# Intel Tumbles on PC Outlook, AWS CEO on AI Spending and Growth

An episode of [Bloomberg Tech](https://invisiblenetwork.ai/podcast/1611767434), 30 Oct 2026, 47 min.

Audio: https://podtrac.com/pts/redirect.mp3/traffic.omny.fm/d/clips/e73c998e-6e60-432f-8610-ae210140c5b1/41764a4f-fc64-4e11-89ba-ae7c0030ab5e/0931cc16-291a-444e-9447-b4f3011a058d/audio.mp3?utm_source=Podcast&in_playlist=9caafc41-289c-4115-995d-ae7c0030ab75

Bloomberg’s Ed Ludlow looks at the fall in Intel shares as a bleak outlook exposes the chipmaker's struggle with a deepening PC slump and difficulty meeting CPU demand. Plus, AWS CEO Matt Garman discusses the cloud giant's accelerating growth and massive AI buildout. And, Bitcoin tumbles after hackers siphoned off $6.3 billion from Binance in an attack that rocked the crypto world. See omnystudio.com/listener for privacy information.

## Transcript

[0:02] Speaker 1: Bloomberg Audio Studios.
[0:04] Speaker 2: Podcasts. Radio.
[0:06] Speaker 3: News.
[0:13] Speaker 4: Bloomberg Tech is live from the heart of Silicon Valley
[0:17] Speaker 4: with Ed Ludlow in San Francisco. This is Bloomberg Tech
[0:24] Speaker 4: coming up. Intel shares plunge, set for the biggest drop
[0:27] Speaker 4: on record as a bleak outlook exposes the chipmaker's struggle
[0:31] Speaker 4: with a deepening PC slump and inability to meet CPU demand. Plus,
[0:36] Speaker 4: we speak exclusively with AWS CEO Matt Garman about the
[0:40] Speaker 4: cloud giant's accelerating growth and massive AI build-out. And Bitcoin
[0:44] Speaker 4: dips below $ 77, 000 as a record $ 6 billion Binance hack
[0:49] Speaker 4: rocks the crypto world. Let's get straight to our top story,
[0:52] Speaker 4: and that is Intel. Intel shares down about 28%, 29%
[0:56] Speaker 4: right now, easily putting them on track for their biggest
[0:59] Speaker 4: one-day drop since records began back in 1982. That's the
[1:03] Speaker 4: data that Bloomberg has. I'm giving you a five-day chart
[1:05] Speaker 4: for two reasons. The severity of the drop right now
[1:08] Speaker 4: after earnings. But also, this is a stock that posted
[1:11] Speaker 4: big gains Friday through Wednesday because of enthusiasm about what
[1:15] Speaker 4: they might achieve. Then the outlook for the fourth quarter game,
[1:18] Speaker 4: and it was weak. Joining us now, Bloomberg's Ian King,
[1:20] Speaker 4: who leads our coverage of semiconductors. Let's start with the
[1:23] Speaker 4: fourth quarter outlook. Explain it to me, how it misses
[1:25] Speaker 4: relative to estimates and where the weakness is.
[1:29] Speaker 5: Yeah, I mean, they're forecasting, oh, it's going to be
[1:32] Speaker 5: up 16% from a year ago. The problem is the
[1:35] Speaker 5: expectations are much greater than that. You know, $ 2, $ 3
[1:39] Speaker 5: billion more than they said they're going to post. And
[1:42] Speaker 5: that's the problem. This is the time when they should
[1:45] Speaker 5: be cashing in on the big data center spending boom.
[1:49] Speaker 5: And unfortunately, they're not able to do that as much
[1:51] Speaker 5: as people want them to.
[1:53] Speaker 4: Okay, the weaknesses in the client segment, which is PCs.
[1:56] Speaker 4: You and I got on the phone last night with
[1:57] Speaker 4: Lit Bhutan, the CEO of Intel, and the explanation for
[2:01] Speaker 4: the weakness in PC is pretty clear, right? What is it?
[2:04] Speaker 5: Yeah, I mean, this is not a surprise. Everybody expected
[2:07] Speaker 5: this horrible train to come down the tracks, and it
[2:09] Speaker 5: really did. Basically, memory, all of that memory that's going
[2:13] Speaker 5: into data centers for AI is not going into PCs, right?
[2:18] Speaker 5: If you're going to buy a laptop right now, it's
[2:19] Speaker 5: going to cost you a lot of money if you
[2:21] Speaker 5: want to put memory in it. So that PC makers themselves,
[2:25] Speaker 5: either they can't get the memory or they're absolutely paying
[2:28] Speaker 5: through the nose for it and having to pass that
[2:30] Speaker 5: along to the consumer, which is killing demand. So what
[2:32] Speaker 5: Intel has said is like, look, the market's going to
[2:34] Speaker 5: be down 12% this year, 10% next year. We don't
[2:37] Speaker 5: really know when it's going to get better. And for Intel,
[2:40] Speaker 5: that's a huge volume market.
[2:43] Speaker 4: And what Libhutan told us on the phone, Ian, was
[2:45] Speaker 4: that this is going to continue into the first half
[2:47] Speaker 4: of next year. What's confusing.
[2:50] Speaker 1: is.
[2:50] Speaker 4: Intel is saying that demand for CPUs that go into
[2:53] Speaker 4: servers and into data centers because of agentic workloads is very,
[2:57] Speaker 4: very high. They're not able to beat that demand. Something
[3:01] Speaker 4: is going wrong. How did LitBootan explain that?
[3:04] Speaker 5: Yeah, no, you're right. We spoke to him and he
[3:06] Speaker 5: was clearly personally disappointed and frustrated when he spoke with
[3:11] Speaker 5: us and then when that came across on the call. Basically,
[3:14] Speaker 5: they're accident prone, unfortunately, at a time when their rivals
[3:18] Speaker 5: are not accident prone. They had a really unfortunate, literally
[3:22] Speaker 5: accident explosion in a plant in Israel. They can't get
[3:26] Speaker 5: enough of what's called a substrate, which is how the
[3:28] Speaker 5: chip is packaged before it goes into the computer. And
[3:31] Speaker 5: their factory in Ireland is getting better. at output but
[3:35] Speaker 5: not as fast as they wanted to so you know
[3:39] Speaker 5: the worst thing for a company is to have orders
[3:41] Speaker 5: that you can't fill and you know unfortunately likely see
[3:44] Speaker 5: them go.
[3:44] Speaker 4: Elsewhere Bloomberg's Ian King. Thank you very much indeed. Let's
[3:48] Speaker 4: stay on Intel. Joanne Feeney, Advisors Capital Partner and Portfolio
[3:52] Speaker 4: Manager joins us now. Actually, someone that does not have
[3:54] Speaker 4: a position with Intel, but has a long history covering
[3:58] Speaker 4: the chip sector and knows Intel inside and out. This
[4:02] Speaker 4: is the third straight quarter that I've got on the
[4:04] Speaker 4: phone with Lit Bhutan, and he's explained to me the
[4:07] Speaker 4: company has not been able to execute to meet demand.
[4:10] Speaker 4: How would you interpret that?
[4:13] Speaker 6: Well, they're trying to resurrect Intel, both on the research
[4:17] Speaker 6: side and the manufacturing side. We're all familiar with the
[4:20] Speaker 6: challenges they've had and Liputan trying to fix those things.
[4:23] Speaker 1: And it just takes time.
[4:25] Speaker 6: We have stayed away from the stock for our clients.
[4:29] Speaker 6: One of the opportunities in this sort of little Intel
[4:34] Speaker 6: disaster is probably for AMD, and yet the market doesn't
[4:38] Speaker 6: seem to be treating the news that way. In fact,
[4:41] Speaker 6: AMD shares are down pretty significantly. It is a stock
[4:43] Speaker 6: we own for clients.
[4:45] Speaker 4: Let me jump in here, Joanne. The conclusion I'm trying
[4:47] Speaker 4: to get to is where is the market share going?
[4:51] Speaker 4: Is Intel regaining leadership in CPU or is AMD running
[4:55] Speaker 4: away with it? Is that the point that you're getting at?
