# Nvidia’s $1 Trillion Slide Sends Valuation to Pre-AI Boom Levels

An episode of [Bloomberg Intelligence](https://invisiblenetwork.ai/podcast/326301337), 8 Jul 2026, 17 min.

Audio: https://podtrac.com/pts/redirect.mp3/tracking.swap.fm/track/UVBrz8bN8aM2Xe47PEPu/traffic.omny.fm/d/clips/e73c998e-6e60-432f-8610-ae210140c5b1/7f534b8c-9e91-429a-b102-ae3c00090c6d/3245157d-9b25-46a6-898e-b4810115a4a5/audio.mp3?utm_source=Podcast&in_playlist=0ad812bf-fd15-4012-a18c-ae3c00090c76

Watch Paul and Scarlet LIVE every day on YouTube: http://bit.ly/3vTiACF. Bloomberg Intelligence hosted by Paul Sweeney and Scarlet Fu -Mandeep Singh, Global Head of Tech Research for Bloomberg Intelligence, discusses top tech stories. Nvidia Corp.'s stock is the cheapest it's been since before the AI boom, after losing roughly $1 trillion in market value in less than two months. The AI rotation trade is also gathering pace in Asia as investors pull money from chipmakers and hunt for cheaper ways to play the technology boom. -Robert Schiffman, Senior Tech Credit Analyst for Bloomberg Intelligence, discusses Amazon’s debt sale. Amazon.com is tapping the US high-grade bond market for the second time this year even after issuing more than $60 billion across currencies and accumulating nearly $155 billion in pro forma cash, underscoring the scale of its AI funding needs. -Anurag Rana, Senior Tech Analyst for Bloomberg Intelligence, discusses Apple saying its expanded agreement with Broadcom is expected to top $30 billion. The deal will involve manufacturing more than 15 billion chips in the US, supporting hundreds of jobs, and Apple will help Broadcom upgrade its production facilities in Colorado. See omnystudio.com/listener for privacy information.

## Transcript

[0:02] Speaker 1: Bloomberg Audio Studios, podcasts, radio news. You're listening to the
[0:10] Speaker 1: Bloomberg Intelligence podcast. Catch us live weekdays at ten am
[0:14] Speaker 1: Eastern on Apple Coarplay and Android Auto with the Bloomberg
[0:17] Speaker 1: Business App. Listen on demand wherever you get your podcasts,
[0:21] Speaker 1: or watch us live on YouTube.
[0:24] Speaker 2: Let's talk a little technology.
[0:25] Speaker 3: Why not Man Deep Seeing, global tech research head from
[0:27] Speaker 3: Bloomberg Intelligence.
[0:28] Speaker 2: He joins us here, so I'm seen as Bloomberg News story.
[0:31] Speaker 3: NVIDIAs stock is at its cheapest it's been since before
[0:34] Speaker 3: the AI boom. After news after losing roughly one trillion
[0:37] Speaker 3: dollars in market value over the less than two months.
[0:41] Speaker 2: What's going on there? Taking some profits?
[0:43] Speaker 4: Yeah, taking profits? And look, I think when it comes
[0:47] Speaker 4: to the chip space, we've heard quite a few announcements
[0:52] Speaker 4: in the last few months, both from the hyperscalers ramping
[0:56] Speaker 4: up their efforts, whether it's Amazon or Microsoft. So it's
[1:00] Speaker 4: not just Google TPUs anymore, BUTTA All these hyperscalers are
[1:04] Speaker 4: focused more on the inference side of things, which is
[1:07] Speaker 4: where you know you're generating most of your revenue if
[1:11] Speaker 4: you're focused on AI. And they've left the training market
[1:15] Speaker 4: to Nvidia. But when it comes to inferencing specifically, you've
[1:19] Speaker 4: got more competition. Even open Ai and Tropic are looking
[1:23] Speaker 4: to make their own chips. Now, the timeline for this
[1:26] Speaker 4: is all you know, two three years out. There will
[1:29] Speaker 4: be nothing imminent in terms of, you know, something that
[1:32] Speaker 4: can compete with Nvidia, but the market is anticipatory in
[1:37] Speaker 4: that sense, and the fact that everyone is looking to
[1:40] Speaker 4: build their own chips is a sign that there will
[1:43] Speaker 4: be more competition for Nvidia down the line.
