- 3 hours ago
Category
đź—ž
NewsTranscript
00:00Ed Ludlow, host of Bloomberg Tech, joins us now here in New York for more on the two giants relationship
00:04souring.
00:05This is kind of a window into what you've talked so much about, this war for talent in Silicon Valley
00:11right now, centering on AI.
00:13Christina mentioning those names, Tang Tan, Johnny Ive, legends kind of of Apple's hardware operation have been brought over to
00:21open AI to work on new products there.
00:22What does this story say about that broader fight that's taking place?
00:25Well, I mean, this is a big story, and I would say probably actually as much about Silicon Valley's look
00:31to the future of hardware as much as it is the people.
00:34You know, if you think about the landscape of Apple and Meta, as examples, trying to work out how people,
00:42consumers, use AI through different device form factors.
00:45The smartphone is what we mostly use, but glasses are coming in, people have tried pins.
00:51And that's kind of the underlying layer of the story, which is interesting.
00:55But in terms of Apple versus open AI, the sort of full economic and litigating might of Apple against open
01:03AI, it's very serious.
01:05You just referenced it, but what does this mean for open AI when it comes to trying to expand into
01:10hardware?
01:10And when it comes to actually making money from people using AI, which is a bit of a fair point.
01:17Yeah, I mean, in the suit, you know, Apple describes open AI's hardware effort as nascent, but you gave the
01:23key details, which is that Apple claims there are about 400 Apple alumni now working at open AI.
01:30And the specifics, which, you know, open AI executives coached those hires as the suit alleges on how to sort
01:39of leave Apple smoothly with, as the suit alleges documents, et cetera.
01:44You know, we don't know much about the form factor or where open AI is at with its device progress.
01:54But the backstory is very much bigger than that.
01:57You know, a couple of years ago, the company started working together on software for the iPhone, integrating open AI's
02:04AI technology into Siri.
02:05Then Johnny Ives struck up a relationship with Apple and with open AI, sorry.
02:11And it's kind of soured from there.
02:13The net result, we don't know.
02:15Yeah, there's kind of a with friends like you who needs friends quality to all of this.
02:19But what is the status of the kind of business relationship or partnership between these two companies?
02:24Well, I mean, now Apple wants to go to trial with the jury trial.
02:28Fair enough.
02:29And, you know, it's worth pointing out that, well, what is it that Apple wants to achieve through this suit?
02:34They want open AI to stop what they're alleging they're doing, which is misuse of trade secrets.
02:39But more than that, the suit requests that open AI start again, re-engineer any hardware that Apple believes has
02:48their own technology in it.
02:49But basically, you know, it has been souring, I think is the phrase you used, David.
02:55That's right and it's probably kind.
02:58There was a middle ground or a middle stop between Apple's developer conference two years ago where Sam Altman was
03:04there and Apple executives were calling open AI the leader in the field.
03:08And they started this work on the software side to open AI being very frustrated with Apple because they felt
03:15that Apple wasn't moving on the things that that initial agreement had put in place.
03:20I want to point out one piece of reporting that is important.
03:23Apple had made an attempt to get open AI to change this behavior outside of court and basically, you know,
03:29cease and desist.
03:30And clearly, that was some months ago and that didn't work.
03:33So now we await next steps.
03:36But that's what Apple wants.
03:37It's so interesting to me because this is it solidifies kind of the environment in Silicon Valley.
03:42Right. But this is nothing new.
03:43Like you can think back to like GE competitors doing this and any big successful company.
03:48I feel like this always happens.
03:49You're the trendsetter.
03:50You're the gold standard.
03:51And then people defect and start their own things or they poach from there because that's where it started.
03:55Like what is the threshold for what they have to prove?
03:57Do they have to prove that they've actually stolen documents?
04:00They have to prove that they that it is very common for talented people to go from one Silicon Valley
04:06technology company to another so long as they are acting within the confines of the law.
04:13And, you know, Apple speaks to this in the suit.
04:16You know, they basically accuse open AI of coaching these people to avoid the dreaded walkout.
04:21So when you give your notice, everyone wants that two week period where you stay in post.
04:26And not get walked off the premises.
04:30But as a part of that, you know, Apple's allegation is that open AI encouraged those hires from Apple to
04:37use the window to access documentation.
04:41Specifically, though, about unreleased products and non-disclosed technology at Apple.
04:46That's the underlying point here.
04:47Let me make what I hope will be an elegant segue here.
04:50Please.
04:50So you're talking about the adventures the companies are embarking on in new hardware related to AI.
04:54And you're here in New York because I sat down with the chairman of the SK group.
04:58Yes.
04:58Within that organization, there's SK Hynix, which makes these memory chips.
05:03Crucially important as we look at kind of where AI is going and where AI related hardware is going as
05:08well.
05:08So you sat down with the chairman of that company yesterday.
05:11What did he say about the prospects for that company in this field?
05:16What's it going to mean for the production of chips broadly, for memory chips in particular going forward as we
05:20see this expansion?
05:21Yeah.
05:21This was the biggest U.S. first-time share sale by a foreign company.
05:25So they have their own shares in Korea, common shares, the list in Seoul.
05:29And now they have American depository receipts.
05:32One ADR represents a tenth of one Korean share.
05:35But it set records.
05:36They raised money for a reason.
05:38They need to build more capacity for memory chips, memory chips that go into consumer electronics, like Apple makes with
05:44the iPhone.
05:46But more critically in the AI age, high bandwidth memory, which is lots of layers of DRAM memory into data
05:52centers.