[4:57] Speaker 6: I think there are two things going on. I think
[4:59] Speaker 6: the PC market is in real trouble. The higher memory
[5:01] Speaker 6: costs are a problem. It is going to squelch demand,
[5:04] Speaker 6: and I think that's probably why AMD is down. They're
[5:06] Speaker 6: going to be hurt by that just as much as
[5:08] Speaker 6: Intel is, potentially, although AMD has been taking share and
[5:11] Speaker 6: probably are going to continue to do that.
[5:12] Speaker 1: The real opportunity might be on the server side.
[5:15] Speaker 6: Now, if you're ordering a bunch of Intel chips, you're
[5:17] Speaker 6: not going to just probably be able to immediately just
[5:19] Speaker 6: flip over to AMD chips.
[5:21] Speaker 1: But if Intel is having these execution.
[5:23] Speaker 6: Problems, you have to worry that customers will say, well,
[5:25] Speaker 6: I can't rely on Intel to deliver.
[5:27] Speaker 1: I'm going to reconfigure things. I'm going to use more AMD.
[5:30] Speaker 6: Because we know that their server chips and particularly the
[5:33] Speaker 6: way they're able to combine high-end CPUs with the GPUs
[5:36] Speaker 6: with the AI accelerators makes AMD, I think, a more
[5:40] Speaker 6: compelling opportunity.
[5:42] Speaker 4: You don't hold the stock on behalf of clients. I
[5:45] Speaker 4: would point out that the United States government or the
[5:49] Speaker 4: United States of America full stop has 5.2% of Intel.
[5:52] Speaker 4: It's something I always reflect on, particularly during the hour
[5:56] Speaker 4: of this program. There's an irony here. AI is making
[6:01] Speaker 4: memory prices expensive, which is putting the consumer off laptops
[6:05] Speaker 4: and PC's. And it's because of all of the memory
[6:08] Speaker 4: demand that's going into the data center. I mean, for Intel,
[6:12] Speaker 4: that is two opposing businesses going in different directions. Why
[6:16] Speaker 4: isn't anyone giving them the benefit of the doubt on that?
[6:21] Speaker 6: I think there's that broader impact, Ed, that the AI
[6:25] Speaker 6: build-out is creating more expenses in a lot of places.
[6:29] Speaker 6: Not just memory, but raw materials. Interest rates are being
[6:32] Speaker 6: driven up by the massive demand for borrowing from the
[6:35] Speaker 6: AI build-out. And this is causing problems across the economy.
[6:39] Speaker 6: Intel could have been able to take advantage. of the
[6:42] Speaker 6: massive demand for CPUs, as everybody has remarked, particularly Intel
[6:46] Speaker 6: and AMD, there would be a shift in the mix.
[6:48] Speaker 6: GPU to CPU ratio was going to drop, was going
[6:51] Speaker 6: to favor.
[6:52] Speaker 1: Intel and AMD.
[6:53] Speaker 6: But if you can't execute to that demand, then you're
[6:55] Speaker 6: leaving a lot of money on the table.
[6:58] Speaker 4: I should point out, at the index level, it's pretty
[7:01] Speaker 4: ugly out there this Friday. The NASDAQ 100 on the
[7:05] Speaker 4: week has been choppy, but net-net we've ended Friday with
[7:09] Speaker 4: pretty severe pressure on technology stocks and the AI trade.
[7:13] Speaker 4: The story changed really quickly. Coming out of Microsoft in particular,
[7:17] Speaker 4: there was so much positivity. Now we're back to the
[7:20] Speaker 4: same calculus, right, of spending and the payoff. It's really
[7:24] Speaker 4: hard to draw a conclusion for me, Joanne. If we
[7:27] Speaker 4: had a good week in earnings, what has the net
[7:29] Speaker 4: result been?
[7:30] Speaker 1: Yeah, it's really mixed, right?
[7:32] Speaker 6: So if you just look at the AI side of
[7:34] Speaker 6: things and the revenue, you know, you look at AWS, right?
[7:38] Speaker 6: You look at Google Cloud and you look at Microsoft Azure, right?
[7:45] Speaker 6: All of those businesses saw accelerating revenue growth. So that's
[7:50] Speaker 6: a really positive sign for what's going on in AI.
[7:52] Speaker 6: But what you had on the flip side were there
[7:54] Speaker 6: other businesses that saw some weakness, whether it was advertising
[7:58] Speaker 6: for Google, for Amazon. Microsoft's pretty clean, actually, and they're
[8:02] Speaker 6: keeping spending under control, so a lot of people.
[8:04] Speaker 1: Were encouraged by that.
[8:05] Speaker 6: But the broader question still remains, is there enough future
[8:09] Speaker 6: revenue growth to cover all these capital spending opportunities? And
[8:13] Speaker 6: then you have what I call sort of creative financial
[8:17] Speaker 6: solutions to what would otherwise be a really sticky problem
[8:20] Speaker 6: in financing a very large buildup involving some relatively new
[8:24] Speaker 6: companies like Anthropic and OpenAI. I personally see this so-called
[8:29] Speaker 6: circular financing as a really positive development because it solves
[8:32] Speaker 6: a massive capital markets imperfection problem, right?
[8:36] Speaker 1: Getting the information. You need to fund relatively new companies.
[8:39] Speaker 6: So you have the big guys with big balance sheets
[8:41] Speaker 6: coming in and helping to smooth the way and solving
[8:44] Speaker 6: some of these information problems. So I sort of see
[8:46] Speaker 6: this as the underlying numbers as a positive, seeing this
[8:49] Speaker 6: accelerating growth as a good thing. But investors are a
[8:53] Speaker 6: little bit tentative, maybe because they're overexposed and they need
[8:56] Speaker 6: to balance their strategies. What we're doing here at Advisors
[9:00] Speaker 6: Capital is making sure not to overweight these things, like
[9:03] Speaker 6: have good exposure, but don't just own AI.
[9:06] Speaker 1: It's going to be, I believe.
[9:07] Speaker 6: a good growth driver continuing for many, many quarters. But
[9:11] Speaker 6: you also want to own something in, say, the defensive space,
[9:14] Speaker 6: something in consumers, something in industrials and particularly healthcare.
[9:18] Speaker 1: So perhaps investors just got too far.
[9:20] Speaker 6: Over their skis on this and they need to rebalance
[9:23] Speaker 6: their strategies before they move into the next year.
[9:26] Speaker 4: The Nasdaq 100's off session lows, but still at that
[9:29] Speaker 4: session low, was on track for its biggest drop since June.
[9:31] Speaker 4: So there's some anxiety out there. Joanne Feeney, it's great
[9:34] Speaker 4: to have you back on the show of Advisors Capital.
[9:37] Speaker 4: The other big story out there today is Amazon. There's
[9:40] Speaker 4: Amazon AWS, which kind of beat expectations and will grow 40%
[9:45] Speaker 4: next year and the year after. And then there's the
[9:47] Speaker 4: bit that people talk about less, the retail business, which
[9:51] Speaker 4: frankly was also astonishing. much more profitable at enormous scale.
[9:56] Speaker 4: I want to bring in Poonam Goyal from Bloomberg Intelligence.
[9:59] Speaker 4: This is a company where North America operating income jumped 82%, right?
[10:03] Speaker 4: And if you look across the company, margin's really good,
[10:06] Speaker 4: but that retail business, 7.7%. Is there an explanation for that?
[10:11] Speaker 7: Yeah, absolutely. The reason why margins are much better are
[10:14] Speaker 7: twofold in my view. The first is Prime Day shifted
[10:17] Speaker 7: out of the third quarter into 2Q, so therefore you
[10:19] Speaker 7: didn't have as many steep discounts. The second is the
[10:22] Speaker 7: ad business. The ad business is still a driver of
[10:25] Speaker 7: profitability for the retail business. And the ad business, even
[10:28] Speaker 7: though it only grew 12%, that was also a reflection
[10:32] Speaker 7: of the Prime Day shift, otherwise would have been 20%,
[10:35] Speaker 7: does contribute anywhere from 60% to 80% of that top
[10:38] Speaker 7: line to the bottom line.