[1:46] Speaker 5: That makes it sound like it's easy for them to
[1:47] Speaker 5: make their own ships. They partner up with someone and
[1:49] Speaker 5: they just make it.
[1:50] Speaker 2: It can't be that easy, is It.
[1:52] Speaker 4: Can't be that easy, which is why the timelines are
[1:55] Speaker 4: already two years out. There's nothing that I could say
[1:58] Speaker 4: over the next six months, you know there will be
[2:01] Speaker 4: a competitor to Nvideo or Google TPUs. Google TPU is
[2:04] Speaker 4: what is inspiring all these companies to build their own
[2:08] Speaker 4: chips because they've seen Google has released eight versions of
[2:12] Speaker 4: their own chip, and the Entropic models, which are the
[2:15] Speaker 4: best in class models, were trained on Google TPUs and
[2:20] Speaker 4: also they use Google TPUs for inferencing for the most part,
[2:24] Speaker 4: So now they have started to use Nvidia. But your
[2:27] Speaker 4: best model out there wasn't using Nvidia, and so from
[2:31] Speaker 4: that perspective, there's a case to be made that there
[2:34] Speaker 4: could be potentially more competition given what Google has accomplished
[2:38] Speaker 4: over the years.
[2:40] Speaker 3: The AI rotation trade is gathering pace in Asia as
[2:43] Speaker 3: investors pull money from chip makers and hunt for cheaper
[2:45] Speaker 3: ways to play the technology boom.
[2:47] Speaker 2: What are the cheaper ways to play the technology boom?
[2:50] Speaker 3: Might just go buy a utility company.
[2:53] Speaker 2: I don't think so.
[2:54] Speaker 4: Look, I think right now we are in that risk
[2:57] Speaker 4: of sentiment when it comes to the AI infrastructure trade.
[3:02] Speaker 4: People seem to be panicking, and they are, and whenever
[3:05] Speaker 4: something like this happens, there's indiscriminate selling of all AI
[3:09] Speaker 4: infrastructure names. People don't try to sort things out of
[3:13] Speaker 4: this company like Memory, for example, I mean all their earnings,
[3:17] Speaker 4: whether it's Samsung or Micron. These are companies that have
[3:22] Speaker 4: the highest margins they've ever had, shown incredible pricing power,
[3:29] Speaker 4: and you know their customers are looking to lock in
[3:32] Speaker 4: the pricing at this point because they don't want to
[3:35] Speaker 4: pay higher prices three months from now. So in terms
[3:38] Speaker 4: of pricing power, if there's any sector that has the
[3:41] Speaker 4: best pricing power right now, it's memory. And so from
[3:44] Speaker 4: that perspective, I don't see a reason why you would
[3:47] Speaker 4: want to, you know, sell these companies, especially companies with
[3:52] Speaker 4: pricing power, with margin expansion, with probably the best margins
[3:56] Speaker 4: they ever had, just because you feel, you know, they
[3:59] Speaker 4: they're hitting Peaker. I don't see that as a reason.
[4:02] Speaker 5: We were talking about in Nvidia earlier, and Vidia shares
[4:04] Speaker 5: have shot up. They're now up about nine tens of
[4:07] Speaker 5: one percent from being in the red. And this is
[4:10] Speaker 5: on a headline from the Information, which is a tech publication,
[4:13] Speaker 5: saying that China will let its top AI companies buy
[4:16] Speaker 5: some in Nvidia H two hundred chips.
[4:19] Speaker 2: This is the.
[4:20] Speaker 5: Lower grade in Nvidia AI chips, right Mande, Yes, Okay.
