 100 systems designs and design engineers.
  • 00:25 That's the piece that AMD wants.
  • 00:26 They're actually going to sell
  • 00:28 the manufacturing piece post codes, they're going to, they're going to list it as discontinued operations and then in, in find a strategic partner to sell it.
  • 00:36 So the four $4.9 billion value of this, it'll be like less than that'll less whatever they can get to that business.
  • 00:41 But
  • 00:42 they don't want that piece.
  • 00:43 It's low margin and puts them in direct competition with their customers.
  • 00:45 What they want is the system's expertise.
  • 00:48 If if you look at the things that they're selling,
  • 00:50 these enormous AI clusters are getting more and more complex
  • 00:54 and their large customers, especially their large hyperscale customers, they don't really buy single GPUs, they buy systems
  • 01:00 and they interlock like tons and tons of these things together.
  • 01:03 And so you really need that systems expertise that their larger competitor has in spades.
  • 01:08 I almost view this, it's kind of a concession on A and BS part that they do not have what they need
  • 01:14 given the demands of their of their large customers, they need
  • 01:17 to get it.
  • 01:18 And so this is how they are doing it.
  • 01:19 So in some sense it's a confession
  • 01:21 on that, But I mean better, better to do it if you need it then then not do it.
  • 01:25 Well, that's what I'm curious about Stacey, like where they kind of stand and all, and it's sort of this all AI
  • 01:29 hype and like, and we've obviously broadened that out because I know it's just bigger than just AI itself.
  • 01:33 But where do they stand in that?
  • 01:35 Because all the focus right now is on Jensen Wong and what he's doing over at NVIDIA.
  • 01:38 Where is AM DS place in this?
  • 01:40 Yes, I mean they're they're a second source, but they're they're small.
  • 01:43 And then again, I don't want to knock AMD too hard.
  • 01:45 So that they they said that they would do 4 1/2 billion dollars of of AI sales this year.
  • 01:50 And
  • 01:51 for some comparison, I mean their larger competitor probably do $100 billion worth this year.
  • 01:54 So they'll do four and a half five whatever.
  • 01:56 And it
  • 01:56 it was 0 a year ago.
  • 01:58 So
  • 02:00 even ramping up in
  • 02:01 in objectively that's a very good performance.
  • 02:04 I mean, relative to what's going on the market, it's very, very small.
  • 02:07 But I mean, objectively it was 0 it, it's good.
  • 02:09 However,
  • 02:11 it's not enough to just have a good chip.
  • 02:12 And I, I think they've got a decent chip.
  • 02:14 I don't know that it's as as good as the other, but they've got a decent chip.
  • 02:18 The chip and it's the software ecosystem and the networking and the hardware and the systems and everything.
  • 02:24 So they've been assembling all of those, try to assemble all those pieces.
  • 02:27 They have the chip internally.
  • 02:28 They bought a networking company a couple years ago called Pensando.
  • 02:32 They've been doing a lot of work on the software side.
  • 02:33 They bought some AI software companies recently.
  • 02:35 And now they've, they've got this to try to get the system's expertise.
  • 02:37 And so they're trying to assemble
  • 02:39 the building blocks to be able to offer the whole ecosystem that their competitors has has delivered already in in state large amount.
  • 02:47 You just mentioned that they're going to sell off the manufacturing part of the business to someone else.
  • 02:51 But
  • 02:51 in doing so, it'll still keep about 1000 engineers.
  • 02:54 So how much of this is
  • 02:55 a
  • 02:55 play for talent where AMD is stocking up on talent?
  • 02:59 Yeah,
  • 02:59 all of it, all of it.
  • 03:00 This is like I said, this is an Aqua hire is is my interpretation of it.
  • 03:04 They're
  • 03:04 paying again headline numbers a little under 5 billion once they sell it and maybe it's a few billion, they're paying
  • 03:10 for the privilege of bringing over 1000, you know, system design engineers and
  • 03:15 it's going to be $150 million in incremental OpEx
  • 03:18 and and very little revenue right now.
  • 03:21 The the revenue synergies such as they are
  • 03:23 and they are very honest.
  • 03:24 It's it's the hope that they will be able to sell more GPUs going forward.
  • 03:28 They have this expertise in house.
  • 03:29 Does this put any pressure on NVIDIA to make a any kind of acquisition to, you
  • 03:34 know, to to
  • 03:34 keep its lead, maintain its lead?
  • 03:36 I don't think so no.
  • 03:37 I mean, look, Nvidia's already got everything they need on these.
  • 03:40 They developed it in house over over many, many years.
  • 03:43 So no, I, I think this is others trying
  • 03:46 to close the gaps and fill the, the expert, the, the, the expertise holes in the functional holes that they just that they didn't have that's they're they're trying to do.
  • 03:53 But even even there
  • 03:54 NVIDIA is fully integrated and like I said, they developed this.
  • 03:57 It's it's like a well oiled
  • 03:59 machine.
  • 03:59 At least it looks like that from
  • 04:02 AMD is going to have to assemble this together from the various parts that they've been acquiring.