[10:40] Speaker 4: Yeah, I mean, online store revenue only grew 2%. What
[10:42] Speaker 4: was interesting is the interest of analysts about the grocery business.
[10:46] Speaker 4: You know, where's your head at with this grocery business
[10:48] Speaker 4: with Amazon right now?
[10:50] Speaker 7: The grocery business is one that Amazon has been targeting
[10:52] Speaker 7: for many years now. I think they finally have a
[10:55] Speaker 7: place where they know where they're headed, focusing on the
[10:59] Speaker 7: online piece a bit more. So the grocery business is
[11:01] Speaker 7: growing fast, and we think that's a great avenue for
[11:04] Speaker 7: Amazon to convert grocery buyers into also general merchandise buyers.
[11:08] Speaker 7: In fact, they said, you know, the grocery shopper that
[11:11] Speaker 7: shops grocery shops the other categories two times. So we
[11:15] Speaker 7: think as grocery scales, and we do think they finally
[11:18] Speaker 7: have the right strategy online, Making it part of the
[11:21] Speaker 7: regular Amazon ecosystem. So before when you went to grocery,
[11:24] Speaker 7: you went to Whole Foods or you went to, you know, Fresh, etc. Now,
[11:29] Speaker 7: when you add a phone charger, you can add an apple.
[11:32] Speaker 7: You couldn't do that before. So I think that change
[11:34] Speaker 7: has really accelerated that grocery push.
[11:38] Speaker 4: Later in the program, we're going to go really, really
[11:40] Speaker 4: deep on AWS. It accounts for the bulk of operating income.
[11:44] Speaker 4: AWS and cloud is where the story's been at. We'll
[11:47] Speaker 4: get to that. One of the surprises to me was
[11:49] Speaker 4: Prime Video is profitable. And I think you would say
[11:53] Speaker 4: probably the advertising is the explanation there, too.
[11:56] Speaker 7: Yeah, the advertising is the explanation, but it's also the content, right?
[12:00] Speaker 7: So where will you get ad dollars? It's when you
[12:02] Speaker 7: have the right content. They have NFL. They'll get NBA
[12:05] Speaker 7: back in the fourth quarter. They've had a great lineup
[12:08] Speaker 7: of shows. The Samurai turned pretty, etc. So when you
[12:11] Speaker 7: get this content in, the ads follow, and that's what
[12:14] Speaker 7: makes it profitable.
[12:16] Speaker 4: Rings of Power, season three, Wednesday, November 11th. It's marked
[12:20] Speaker 4: on my calendar. I don't know if it came up
[12:21] Speaker 4: in the call. Poonam Goyal of Bloomberg Intelligence. Thank you
[12:24] Speaker 4: very much indeed. So coming up later in the show,
[12:26] Speaker 4: AWS CEO Matt Garman joins us. And we have a
[12:28] Speaker 4: lot of questions about the visibility they have. Like, how
[12:31] Speaker 4: do they know about the growth that's going to come
[12:33] Speaker 4: in 27 and 28? And where are the bottlenecks and
[12:36] Speaker 4: how much are they going to spend and where? Lots
[12:38] Speaker 4: of questions to follow. Coming up right now, we're going
[12:40] Speaker 4: to speak with Roblox CEO Dave Bazzucchi. Slowing growth. Child
[12:45] Speaker 4: safety is still a big part of the conversation. And
[12:47] Speaker 4: then the company's increasing bet on AI. That's next. This
[12:50] Speaker 4: is Bloomberg Tech. Okay, let's look at shares of Roblox,
[13:06] Speaker 4: down about 20%. I think that puts them on track
[13:08] Speaker 4: for their biggest drop since late July. The online gaming
[13:11] Speaker 4: company is seeing some growing pains, rolling out tougher child
[13:15] Speaker 4: safety measures, and then daily active users falling 9% last quarter,
[13:19] Speaker 4: bookings dropping 12% year on year. But the company is
[13:23] Speaker 4: betting that AI can help reignite growth, with more than 80%
[13:26] Speaker 4: of its top games already using AI tools. Delighted to
[13:29] Speaker 4: welcome back to the program, Roblox founder and CEO, Dave Bazzucchi. Actually,
[13:34] Speaker 4: like sequentially, quarter on quarter, you guys have some underlying
[13:38] Speaker 4: momentum here. But, you know, is this policy impacting behavior
[13:42] Speaker 4: on the user side? Dave, what are you seeing?
[13:45] Speaker 8: Hey, thanks for having me on. Just want to highlight
[13:48] Speaker 8: we came off a $ 1. 7 billion quarter and beat the
[13:52] Speaker 8: high end of our guidance with 138 million daily actives.
[13:57] Speaker 8: A couple really interesting things to highlight here are 18
[14:01] Speaker 8: and up users. The DAUs there are growing 18% year
[14:05] Speaker 8: on year. We see incredible strength around the world. India's
[14:08] Speaker 8: up 37% year on year. And as you correctly noted,
[14:14] Speaker 8: we are in the midst of an AI fueled content
[14:18] Speaker 8: acceleration on the platform. Some of the metrics we use
[14:22] Speaker 8: for new content on Roblox, like how many assets are
[14:26] Speaker 8: being created, have literally tripled in the last month. So
[14:30] Speaker 8: our business is built on great content and we're just
[14:33] Speaker 8: seeing an incredible expansion of that content.
[14:37] Speaker 4: Dave, on the over 18 user cohort, does that sort
[14:42] Speaker 4: of on a per user basis carry higher revenue or
[14:46] Speaker 4: it's more monetizable for you, the growth in that segment?
[14:51] Speaker 8: Yeah, we've shared our vision of having 10% of the
[14:54] Speaker 8: global gaming marketplace run on Roblox. Right now in the US,
[14:59] Speaker 8: under 18, that's about 20% of the gaming market. 18
[15:05] Speaker 8: and up has incredible headroom. And we shared on the
[15:07] Speaker 8: earnings call, our 18 and up users spend on aggregate
[15:11] Speaker 8: about 50% more per engagement. So that is a higher
[15:16] Speaker 8: monetizing market for us. It's about four times as big
[15:21] Speaker 8: as the under 18 market, and it's growing really solidly.
[15:27] Speaker 4: Dave, have you played steal an egg?
[15:32] Speaker 8: I have played Steel and Egg, and I want to
[15:34] Speaker 8: highlight this Saturday, if you come by on Halloween at
[15:38] Speaker 8: 8 a.m., you'll find the Jonas Brothers in Steel and Egg,
[15:42] Speaker 8: and that's 11 o'clock Eastern time. So really fun, really innovative.
[15:48] Speaker 8: Just one of many new hits that are hitting the platform.
[15:52] Speaker 8: And as we get back to AI acceleration, I'll highlight
[15:55] Speaker 8: for the first time ever, we're seeing primarily AI-created experiences
[16:01] Speaker 8: happening hit our top 20. So both Break and Steal
[16:05] Speaker 8: an Egg and Hustle Valley hit the top 20, mostly
[16:09] Speaker 8: created by AI.
[16:11] Speaker 4: Dave, speaking of Halloween and with Steal an Egg, is
[16:15] Speaker 4: it scary for the company that a game like Steal
[16:19] Speaker 4: an Egg is basically half of the platform's current concurrent users?
[16:24] Speaker 4: Like the scale of it is kind of wild.
[16:27] Speaker 8: Yeah, that's not correct. It is not half of the
[16:30] Speaker 8: concurrent users. And we do see hits constantly come and
[16:34] Speaker 8: go on our platform and have a fair amount of concurrency.
[16:38] Speaker 8: The distribution of games on our platform has never been healthier.
[16:44] Speaker 8: And we see with our discovery system, roughly about 40%
[16:48] Speaker 8: of the weekly top 100 games are created within the
[16:51] Speaker 8: last three months. So if anything, the work we've done
[16:55] Speaker 8: with discovery and highlighting well-retaining games is creating a stronger
[17:01] Speaker 8: content platform.