[4:24] Speaker 4: This is the Hopper, the not the Nerve version, but
[4:28] Speaker 4: the version that probably they were allowed to sell in China.
[4:31] Speaker 4: And yeah, the US companies have the better versions available.
[4:35] Speaker 5: Okay, so it's a downgraded version from what Nvidia offers
[4:39] Speaker 5: to the US customer. But the US government has allowed
[4:43] Speaker 5: in Nvidia to sell to China. China just didn't allow
[4:46] Speaker 5: its own companies to buy these chips up until now. Correct,
[4:49] Speaker 5: easy come, easygo, China can turn on permission and turn
[4:52] Speaker 5: off permission like a switch, right. I mean, that's the risk.
[4:55] Speaker 2: Here, that's the risk.
[4:57] Speaker 4: And look, when it comes to the Chinese open source
[5:00] Speaker 4: models so far, what we have seen from them is
[5:03] Speaker 4: they are probably six to nine months behind when it
[5:06] Speaker 4: comes to the capabilities of the models. And even though
[5:10] Speaker 4: they didn't have access to the latest in video chips,
[5:13] Speaker 4: at the end of the day, these chips are deployed
[5:16] Speaker 4: on a cloud somewhere and that cloud may not be
[5:19] Speaker 4: in China, but the customers could be these Chinese model companies.
[5:24] Speaker 4: So from that perspective, there is that aspect around accessing
[5:28] Speaker 4: the compute wherever it's available on the cloud, and that
[5:33] Speaker 4: may go beyond the geographic boundaries.
[5:35] Speaker 2: Thirty seconds s khinex.
[5:36] Speaker 3: They want to do a twenty billion dollar equity offering
[5:38] Speaker 3: here in the US to trade Friday.
[5:40] Speaker 2: Is that going to happen?
[5:41] Speaker 4: I think so yes.
[5:42] Speaker 2: I mean you've taped so that it. Yeah.
[5:44] Speaker 4: But look, I think investors who believe in this memory
[5:48] Speaker 4: cycle that it's going to be durable and long, they
[5:52] Speaker 4: will want a piece of it.
[5:54] Speaker 2: Stay with us. More from Bloomberg Intelligence coming up after this.
[6:01] Speaker 1: You're listening to the Bloomberg Intelligence podcast. Catch US Live
[6:05] Speaker 1: weekdays at ten am Eastern on Apple Coarclay, and Android
[6:08] Speaker 1: Auto with the Bloomberg Business app. Listen on demand wherever
[6:11] Speaker 1: you get your podcasts, or watch us live on YouTube.
[6:16] Speaker 3: One of the big stories obviously in capital markets over
[6:18] Speaker 3: the last really a couple of months, it's just been
[6:20] Speaker 3: the issuance of investment grade debt by technology companies associated
[6:24] Speaker 3: with their AI capital spending plans. The most recent deal
[6:27] Speaker 3: is Amazon coming with a big deal at twenty five
[6:31] Speaker 3: billion dollars. That's issuing more than sixty billion across currencies.
[6:34] Speaker 3: You know, just recently, Rob Shifman, he's been on fire
[6:37] Speaker 3: on this stuff, way ahead of it. Senior tech credit
[6:40] Speaker 3: analysts for Bloomberg Intelligence.
[6:42] Speaker 2: Amazon, just talk to us. That's the most recent deal. Rob,
[6:45] Speaker 2: Just give us a sense of how that was received.
[6:47] Speaker 3: In the marketplace and how the market's thinking about some
[6:50] Speaker 3: of these tech players that have really become big issuers.