05:52And the biggest takeaway for me from that is despite their best efforts and being on the hook for hundreds
05:58of billions of dollars,
05:59they say their supply may never catch up with the growing demand.
06:03That's crazy to me.
06:04He was also, the company's been talking about how it's kind of a cyclical demand, which also was confusing to
06:09me because I feel like, well, that's what I mean.
06:11Like, right now it seems like there is just demand.
06:14So historically, memory chips are boom and bust.
06:17When there is great demand for consumer electronics, the memory chip companies ramp up capacity.
06:23And people tend to have stockpiles.
06:26Then when consumer electronics demand falls away because of an economic cycle or it's just not in vogue,
06:33sometimes those memory chip makers have been, oh, no.
06:36Now what do we do with all these big factories we've built and no one wants any memory chips?
06:40So they've tried to avoid that.
06:42Is the AI era different where the data center demand for memory is sort of more linear and keeps going?
06:47And they would say, yes, it is.
06:49A fascinating moment when he brought up the concept of memory as a service.
06:53So we think of memory as a commodity of these chips.
06:56And you teased a bit out of him in that regard.
06:59Later on, I was filling in for Ed on the show yesterday on Bloomberg Tech yesterday.
07:02Talked to a professor at you.
07:03You're welcome.
07:03All right, you two.
07:04I need to be here for this.
07:05At UCLA, he was really pioneered in memory work.
07:07He said, this is the next frontier.
07:08Explain what it is and sort of how that stands to change the way that we think about the compute
07:12that goes into AI.
07:13Honestly, I don't know.
07:14He kind of did it with his own volition, but, you know, in the world of AI, when you think
07:19about AI workloads, so not training AI models, but running them, inference, you don't sort of necessarily as an everyday
07:26consumer or even a software company, any company that's using AI say, okay, I better go out and buy myself
07:31lots of chips to run this AI stuff.
07:33You rent cloud computing capacity from Amazon or Microsoft, Google, et cetera.
07:38And I think what he was basically saying is instead of us selling the memory chips, kind of avoiding that
07:44boom and bust cycle, why don't we just rent out the memory capacity?
07:47Now, I think there's a lot of questions about how that would work in practice.
07:50And, you know, I tried to say to him, how would that work in practice?
07:53It's been really interesting because the Korean press has really picked this up and run with it.
07:57You know, it's an interesting idea, potentially higher margin.
08:02We don't know.
08:03But, you know, that is the direction that computing is going in.
08:07And that's exciting.
08:08And we also have some sound from your interview because the chairman said he had big plans for the U
08:12.S.
08:13Yes.
08:14They're already investing in more than $35 billion in the U.S. side.
08:19My plan is a much bigger number.
08:21Bigger than $35 billion.
08:22Yeah, well, much, much, much bigger than $35 billion.
08:26So much, much bigger than $35 billion.
08:29Much, much.
08:29What is he teeing up?
08:30Yeah, right now, again, SK, they have 60% market share, basically, in high bandwidth memory, these data center memory
08:39chips we've been talking about.
08:40In Korea, they've committed hundreds of billions of U.S. dollars to expand their domestic capacity, sorry.
08:46Here in the U.S., across SK Group, to date, $35 billion.
08:51So they raised $26.5 billion selling U.S.-listed shares.
08:54Kind of seems like nothing.
08:55Yeah.
08:56In the grand scheme of things.
08:57But this is where the customers are, right?
09:02NVIDIA is the biggest driver of HBM adoption.
09:05For every one NVIDIA GPU, there are eight corresponding HBM chips.
09:09So if NVIDIA is going up like that, it's logical.
09:11So would SK.
09:12So they know that's a part of it.
09:14They need to have capacity here to serve the data center build out in the United States.
09:18He just didn't give me a dollar figure.
09:20Imagine if he had.
09:20You know, we sent a red headline on the Bloomberg terminal with that because it's a very important political thing
09:25as well, right?
09:27This nation and this administration has been very proactive in getting outside capital to commit to building things here in
09:34America.
09:35You've got about a minute left.
09:36So you say that they have the largest share of high bandwidth memory production.
09:40There's Micron.
09:41There's Samsung as well.
09:41We had Howard Ludnick, the Commerce Secretary, talking about his desire for Samsung, SK Hynix, to do more here in
09:48the U.S.
09:49How resonant is that?
09:50And when you listened, and you asked about politics explicitly, his relationship with the Trump administration.
09:55What's your read on kind of the resonance of calls like that from the Trump administration to do more here
10:00in the U.S. with a company like SK Hynix?
10:02Yeah, and if you expand it beyond memory, TSMC as well.
10:05You know, the world is highly dependent on Taiwan for the production of the latest chips on the logic side,
10:15GPUs, compute.
10:16That is a security risk, many would say.
10:19So there's a great willingness to invest in America.
10:22The president has been a good dealmaker.
10:24But the debate that exists is would they ever, Samsung, TSMC, bring their latest and greatest to America prior to
10:32bringing it to Korea or Taiwan?
10:35That seems unlikely.
10:38So you're waiting for the great American champion to crack the technology side.
10:42But, yeah, that is the direction of travel.
10:45Ed, we've only got about 30 seconds left before we have to send you on your way back home.
10:49What other stories should you be watching on your beat?
10:50What are you looking at coming up?
10:52Earnings, technology company earnings.
10:53And all the American public needs to know this weekend is how much money do they spend on AI?
10:59And now, how much money comes out the other side?
11:01Like, what's the net result of all of this spending?
11:04This quarter seems like that will be more in focus.
11:07It's not getting to be more in focus.
11:07I think it's going to be more in focus.
11:07The end result of all of this.
Comments