[17:04] Speaker 4: One of the things I've really enjoyed talking to you
[17:05] Speaker 4: about is what life is like at Roblox. You and
[17:08] Speaker 4: I have chatted about, on a Friday night, the GPUs
[17:13] Speaker 4: being close to melting. Could you just talk a little
[17:16] Speaker 4: bit about the behaviors on your platform right now when
[17:18] Speaker 4: you guys are seeing the bursts of activity, when there
[17:22] Speaker 4: is big weekend momentum relative to the midweek? I find
[17:25] Speaker 4: that stuff really interesting data to follow.
[17:29] Speaker 8: There's a lot of fun 24-7 action on our platform.
[17:33] Speaker 8: I want to highlight we run a hybrid infrastructure. The
[17:36] Speaker 8: vast majority of our infrastructure, we built our own, our
[17:39] Speaker 8: own CPUs, our own GPUs.
[17:42] Speaker 4: But we have great cloud partners.
[17:44] Speaker 8: And in the midst, for example, of a peak on
[17:46] Speaker 8: a Saturday, we mix our own infra with our cloud
[17:50] Speaker 8: providers to support that growth. For our creators, they don't
[17:55] Speaker 8: have to worry about it. They make amazing content and
[17:57] Speaker 8: content drives their business. And they get that auto scaling
[18:01] Speaker 8: as part of our platform.
[18:04] Speaker 4: Dave Bazzucchi of Roblox. Really grateful to have you back
[18:06] Speaker 4: with us on Bloomberg Tech this Friday. Thank you very much.
[18:08] Speaker 3: Great to be here.
[18:09] Speaker 8: Thank you.
[18:10] Speaker 4: Okay, coming up, more than $ 6 billion of assets affected
[18:13] Speaker 4: in the biggest crypto hack on record. We have the
[18:16] Speaker 4: details next. This is Bloomberg Tech. Foxconn wants to capture
[18:32] Speaker 4: half the AI server market within the next two years
[18:35] Speaker 4: by expanding in the US and partnering with OpenAI. The
[18:39] Speaker 4: company's chairman, Young Liu, told Bloomberg TV that the demand
[18:42] Speaker 4: for AI hardware is real and he's confident it will
[18:45] Speaker 4: generate earnings.
[18:47] Speaker 3: We're all facing similar problems because this time, this new
[18:52] Speaker 3: industry is coming so big and so fast so that
[18:58] Speaker 3: the whole industry is all have the similar problem. But
[19:04] Speaker 3: the problem is not, this is not fake demand. This
[19:10] Speaker 3: is not imaginary demand.
[19:12] Speaker 4: This is real.
[19:13] Speaker 3: So that's a good problem. As long as we are
[19:16] Speaker 3: able to keep up with the demand, we'll make money.
[19:20] Speaker 4: Lou also confirmed meeting with OpenAI CEO Sam Altman, saying, quote,
[19:24] Speaker 4: there's still a lot of details we don't know about
[19:27] Speaker 4: the partnership. A huge story. The crypto industry is suffering
[19:31] Speaker 4: what is estimated to be its biggest ever hack. Binance
[19:34] Speaker 4: says $ 6. 3 billion worth of funds were affected, with $ 3
[19:38] Speaker 4: billion of assets frozen and now suspending all withdrawals as
[19:41] Speaker 4: it kind of completes this security review and revamp. Joining us,
[19:45] Speaker 4: Bloomberg Digital Finance Reporter Emily Nicole. Let's start with just
[19:48] Speaker 4: the basics of what happened. What do we know about
[19:51] Speaker 4: this hack?
[19:53] Speaker 9: Well, we know so far is that finance said an
[19:56] Speaker 9: employee's laptop was targeted, among some other unnamed additional points
[20:00] Speaker 9: of breach. And from that, they were able to gain
[20:03] Speaker 9: access to a Binance wallet and that they were able
[20:05] Speaker 9: to then send quite a lot of assets out to
[20:08] Speaker 9: their own wallets, about $ 6. 3 billion in total. And then
[20:12] Speaker 9: we've had several blockchain analytics firms come forward, Elliptic, TRM, Chainalysis,
[20:17] Speaker 9: and say that the way that these funds are now
[20:18] Speaker 9: being laundered points to the DPRK in North Korea as
[20:23] Speaker 9: the perpetrators.
[20:23] Speaker 4: Okay, interesting. Our reporting, hackers siphoned off $ 6. 3 billion from
[20:30] Speaker 4: crypto exchange Binance. And now we're trying to explain what
[20:34] Speaker 4: Binance can do in response to this. What's the action
[20:38] Speaker 4: the company is taking?
[20:41] Speaker 9: So it's coordinating with law enforcement.
[20:42] Speaker 1: That's what they say.
[20:43] Speaker 9: About half of that has already been frozen. So there
[20:46] Speaker 9: are certain crypto assets in this world where the issuer
[20:48] Speaker 9: still maintains a certain degree of control over the tokens themselves.
[20:51] Speaker 10: That's assets like.
[20:52] Speaker 9: Stablecoins, Tether, Circle. And about 3 billion of those, that's
[20:56] Speaker 9: 6 billion, has been frozen. So those assets can't be moved.
[20:59] Speaker 9: The rest, the hacker has gotten away with and it's
[21:02] Speaker 9: what they've got. But law enforcement and Binance and other
[21:05] Speaker 9: parties are all working to trace that down, make sure
[21:08] Speaker 9: that those funds also can't be used if possible or
[21:10] Speaker 9: recover them at best. But Binance says that it has
[21:13] Speaker 9: enough company funds set by to account for the full
[21:16] Speaker 9: 6 billion if it needs to. So there shouldn't really
[21:19] Speaker 9: be any kind of haircut for customers here.
[21:23] Speaker 4: Emily, what are we seeing in Bitcoin, other tokens, sort
[21:26] Speaker 4: of in reaction to it?
[21:29] Speaker 9: So initially, when the news of this started to come
[21:31] Speaker 9: out yesterday, Bitcoin did sell off a little bit. It
[21:33] Speaker 9: went to a multi-month low. But since then, it's been
[21:36] Speaker 9: pretty stable. And that's kind of going the same for
[21:38] Speaker 9: other currencies as well, cryptocurrencies. It's quite surprising, actually. Normally,
[21:43] Speaker 9: when the biggest exchange in your industry gets hacked, you'd
[21:45] Speaker 9: think there'd be a massive market panic. But actually, most
[21:48] Speaker 9: people have been quite calm. And that's likely because from
[21:50] Speaker 9: what we're hearing anyway, because Binance is the biggest, we
[21:53] Speaker 9: all know that it has quite a lot of profits
[21:55] Speaker 9: put by, you know, not specifics. I don't disclose that,
[21:58] Speaker 9: but it's generally seen as a very profitable company. They
[22:00] Speaker 9: say that their revenue for this year alone could cover
[22:03] Speaker 9: that loss in one go. So generally, people are quite
[22:06] Speaker 9: calm about the fact that this has happened, that Binance
[22:08] Speaker 9: will sort it out. Withdrawals are still suspended for now,
[22:10] Speaker 9: but they expect to reopen by Sunday.
[22:12] Speaker 3: So.
[22:14] Speaker 4: Okay, this is a must read. Go to Bloomberg.com, the
[22:17] Speaker 4: terminal to get it. Big thank you to Bloomberg's Emily
[22:19] Speaker 4: Nicole and the team who I'm sure will be working
[22:21] Speaker 4: on it over the weekend. Coming up, Matt Garman, CEO
[22:23] Speaker 4: of Amazon Web Services, to discuss earnings where AWS, that's
[22:27] Speaker 4: where the data points were. Earnings is what is driving
[22:30] Speaker 4: markets this Friday. Actually, overnight, basically, the story has changed
[22:35] Speaker 4: and it has soured. Intel is headed for its biggest
[22:37] Speaker 4: drop on record. Our data goes back to 1982. It
[22:40] Speaker 4: is down 28%. That is dragging the indexes. Amazon also
[22:44] Speaker 4: under some pressure, although again, positive set of results. It's halftime, guys.
[22:49] Speaker 4: We're in the city of San Francisco. We'll be right back.