[6:52] Speaker 6: Sure, you know, if you read Bloomberg News, it reads
[6:56] Speaker 6: like it was received terribly and quite frankly, I don't
[6:58] Speaker 6: think that was the case. You know, when they've actually
[7:01] Speaker 6: issued almost one hundred billion dollars a debt this year
[7:04] Speaker 6: across currencies, when you issue that much paper, every time
[7:08] Speaker 6: you do a deal, you need to pay a little
[7:09] Speaker 6: bit more. So new issue concessions on this were higher
[7:13] Speaker 6: than traditionally. That being said, they were probably about ten
[7:17] Speaker 6: or fifteen beeps higher. And when you're talking about a
[7:19] Speaker 6: company that's raising thirty year money at spreads over treasuries
[7:25] Speaker 6: at one hundred and ten basis points, quite frankly, it's
[7:28] Speaker 6: just a rounding error. The ability to raise twenty five
[7:31] Speaker 6: billion dollars in a couple of hours. I don't care
[7:34] Speaker 6: if it's oversubscribed by two times or five times. Amazon
[7:38] Speaker 6: could raise as much money as they want whenever they want.
[7:41] Speaker 6: They just might have to pay a little bit more.
[7:43] Speaker 5: So that paying a little bit more, you're attributing it
[7:45] Speaker 5: to the fact that they've already raised quite a bit
[7:47] Speaker 5: of money. And I think you're referring to the big
[7:49] Speaker 5: bond sale they did in Canada, the Luney offering last month.
[7:53] Speaker 5: Or is it that maybe sentiment has turned a little bit.
[7:56] Speaker 5: I think that's what we're trying to get at.
[7:57] Speaker 6: Yet, No, I don't think sentiment is turned. Doing alternative currencies,
[8:02] Speaker 6: non dollar currencies is actually good from a US dollar
[8:05] Speaker 6: trading perspective. Because for every dollar that they issue that
[8:08] Speaker 6: is in alternative currencies, they don't have to do in
[8:11] Speaker 6: US dollars. What starts to happen is large money managers
[8:14] Speaker 6: just get full on the name. You know their Amazon
[8:16] Speaker 6: has about two hundred million dollars of debt outstanding now,
[8:19] Speaker 6: and these are large positions, so incentive to incentivize people
[8:23] Speaker 6: to take on more, you just have to pay a
[8:25] Speaker 6: little bit more. But what I'm saying is the overall
[8:28] Speaker 6: cost of borrowing for them from a spread perspective, is
[8:31] Speaker 6: still somewhat tiny. So yes, sentiment is getting a little
[8:36] Speaker 6: bit weaker. That happens when you have record issuance, and
[8:39] Speaker 6: it's likely to continue. Also, Amazon's not ruling out they're
[8:42] Speaker 6: going to be back in the markets later this year.
[8:44] Speaker 6: That said, you know, if they did an alphabet style
[8:47] Speaker 6: equity deal and raised seventy five eighty five billion dollars
[8:50] Speaker 6: of equity next month, I think you're going to see
[8:53] Speaker 6: spreads tighter. And in fact, you know, with all this
[8:56] Speaker 6: Middle East news, it's actually somewhat mildly positive or credit spreads.
[9:00] Speaker 6: What happens is yields end up going higher, so the
[9:04] Speaker 6: bid from the byside, particularly out the curve goes up.
[9:08] Speaker 6: You can basically buy the same names at the same
[9:11] Speaker 6: or tighter spreads, but higher yields. So you're seeing some
[9:15] Speaker 6: of that today. So Amazon is tighter today, and even
[9:18] Speaker 6: names like SpaceX are trading tighter today. So this concept
[9:21] Speaker 6: that the large hyperscalers are not going to be able
[9:25] Speaker 6: to raise as much money as we say that they're
[9:27] Speaker 6: going to need, I think that's just sort of a
[9:30] Speaker 6: misnomer because it cost them a little bit more this
[9:32] Speaker 6: time than it did last time.
[9:33] Speaker 3: You mentioned SpaceX and Tom follows those bonds. Tom King
[9:36] Speaker 3: follows those bonds, and he's been noting that they've been
[9:39] Speaker 3: trading lower relative talk to us about how they've been trading.