[22:51] Speaker 4: This is Bloomberg Tech. Welcome back to Bloomberg Tech. Technology
[23:04] Speaker 4: shares are under pressure and earnings is a big part
[23:07] Speaker 4: of the story there. The Nasdaq 100 is off by
[23:10] Speaker 4: almost 3 percent. The Philadelphia Semiconductor Index off by almost
[23:13] Speaker 4: 7 percent. By the way, the SOX is now in
[23:15] Speaker 4: negative territory for the week, putting on track for its
[23:18] Speaker 4: fourth straight weekly drop. which is its longest run of
[23:22] Speaker 4: weekly declines going back to April of 2025. So there
[23:24] Speaker 4: is actually a little bit of sustained pressure there in Intel,
[23:27] Speaker 4: which is down on track for its biggest drop on record. Frankly,
[23:31] Speaker 4: our data goes back to 1982. It IPO'd in 71,
[23:32] Speaker 4: but the stock's down almost 30%. after very weak outlook
[23:39] Speaker 4: and guidance, largely because of the PC or client segment.
[23:42] Speaker 4: A really good set of numbers comes from Amazon, particularly
[23:45] Speaker 4: in AWS. We went over the retail business earlier. The
[23:48] Speaker 4: stock's down, but this is a company that says AWS
[23:52] Speaker 4: is going to grow 40% in each of the next
[23:55] Speaker 4: two years, and the big AI build-out is well underway.
[23:58] Speaker 4: Let's talk through it with the CEO of Amazon Web Services,
[24:01] Speaker 4: Matt Garman. Matt, it's great to have you back with
[24:03] Speaker 4: us on Blue Bag Tech. Thanks for having me. Andy
[24:06] Speaker 4: was talking about this path to AWS in and of itself,
[24:11] Speaker 4: being a $ 1 trillion revenue business in five to 10 years.
[24:15] Speaker 4: I know that he was asked, but I'd invite you
[24:17] Speaker 4: to walk us through how you get there, what the
[24:20] Speaker 4: roadmap is from here. Yeah.
[24:23] Speaker 10: Well, I think, look, if you look at the business
[24:24] Speaker 10: that we have today, we're $ 184 billion run rate business
[24:28] Speaker 10: that's growing 39%. And we've accelerated every single quarter for
[24:32] Speaker 10: the last six quarters. And so it's really unusual. In fact,
[24:35] Speaker 10: it's very unheard of for a company or a business
[24:39] Speaker 10: of that size to be growing at 39%. As we
[24:42] Speaker 10: look forward to the demand we're seeing and what we
[24:44] Speaker 10: hear from customers, we see 40% plus growth rates for
[24:48] Speaker 10: the next two years as well. As you start to
[24:51] Speaker 10: compound on that really large business, that gets really big
[24:54] Speaker 10: really quickly. It's something that we're quite excited about and
[24:58] Speaker 10: a really big potential for Amazon and for AWS and
[25:02] Speaker 10: for our customers.
[25:03] Speaker 4: I think what was so interesting about what was said is,
[25:07] Speaker 4: in my experience, Amazon's been reluctant to forecast AWS growth
[25:12] Speaker 4: that further out and further afield. So you're saying 40%
[25:16] Speaker 4: growth in both 2027 and 2028. What gave you confidence
[25:21] Speaker 4: to say that? What are the data points that you're
[25:23] Speaker 4: looking at most closely? Yeah, it's a good question.
[25:27] Speaker 10: And what we're seeing is as we get more and
[25:29] Speaker 10: more really large enterprises building on top of AWS, we
[25:32] Speaker 10: get better visibility into what that business looks like for
[25:36] Speaker 10: the next year, two, three, four, five. And so as
[25:39] Speaker 10: that business and that base started to grow and we
[25:41] Speaker 10: get more and more commitments, if you see in our backlog,
[25:44] Speaker 10: it's grown to be $ 580 billion. So we have that
[25:47] Speaker 10: much committed to spend on AWS and it's growing really
[25:50] Speaker 10: rapidly every single quarter. And so for us, we're investing
[25:54] Speaker 10: lots of capacity, lots of capital to grow that business
[25:58] Speaker 10: so that we have capacity for all of our customers.
[26:00] Speaker 10: And so our goal is to manage that capacity and
[26:03] Speaker 10: make sure that we have that. We're investing a lot
[26:06] Speaker 10: and we have lots of capacity for our customers. In fact,
[26:09] Speaker 10: as we talk to our enterprises and startups that are
[26:11] Speaker 10: all building on AWS, we want to be sure and
[26:14] Speaker 10: give them assurance that they have capacity to grow as
[26:16] Speaker 10: they see this big opportunity in front of them. And
[26:19] Speaker 10: so the combination of those really gives us confidence in
[26:22] Speaker 10: the business growing.
[26:23] Speaker 4: And Matt, just to be clear, you used the word committed,
[26:26] Speaker 4: but is the idea that this is, reflective of the backlog,
[26:29] Speaker 4: this is contracted demand rather than anticipated demand?
[26:35] Speaker 10: A lot of it is contracted, not all of it,
[26:36] Speaker 10: of course. We see lots of startups coming in. We
[26:38] Speaker 10: see customers growing. But we see this base of customers.
[26:42] Speaker 10: And so the typical pattern that we see is a
[26:45] Speaker 10: customer will come in and make a big commitment to
[26:47] Speaker 10: us and say, we're their cloud platform. And it might
[26:49] Speaker 10: be customers like ADP. It might be customers like actually Roblox,
[26:54] Speaker 10: who was just on earlier with you, that are seeing
[26:56] Speaker 10: what their demand looks like and seeing what AI is driving.
[26:59] Speaker 1: That's right.
[27:00] Speaker 10: And we get really that flywheel effect where we understand
[27:02] Speaker 10: what their baseline growth is as well as what their
[27:04] Speaker 10: potential growth is. And we kind of layer in all
[27:07] Speaker 10: of those customers, whether they're enterprises, whether they're startups, whether
[27:10] Speaker 10: they're AI labs. And then we make sure that we
[27:13] Speaker 10: have capacity and growth for all of those. And that
[27:15] Speaker 10: gives us good visibility into what's happening for the next
[27:17] Speaker 10: couple of years.
[27:19] Speaker 4: Matt, Andy was saying, and I've heard you say, and
[27:22] Speaker 4: he really emphasized, this is the early days of AI.
[27:25] Speaker 4: We're in the early innings. This is early. And while
[27:29] Speaker 4: it's sort of a bit of a paradox, what is
[27:31] Speaker 4: it that hasn't happened yet that we can keep saying
[27:34] Speaker 4: we're in the early days of AI? Yeah.
[27:37] Speaker 10: It's super interesting because we've never seen a business grow
[27:40] Speaker 10: this fast. We've never seen a technology grow this fast.
[27:43] Speaker 10: And for those of us that are on the inside
[27:45] Speaker 10: and can see what the technology is capable of, it
[27:48] Speaker 10: really is transformative across every single thing that our customers
[27:51] Speaker 10: are doing. And many of our customers are just in
[27:54] Speaker 10: the early stages of really figuring out how they get
[27:56] Speaker 10: value out of this. And I spend time with customers
[27:59] Speaker 10: and enterprises and startups every single day, and they are
[28:02] Speaker 10: seeing how they do business, how their company runs, how
[28:05] Speaker 10: their products work, completely be transformed by AI inside of
[28:08] Speaker 10: those companies. But they're at the very early stages and
[28:11] Speaker 10: they're getting so much value even out of the 1, 2, 3, 4, 5%
[28:15] Speaker 10: that they're transforming. And so we see customers on that
[28:19] Speaker 10: journey where 3% is going to grow to 10%, is
[28:22] Speaker 10: going to grow to 50% of their business that's transformed.
[28:25] Speaker 10: And so that is a massive opportunity for better efficiency,
[28:29] Speaker 10: faster growth for all of our customers, and in manifests
[28:33] Speaker 10: into more demand for AI and cloud services.
[28:37] Speaker 4: We're live on Bloomberg Television and on Bloomberg Radio and
[28:40] Speaker 4: speaking to the CEO of AWS, Matt Garman. And AWS
[28:44] Speaker 4: gave a really strong picture of the future in terms
[28:46] Speaker 4: of growth. The point that you just made is really
[28:49] Speaker 4: interesting because those are across the startup and the enterprise, right?
[28:54] Speaker 4: The thing that people want to know is who specifically
[28:57] Speaker 4: is driving the growth at the high level right now.
[29:00] Speaker 4: And so is it just those two big AI labs
[29:03] Speaker 4: or is it more widespread than that?
[29:06] Speaker 10: Yeah, and that's honestly what gives us confidence is that
[29:09] Speaker 10: widespread of growth. And I think for our business in particular,
[29:12] Speaker 10: We have a really large diversity of customers. And so
[29:15] Speaker 10: we work deeply, of course, with the large AI labs,
[29:17] Speaker 10: Anthropic and OpenAI. We work really well with the large
[29:21] Speaker 10: technology companies, whether it's a Meta or a Salesforce or
[29:25] Speaker 10: Apple or the large companies. But we have customers across
[29:29] Speaker 10: the whole portfolio, whether they're banks, whether they're healthcare companies,
[29:32] Speaker 10: whether it's a two-person startup that's just getting started in
[29:35] Speaker 10: an apartment somewhere. Across that whole industry, that's the demand
[29:39] Speaker 10: that we see. And so we have tens, hundreds of
[29:41] Speaker 10: thousands of customers that are spending more and more every year.
[29:45] Speaker 10: We have more customers that have committed to billion-dollar spends
[29:48] Speaker 10: with us this year than we ever have before.
[29:50] Speaker 4: How big is that cohort, Matt? How big is the
[29:53] Speaker 4: 1 billion total?
[29:55] Speaker 10: But it's pretty significant. And we see it growing regularly
[29:59] Speaker 10: over and over again. And so that growth is what
[30:01] Speaker 10: really gives us confidence that it's a diversity. It's not just,
[30:04] Speaker 10: you know, I think you look at some of the
[30:06] Speaker 10: others that are growing, they have really high concentration with
[30:09] Speaker 10: just a couple of these labs. For AWS, that's not
[30:11] Speaker 10: the case. We're spread across almost every single industry that
[30:14] Speaker 10: you can imagine. And As those industries start to move
[30:18] Speaker 10: to the cloud, they want to take advantage of AI.
[30:20] Speaker 10: AWS is the place that they want to launch. And
[30:22] Speaker 10: it's because they want to grow. They trust us. They
[30:25] Speaker 10: trust us for enterprise security. They trust us for availability.
[30:28] Speaker 10: And they trust us for the set of technologies that
[30:30] Speaker 10: we have available. And that's really what's leading to that
[30:33] Speaker 10: sustained growth and gives us confidence across a wide set
[30:36] Speaker 10: of customers.
[30:37] Speaker 4: Is it quantifiable, specifically on Anthropic? Is this case, say,
[30:41] Speaker 4: how much of that 40% growth right now is Anthropic
[30:44] Speaker 4: spending on compute from AWS?
[30:48] Speaker 10: The interesting thing is, and you can read their S1,
[30:51] Speaker 10: and they share some of what they spend with us
[30:53] Speaker 10: in that filing. But I'll say that most of that
[30:56] Speaker 10: spend is actually spread across customers, too, because a lot
[30:59] Speaker 10: of that spend is in Bedrock. And so it's actually
[31:01] Speaker 10: our customers that are spending on Anthropic in Bedrock. And
[31:05] Speaker 10: so for us, well, a lot of that can even
[31:07] Speaker 10: look like it's coming directly to Anthropic, and it is.
[31:09] Speaker 10: They're a fantastic partner of ours. It's spread across tens
[31:12] Speaker 10: to hundreds of thousands of customers that are all doing
[31:14] Speaker 10: inference inside of their applications on top of Bedrock, on
[31:18] Speaker 10: top of AWS. And for us, it's actually more diverse
[31:21] Speaker 10: than it even looks from those numbers.
[31:23] Speaker 4: You are adding capacity as quickly as you can and
[31:26] Speaker 4: trying to give confidence to the customers that it will
[31:28] Speaker 4: be there when they need it. What is the biggest
[31:31] Speaker 4: constraint today, Matt, as you and I speak now? Chips,
[31:34] Speaker 4: getting the data center built, power?
[31:37] Speaker 2: Yeah.
[31:38] Speaker 10: Look, the whole industry is ramping up really aggressively, and
[31:42] Speaker 10: we are building really, really fast. And so it's day-to-day
[31:46] Speaker 10: as to what the biggest bottleneck is. I think we're
[31:49] Speaker 10: running really fast across all of those that you mentioned.
[31:52] Speaker 10: It's power and data centers to make sure that we
[31:54] Speaker 10: have enough space to run for our customers. It's DRAM
[31:58] Speaker 10: some days. It's chips other days. Sometimes it's network components
[32:01] Speaker 10: other days. And part of what we do for our
[32:03] Speaker 10: customers is actually really understand all the way down three
[32:06] Speaker 10: levels into that supply chain to really understand deeply what
[32:10] Speaker 10: are the components our customers are going to need, how
[32:13] Speaker 10: do we get commitments from our suppliers multiple years out,
[32:17] Speaker 10: and then ensure that we have all of the processes
[32:19] Speaker 10: together so that we can land these in a really
[32:21] Speaker 10: efficient way. I think we shared recently, we're installing an
[32:24] Speaker 10: accelerator chip, whether a GPU or a tranium chip, every
[32:27] Speaker 10: 14 seconds. It is a massive and rapid scale-up. And
[32:32] Speaker 10: today we're still supply constrained and we're scaling really rapidly
[32:35] Speaker 10: for customers and we wish we could do it faster.
[32:39] Speaker 4: The next big evolution of the business is to sell
[32:42] Speaker 4: Tranium outside of AWS next year. I think a lot
[32:45] Speaker 4: of people misunderstand how that works in practice, what we
[32:48] Speaker 4: mean when we say that, Matt. But it must be
[32:50] Speaker 4: a really hard choice because the demand internally is also
[32:53] Speaker 4: there for the use of Tranium.
[32:55] Speaker 7: Yeah.
[32:56] Speaker 2: Yeah.
[32:56] Speaker 10: It's true, and that's actually a debate that we have internally,
[32:59] Speaker 10: but I think it's an important strategic move for us,
[33:01] Speaker 10: both for us and for our customers. When we talk
[33:03] Speaker 10: to customers, they say, look, I want to run the
[33:06] Speaker 10: majority of my workloads inside of AWS, and Tranium is
[33:11] Speaker 10: a really powerful and great cost-effective platform at the very
[33:15] Speaker 10: highest performance level for customers to run their AI workloads.
[33:19] Speaker 10: But certain sets of customers also tell us that they
[33:21] Speaker 10: want to run these things on premise. And so there is,
[33:23] Speaker 10: you know, not everything is going to be in the
[33:25] Speaker 10: cloud in the near term. Sometimes customers want to have
[33:29] Speaker 10: their own data centers. They want to run on prem
[33:31] Speaker 10: and they actually want to run Tranium in both locations
[33:33] Speaker 10: so they can run in AWS and in their own
[33:36] Speaker 10: data centers. Or in other places. And so for us,
[33:40] Speaker 10: you know, we have to balance exactly what that looks like.
[33:42] Speaker 10: But we're quite excited about that business. I think it's
[33:44] Speaker 10: going to be really interesting. And it's going to enable
[33:47] Speaker 10: our customers who really want to build on that platform.
[33:49] Speaker 10: They love the Tranium platform. They see the cost performance
[33:52] Speaker 10: benefits of it. And they want to run across a
[33:55] Speaker 10: bunch of different places. And we want to make that
[33:57] Speaker 10: possible for them.
[33:58] Speaker 4: Matt, to end our conversation, the 10Q hit, and I
[34:01] Speaker 4: was reading through it as one must. Amazon's arranged about
[34:06] Speaker 4: $ 80 billion of supplier financing for technology infrastructure starting next year.
[34:11] Speaker 4: What is that for? Why do you need to do
[34:13] Speaker 4: that mechanism?