[9:43] Speaker 6: Why, Well, it's the nomenclatures is important. So lowers, are
[9:48] Speaker 6: you talking about spreads or are you talking about price right,
[9:50] Speaker 6: because it's the opposite. So, you know, spreads on the
[9:53] Speaker 6: break widen twenty to twenty five basis points, and now
[9:57] Speaker 6: they've settled in, you know, a nickel or so tighter,
[10:00] Speaker 6: so they're in a much more narrow trading range. Again,
[10:03] Speaker 6: I think that these these these basis point moves are
[10:06] Speaker 6: reasonably small, and they're not telling a story that I
[10:10] Speaker 6: think we see across news, which is a news knows
[10:13] Speaker 6: bad news sells, and they put out the sky is falling,
[10:16] Speaker 6: and quite frankly, it just isn't. There's a deep bid
[10:19] Speaker 6: for SpaceX bonds, there's a bid for Oracle bonds, and
[10:22] Speaker 6: there's a bid for these double A hyperscaler bonds. It's
[10:25] Speaker 6: just at a little bit of an incremental cost over
[10:27] Speaker 6: what historically they had paid. And it makes sense. Part
[10:30] Speaker 6: of the SpaceX issue is that they're going to come back,
[10:33] Speaker 6: and they're going to come back again and again and again,
[10:36] Speaker 6: and people just need to position that. You know, if
[10:38] Speaker 6: you're a triple bing aame, you know, where should you
[10:41] Speaker 6: be traded? Well, they're trading inside of where Oracle is.
[10:43] Speaker 6: So all of a sudden, I think Oracle levels, which
[10:45] Speaker 6: are you know, thirty forty wider create a backstop where
[10:48] Speaker 6: I think SpaceX might be going.
[10:50] Speaker 5: Stay with us more from Bloomberg Intelligence.
[10:52] Speaker 2: Coming up after this.
[10:56] Speaker 1: You're listening to the Bloomberg Intelligence podcast. Catchup live weekdays
[11:01] Speaker 1: at ten am Eastern on Apple, Cocklay and Android Auto
[11:04] Speaker 1: with the Blueberg Business app. Listen on demand wherever you
[11:07] Speaker 1: get your podcasts, or watch us live on YouTube.
[11:11] Speaker 2: Let's move over to the tech space here.
[11:13] Speaker 3: Apple chip deal with Broadcom expected to exceed thirty billion dollars.
[11:18] Speaker 2: Big move for Broadcam, big mood for Apples.
[11:20] Speaker 3: Check in with Ana rog Rana, senior tech analys for
[11:22] Speaker 3: Bloomberg Intelligence, An Rock talk to us about this Apple deal.
[11:27] Speaker 7: Soon you look at it.
[11:28] Speaker 8: You know, at first, when we heard about this to
[11:30] Speaker 8: you a couple of days ago, you know, we thought
[11:33] Speaker 8: this would be something new. But it seems a little
[11:35] Speaker 8: bit like you know, similar smaller chips that they have
[11:39] Speaker 8: been buying from Broadcom for a while.
[11:41] Speaker 7: But I think the big portion is they are.
[11:42] Speaker 8: Being built in the US, are being manufactured in the US,
[11:45] Speaker 8: And I think that's really the big difference here because
[11:48] Speaker 8: you know, as we.
[11:48] Speaker 7: Know, most of these parts do come from Asia.
[11:51] Speaker 8: So this is the first part of the commitment Tim
[11:54] Speaker 8: Cook has made to President Trump that is going to.
[11:56] Speaker 7: Be buying more products in the US.
[11:58] Speaker 8: So I think that's the angle that's more important.
[12:01] Speaker 2: Now.
[12:01] Speaker 8: It is important from Broadcom's point of view because Broadcom
[12:05] Speaker 8: didn't make the modem for the iPhone and Apple's beIN
[12:09] Speaker 8: it's in housing that particular piece now that was not
[12:12] Speaker 8: good for it was a big overhand for Broadcom for
[12:14] Speaker 8: a while, but this deal for them says Apples still
[12:17] Speaker 8: going to them for other chips and other paths. So
[12:20] Speaker 8: it's a good deal more so for Broadcom and for Apple.