[34:15] Speaker 10: Well, look, when we see such a really large opportunity
[34:18] Speaker 10: in front of us, We're looking for ways to also
[34:21] Speaker 10: responsibly fund that so that we can go as big
[34:24] Speaker 10: as possible for our customers. And so we look to
[34:27] Speaker 10: how do we kind of plan the capex spend to
[34:31] Speaker 10: balance with revenue. And if you look at the AWS
[34:33] Speaker 10: business model, it's always been the case that there's a
[34:35] Speaker 10: very large amount of capex that goes up front. Then
[34:38] Speaker 10: we monetize that over periods of time, whether it's five
[34:41] Speaker 10: or six years for servers, whether it's 20 or 30
[34:43] Speaker 10: years for data centers and power. That's the model that
[34:46] Speaker 10: we've always had. We're in such a rapid ramp-up right
[34:49] Speaker 10: now with a business that we really love. We have
[34:52] Speaker 10: a business that's growing at 39% with margins in the
[34:56] Speaker 10: high 30%. That is an incredible business, and we see
[34:59] Speaker 10: such an opportunity to invest more and more there. We
[35:02] Speaker 10: look across our partner ecosystem and others to figure out
[35:05] Speaker 10: how we can balance some of those cash flows as
[35:07] Speaker 10: we're really ramping up aggressively and taking advantage of what
[35:10] Speaker 10: we think is a massive opportunity that, as you mentioned
[35:13] Speaker 10: at the beginning of the show, we think can lead
[35:15] Speaker 10: to a trillion-dollar business over time.
[35:17] Speaker 4: You've been very generous with your time this morning. AWS
[35:21] Speaker 4: CEO Matt Garman, it's been great to have you back
[35:22] Speaker 4: on Bloomberg Tech. Thank you very much. Coming up, we're
[35:25] Speaker 4: going to speak to Peter Singlehurst, who is the head
[35:27] Speaker 4: of private companies and that team at Bailey Gifford. And
[35:30] Speaker 4: we're going to talk about Anthropic. A reminder, the big
[35:33] Speaker 4: story earlier in the week, yes, it was this week,
[35:36] Speaker 4: is Anthropic did file publicly for an IPO. We'll get
[35:39] Speaker 4: to it. This is Bloomberg Tech. it's been a wicked
[35:55] Speaker 4: week for tech news halloween pun intended but it started
[36:00] Speaker 4: with anthropic flipping the s1 public yes that was this
[36:05] Speaker 4: week spooky Peter Singlehurst is the head of private companies
[36:10] Speaker 4: team at Bailey Gifford, the firm backed Anthropic in 2025.
[36:14] Speaker 4: And then again in the Series G and H rounds.
[36:17] Speaker 4: And I guess, Peter, you're at an advantage because when
[36:20] Speaker 4: we had this conversation Monday with different venture firms, our
[36:24] Speaker 4: team here at Bloomberg, you've had five days to read
[36:27] Speaker 4: the thing. I think what's interesting, what have you learned
[36:30] Speaker 4: that's new? What have you come to appreciate about Anthropic's
[36:33] Speaker 4: business that you might have not known before the flip?
[36:38] Speaker 2: I think one of the reasons we invest in companies
[36:40] Speaker 2: privately is so that by the time they come to
[36:43] Speaker 2: the public markets, we're not learning new things in the S1.
[36:47] Speaker 2: If we're learning new things in the S1, it means
[36:49] Speaker 2: that perhaps we've not been following the company as closely
[36:52] Speaker 2: as we should.
[36:52] Speaker 4: But fortunately, there was.
[36:55] Speaker 2: Nothing of particular new news to us in the Anthropic
[36:59] Speaker 2: S1 other than the continued strong execution within the business.
[37:04] Speaker 4: Then let me ask you about what that reflects, the execution.
[37:07] Speaker 4: The data point that I really spent time trying to
[37:10] Speaker 4: understand was in 2023, compute costs were 10x revenue, right?
[37:16] Speaker 4: And if you go to the first half of this year,
[37:18] Speaker 4: compute costs is 55% of current revenue in that 1H period.
[37:23] Speaker 2: What does that tell us? What should we come to
[37:25] Speaker 2: know about that? I think there are a few things
[37:27] Speaker 2: that you can learn from that. The first is it
[37:29] Speaker 2: speaks to the strength of their strategy around compute.
[37:35] Speaker 4: One of the things that we really.
[37:36] Speaker 2: Liked about the business and that got us to invest
[37:39] Speaker 2: was the multi-silicon approach. The fact that they are able
[37:42] Speaker 2: to train their models and run their models across not
[37:46] Speaker 2: just NVIDIA GPUs, but across Tranium, across TPUs, and now
[37:50] Speaker 2: across AMD chips as well. This gives them flexibility on
[37:54] Speaker 2: who they can buy chips from, and it's that that
[37:57] Speaker 2: kind of gives them this ability to scale up their
[38:01] Speaker 2: revenue at a faster rate than scaling up their compute cost.
[38:05] Speaker 2: But I think it also speaks to something that they've
[38:07] Speaker 2: spoken a lot about, which is this compute multiplier effect
[38:11] Speaker 2: that they are seeing, where for every dollar of computes,
[38:17] Speaker 2: both on research and on inference, they're getting more and
[38:21] Speaker 2: more intelligence. The thing that's really fascinating though is that
[38:26] Speaker 2: Whilst historically those improvements have been driven by their researchers,
[38:31] Speaker 2: their engineers driving that efficiency, Claude is now starting to
[38:35] Speaker 2: develop those compute multipliers itself.
[38:38] Speaker 5: Yes.
[38:39] Speaker 2: This kind of gets into this fascinating possibility for the
[38:42] Speaker 2: possibility of recursive self-improvement, which could be one of the
[38:46] Speaker 2: biggest unlocks within AI and is perhaps something that people
[38:49] Speaker 2: aren't really factoring in to the potential just yet.
[38:53] Speaker 4: Now, that is interesting because that is about strategy and
[38:57] Speaker 4: executing on the business. The thing that I think people
[39:01] Speaker 4: find it hard to understand, right, is the public benefit
[39:03] Speaker 4: corp piece of it, that there is a public benefit goal,
[39:07] Speaker 4: a set of public benefit goals. And then there is
[39:10] Speaker 4: the founder's entity having 50.1% of the shareholder voting power
[39:17] Speaker 4: when all is said and done. You know, at the
[39:20] Speaker 4: end of the day, it's an investment. You made it
[39:21] Speaker 4: in the private markets and it will come to fruition
[39:24] Speaker 4: if and when Anthropo goes public. What is your attitude
[39:27] Speaker 4: towards those two mechanisms?
[39:31] Speaker 2: Whenever companies have these dual-class structures, these unusual governance structures,
[39:38] Speaker 2: there has to be a good why that stands behind it.
[39:42] Speaker 2: And when there is a good why or a good reason,
[39:45] Speaker 2: we can really get on board with those. And I
[39:47] Speaker 2: think in the case of Anthropic, there is a very
[39:48] Speaker 2: clear reason as to why it is in the interest
[39:52] Speaker 2: of long-term shareholders for this to be a company which
[39:57] Speaker 2: has that strong founder control. It's a business where there
[40:01] Speaker 2: are going to need to be short-term sacrifices made for
[40:04] Speaker 2: the long-term benefit of the company. And we've seen them
[40:07] Speaker 2: do this on a number of occasions already. It's a
[40:10] Speaker 2: company that will potentially pose significant risks. And the technology
[40:16] Speaker 2: itself is so powerful that it needs to have the
[40:18] Speaker 2: right governance structure around it. Now, if you were to
[40:22] Speaker 2: have an ordinary voting structure, the potential for the company
[40:25] Speaker 2: to start making the wrong sorts of decisions, sorts of
[40:27] Speaker 2: decisions that might lead it towards perhaps maximizing short-term gains
[40:32] Speaker 2: at the expense of both long-term gains and safety, could
[40:35] Speaker 2: be very real.
[40:36] Speaker 1: So in the case of Anthropic, we're really.
[40:38] Speaker 2: On board with the governance structure, with the voting structure,
[40:42] Speaker 2: because we believe it's in the long-term interest of the business.
[40:44] Speaker 2: And anything that is in the long-term interest of the
[40:47] Speaker 2: business is also in the interest of long-term shareholders.
[40:51] Speaker 4: I don't know if you heard the conversation I was
[40:53] Speaker 4: just having with Matt Garman, the AWS CEO, but part
[40:57] Speaker 4: of that was, you know, people keep saying we're in
[40:59] Speaker 4: the very early days of AI, the very early innings, right?
[41:04] Speaker 4: And it's really hard to know when we exit the
[41:06] Speaker 4: early innings, but some of the data from the prospectus,
[41:09] Speaker 4: was you know talking about these different cohorts of anthropic
[41:12] Speaker 4: customers that have grown to from say like 950 to
[41:16] Speaker 4: 12 000 who are spending a hundred thousand dollars a
[41:19] Speaker 4: year or like the penetration of uh fortune 500 spend
[41:24] Speaker 4: not the number of fortune 500 that they do business
[41:26] Speaker 4: with but how much of their spending they captured seemed
[41:28] Speaker 4: like really small to me what did you make of
[41:31] Speaker 4: that Yeah, look.
[41:33] Speaker 2: I think we certainly agree that we're still in the
[41:36] Speaker 2: very early days for AI. I'm reminded actually of the
[41:39] Speaker 2: Amazon shareholder letters where for years and years and years,
[41:42] Speaker 2: Jeff Bezos would still say, look, it's still day one
[41:45] Speaker 2: for e-commerce. And I think with any technological change... These
[41:49] Speaker 2: things happen over decades, not years. If you think of
[41:52] Speaker 2: Moore's Law, it's played out since the 1940s, the 1950s.
[41:57] Speaker 2: The internet is continuing to play out even though it's
[42:00] Speaker 2: been around now for three decades. Mobile, like these things
[42:04] Speaker 2: play out over very long periods of time. I think
[42:06] Speaker 2: it would be utterly bizarre if the opportunities and the
[42:09] Speaker 2: potential of the technology in AI was maxed out over
[42:13] Speaker 2: a few short years. So, yes, I fully agree. I
[42:16] Speaker 2: think we are still in the early innings. We've seen
[42:19] Speaker 2: rapid adoption in coding because, hey, you know, it turns
[42:22] Speaker 2: out that coders adopt technologies very quickly. We're only just
[42:26] Speaker 2: starting to see adoption in cybersecurity, in law, in healthcare,
[42:34] Speaker 2: in in finance, this very much still feels like day
[42:37] Speaker 2: one for artificial intelligence.
[42:41] Speaker 4: Just to clarify my data point, it was that among
[42:44] Speaker 4: Fortune 500 companies already spending at least 100,000 a year
[42:47] Speaker 4: with Anthropic, their Anthropic spending averaged less than 0.1% of
[42:52] Speaker 4: their total corporate expenses. That's my point. It's like, if
[42:55] Speaker 4: they're ahead in that game of enterprise, there's a lot
[42:58] Speaker 4: of the game left to play. You just talked about
[43:00] Speaker 4: specialist models. They did note as a sort of risk
[43:04] Speaker 4: factor that those focused on specialist models might represent a
[43:08] Speaker 4: risk to their models. Is that something that they'd have
[43:11] Speaker 4: to say, really?
[43:12] Speaker 2: It probably is something that you have to say.
[43:14] Speaker 3: I mean....
[43:16] Speaker 4: There are lots of potential future.
[43:17] Speaker 2: Versions for the foundational model technology. We believe that the
[43:23] Speaker 2: most likely one is that the frontier labs will drive
[43:27] Speaker 2: the overall gains and the overall progress in frontier technology.
[43:31] Speaker 2: And that intelligence will then manifest in different harnesses, different
[43:35] Speaker 2: application layers, that you will have open source models behind
[43:40] Speaker 2: or kind of intelligent enough models behind that sit behind.
[43:43] Speaker 2: But the question is not so much of one of volume,
[43:47] Speaker 2: it's one of value. And our investment in Anthropic is
[43:51] Speaker 2: very much predicated on the belief that the majority of
[43:55] Speaker 2: value will accrue to frontier intelligence.
[44:00] Speaker 4: I mean, you've made the case with conviction about that
[44:02] Speaker 4: on the show before. Kind of away from Anthropic, I
[44:06] Speaker 4: know this might seem sort of like, well, why would
[44:08] Speaker 4: you want to know this? But you run the private
[44:10] Speaker 4: companies team at Bailey Gifford, right? Could you just explain
[44:13] Speaker 4: what a big IPO would represent for the team and
[44:16] Speaker 4: then how it kind of gets handed off to the
[44:18] Speaker 4: rest of the firm if your portfolio is just the
[44:20] Speaker 4: private side?
[44:22] Speaker 2: So whilst I run the private companies business, we keep
[44:26] Speaker 2: our public and private investing activities deeply integrated. So if
[44:30] Speaker 2: you were to come and walk around the investment floor
[44:32] Speaker 2: here in Edinburgh, you'd see the private companies team sitting
[44:35] Speaker 2: right next to our North American equities team, which sits
[44:37] Speaker 2: right next to our long-term global growth team, and they're
[44:40] Speaker 2: investing in public companies. When a company goes public from
[44:44] Speaker 2: our private portfolios, some of those portfolios actually can own
[44:48] Speaker 2: for the long term into the public markets. Other portfolios
[44:51] Speaker 2: have a more limited ownership into the public markets. But
[44:55] Speaker 2: what we try to do as a private team is
[44:57] Speaker 2: make sure that our public teams are able to build
[45:00] Speaker 2: an understanding, build relationships with private company management teams so
[45:05] Speaker 2: that as and when those businesses do come to the
[45:07] Speaker 2: public markets, they are able to make the best possible
[45:10] Speaker 2: decisions for their public market clients based on that accrued
[45:15] Speaker 2: understanding and the accrued relationships with these companies.
[45:19] Speaker 4: And we showed the portfolio. My point is, Anthropic now,
[45:22] Speaker 4: but you've got Databricks, Anduril. These are the big names
[45:25] Speaker 4: in private markets right now. Peter Singlehurst, head of the
[45:27] Speaker 4: private companies team at Biddington. It's great to have you
[45:29] Speaker 4: back with us. Thank you very much indeed. Look, that
[45:33] Speaker 4: does it for this edition of Bloomberg Tech. But this
[45:35] Speaker 4: is what markets look like going into Friday. Intel is
[45:37] Speaker 4: the biggest drag. on a points basis at both index levels. Obviously,
[45:42] Speaker 4: it's on track for its biggest drop ever since records began.
[45:45] Speaker 4: And in Bloomberg's case, records begin in 1982. It gave
[45:48] Speaker 4: a very weak fourth quarter outlook and has a lot
[45:51] Speaker 4: of problems specific to it. But that is dragging down
[45:54] Speaker 4: the market. Amazon's also lowered despite, frankly, a pretty incredible
[45:58] Speaker 4: set of numbers. In the program, spoken to the AWS CEO,
[46:02] Speaker 4: Matt Garman, and AWS is growing at a level and
[46:06] Speaker 4: will continue, based on what the company said, grow at
[46:08] Speaker 4: a level that the market on balance would be impressed by.
[46:11] Speaker 4: And then next week... These are the numbers that we're
[46:15] Speaker 4: going to get from these different companies. And it starts
[46:17] Speaker 4: with Apple on Monday. Later in the week, AMD is
[46:21] Speaker 4: going to be important as a follow-on to Intel, Qualcomm,
[46:25] Speaker 4: a lot around the gig economy too, Airbnb, Lyft, Instacart.
[46:29] Speaker 4: We'll have full coverage across the board. Of course we will.
[46:32] Speaker 4: Really do recap the show. It's been an astonishing end
[46:34] Speaker 4: to the week. Do that on the podcast. You know
[46:36] Speaker 4: where to find it, on all the Bloomberg platforms and online.
[46:39] Speaker 4: From San Francisco, thanks for being with us. This is
[46:41] Speaker 4: Bloomberg Tech.

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.