[12:22] Speaker 8: It's basically telling the President Trump that we are buying
[12:26] Speaker 8: more products in the US.
[12:27] Speaker 3: And broadcomstock is up three and a half percent today,
[12:30] Speaker 3: up about eleven percent year to date, you know, of
[12:32] Speaker 3: the trailing twelve months, up about forty one percent. So
[12:34] Speaker 3: it's been well received by the marketplace here. So how
[12:38] Speaker 3: does Apple think about it's supply chain these days? You know,
[12:42] Speaker 3: they've been working so hard since the pandemic and then
[12:44] Speaker 3: with President Trump's pressures to kind of rejiggerate supply chain.
[12:49] Speaker 2: Where are we in that process now on?
[12:51] Speaker 4: Right?
[12:52] Speaker 7: So people do?
[12:53] Speaker 8: I think they are doing the right thing in terms
[12:54] Speaker 8: of if you look at bill of materials of any
[12:57] Speaker 8: product of any big phone, the chip is a very
[13:00] Speaker 8: expensive part of it.
[13:01] Speaker 7: You really, I don't.
[13:02] Speaker 8: Think you can logically think about them trying to assemble
[13:07] Speaker 8: the phone.
[13:08] Speaker 7: You know, that's a very very low.
[13:10] Speaker 8: Cost product somewhere in the US, because it's not mathematically
[13:14] Speaker 8: economical for them, you know, let them do that portion
[13:16] Speaker 8: somewhere in Asia where they have the scale, low cost labor,
[13:19] Speaker 8: and the factories are already set up, but the expensive
[13:22] Speaker 8: parts can be made in the US if we have
[13:25] Speaker 8: funding from the government, if they are bigger factories that
[13:27] Speaker 8: do that kind of stuff, so that I think they're
[13:29] Speaker 8: going about it the right way. We've also heard from
[13:32] Speaker 8: our Kerman. Then there is a possibility that Apple may
[13:35] Speaker 8: work with Broadcom to do some.
[13:36] Speaker 7: AI server chips.
[13:38] Speaker 8: That's going to be actually a very expensive chip if
[13:40] Speaker 8: they gig if they actually go out and do that,
[13:43] Speaker 8: and that can be done in the US also, So
[13:46] Speaker 8: looking at the right part of the supply chain, the
[13:48] Speaker 8: more expensive parts being built, I think that's the right
[13:51] Speaker 8: way to look at it.
[13:53] Speaker 3: What do we know of the role of Tim Cook
[13:56] Speaker 3: these days? What is he doing and what's the folk
[14:00] Speaker 3: spream going forward?
[14:00] Speaker 1: Do you think?
[14:02] Speaker 8: I think going forward it's I think very clear in
[14:04] Speaker 8: the announcement that is going to make that he's going
[14:06] Speaker 8: to be, you know, making shore, taking care of some
[14:09] Speaker 8: of the government affairs, which to me means attending dinners
[14:11] Speaker 8: both in China and in the US with the respective
[14:14] Speaker 8: presidents or basically wherever the president wants him to go.
[14:17] Speaker 7: He will tag.
[14:18] Speaker 8: Along I mean, I think that's really something that.
[14:20] Speaker 7: Even in the first three months of President Drum's you.
[14:23] Speaker 8: Know, inauguration, he was not happy that Kim Cook was
[14:26] Speaker 8: not in one of the particular functions. But since then
[14:29] Speaker 8: he's been seen everywhere. So I think that's going to
[14:31] Speaker 8: be his legacy and work going forward.
[14:34] Speaker 2: And it's hugely valuable, no question about it.
[14:37] Speaker 3: That's the way a lot of CEOs have come to
[14:40] Speaker 3: the They've come to the conclusion that's the best way
[14:41] Speaker 3: to deal with this administration.
[14:43] Speaker 2: Any latest thoughts here on.
[14:45] Speaker 3: Microsoft, you know, it's you know, it goes to that
[14:49] Speaker 3: whole AI as a competitive threat to software broadly defined,
[14:53] Speaker 3: and I can get it for some companies, but boy,
[14:56] Speaker 3: it doesn't ring true for Microsoft.
[14:58] Speaker 2: They just seem to actually big of a mote. But
[14:59] Speaker 2: that's not what the market saying.
[15:01] Speaker 7: I think the market's a little bit worried about Capex.
[15:03] Speaker 8: We published a note yesterday saying that maybe they need
[15:06] Speaker 8: to lighten up a little bit on that narrative because
[15:09] Speaker 8: if you look at the valuation gap, now it's trading
[15:12] Speaker 8: at what less than twenty times earnings and that was
[15:14] Speaker 8: not the case. I think the valuation's down almost thirty
[15:17] Speaker 8: nine to forty percent over the last twelve months. And
[15:20] Speaker 8: I think a lot of art has to do with
[15:22] Speaker 8: the money they're going to spend. By our calculation, they're
[15:25] Speaker 8: going to hit closer two hundred and ninety billion by
[15:27] Speaker 8: calendar twenty twenty six. That's about fifty four percent on revenue,
[15:30] Speaker 8: and that number is going to go up next to year.
[15:33] Speaker 8: So you have to really think that you really need
[15:35] Speaker 8: to spend that much money down the road, especially when
[15:38] Speaker 8: you have a very profitable software business. It's okay to
[15:41] Speaker 8: give back on a little bit of revenue growth, but boy,
[15:44] Speaker 8: you do want to keep that free cash flow going
[15:46] Speaker 8: because that's what some of the investors are looking for.
[15:49] Speaker 3: Yet, you know, we had your credit colleague on earlier today,
[15:53] Speaker 3: Rob Shiftman from Bloomberg Intelligence, and he's basically saying, Hey,
[15:56] Speaker 3: my market's wide open. These guys can keep issuing debt
[15:58] Speaker 3: and build as much as they.
[16:00] Speaker 8: Well, the thing is, it's okay for Google to do
[16:03] Speaker 8: it because they have their own chips. Amazon it has
[16:05] Speaker 8: its own chip. I like that Microsoft is still behind
[16:09] Speaker 8: in that chip war. If Microsoft is able to come
[16:11] Speaker 8: up its own chip and create a data center where
[16:14] Speaker 8: they don't need to go out and.
[16:15] Speaker 7: Pay the in media tacks.
[16:16] Speaker 8: It's okay, but right now Microsoft chip is not at
[16:19] Speaker 8: that level that customers are running to them and say, well,
[16:22] Speaker 8: give me your AI workload, but on your chip.
[16:24] Speaker 7: So if you have to go.
[16:25] Speaker 8: Buy Invidio chips, which is sometimes half of the CAPEX
[16:28] Speaker 8: goes to in media, then you have to really think
[16:30] Speaker 8: about whether why why are you doing that?
[16:32] Speaker 2: Thirty seconds left? Do they have the ability? Do you
[16:34] Speaker 2: think they'll get their own chip at any time soon?
[16:36] Speaker 8: They have their own chip and it's working, but it's
[16:39] Speaker 8: not doesn't have the same customer preference as that what
[16:42] Speaker 8: we saw with Google TPUs or with Amazon.
[16:46] Speaker 1: This is the Bloomberg Intelligence podcast, available on Apple, Spotify,
[16:51] Speaker 1: and anywhere else you get your podcasts. Listen live each
[16:54] Speaker 1: weekday ten am to noon Eastern on Bloomberg dot com,
[16:58] Speaker 1: the iHeartRadio app, tune In, and the Bloomberg Business app.
[17:02] Speaker 1: You can also watch us live every weekday on YouTube
[17:06] Speaker 1: and always on the Bloomberg terminal

Shows are listed from Apple Podcasts' charts and directory, and Podcast Index where it is used. Every show and episode belongs to its publisher and plays from the publisher's own feed; this site hosts none of them.