# China's Moonshot, Netflix's Slump & Greylock's $1.5B Bet

An episode of [Bloomberg Tech](https://invisiblenetwork.ai/podcast/1611767434), 17 Jul 2026, 44 min.

Audio: https://podtrac.com/pts/redirect.mp3/traffic.omny.fm/d/clips/e73c998e-6e60-432f-8610-ae210140c5b1/41764a4f-fc64-4e11-89ba-ae7c0030ab5e/f2c5b397-6fe0-4278-9236-b48a0125ced2/audio.mp3?utm_source=Podcast&in_playlist=9caafc41-289c-4115-995d-ae7c0030ab75

Bloomberg’s Ed Ludlow looks at Chinese startup Moonshot's latest model, which it says can compete with the best from OpenAI and Anthropic. Plus, Netflix shares tumble after the streaming giant warns of slowing sales growth for a second straight quarter, and Greylock - one of the oldest venture firms - just raised $1.5 billion for its 18th fund. We speak with partner Saam Motamedi. See omnystudio.com/listener for privacy information.

## Transcript

[0:00] Speaker 1: Bloomberg Audio Studios, podcasts, radio news. Bloomberg Tech is live
[0:15] Speaker 1: from the heart of Silicon Valley with Ed Lao in
[0:19] Speaker 1: San Francisco.
[0:23] Speaker 2: This is Bloomberg Tech coming up. The global AI race
[0:26] Speaker 2: is accelerating, as Chinese startup Moonshot says its latest model
[0:30] Speaker 2: can compete with the best from Open Ai and m
[0:33] Speaker 2: propic plus, Netflix shares tumble up for the streaming giant
[0:36] Speaker 2: warns of slowing sales growth for a second straight quarter,
[0:40] Speaker 2: and Graylock, one of the oldest venture firms, just raised
[0:43] Speaker 2: one point five billion dollars for its eighteenth fund. We
[0:47] Speaker 2: speak with partner some modern medi Let's get straight to
[0:50] Speaker 2: our top story. Technology stocks are under pressure all around
[0:53] Speaker 2: the world this Friday, and the catalyst is a surprise
[0:57] Speaker 2: out of China. AI startup Moonshot has unveiled Kimmy K three,
[1:01] Speaker 2: a two point eight trillion parameter model it says can
[1:04] Speaker 2: compete with the best from Open Ai and Anthropic. It
[1:07] Speaker 2: charges about three dollars per million input tokens and fifteen
[1:12] Speaker 2: dollars per million output tokens, basically dramatically cheaper to use
[1:17] Speaker 2: than many leading AI models. Despite performing at the frontier,
[1:20] Speaker 2: investors are drawing parallels with last year's deepseek moment and
[1:25] Speaker 2: questioning whether the AI industry is enormous spending spree is
[1:28] Speaker 2: becoming harder to justify. Here's what's moving, and here's where
[1:33] Speaker 2: the pressure is. Right now, we're off session lows, but
[1:35] Speaker 2: the NASDAK one hundred down one point seven percent, the
[1:38] Speaker 2: Socks down almost three percent. The Socks is on track
[1:41] Speaker 2: for its biggest weekly decline since April of twenty twenty five.
[1:46] Speaker 2: Right now, it's trading at its lowest level since about
[1:49] Speaker 2: early to mid May. Then, as that one hundred is
[1:52] Speaker 2: not far off that weekly drop milestone either, about a
[1:54] Speaker 2: percentage point away from its biggest drop since April of
[1:59] Speaker 2: twenty twenty five. The individual movers, their compute names that
[2:03] Speaker 2: we're looking at one point, Apple is now down significantly
[2:08] Speaker 2: modestly two tens percent. It is very close, and at
[2:12] Speaker 2: one point did overtaking Video as Well's most valuable company.
[2:15] Speaker 2: But basically the market's under pressure generally. Let's get to
[2:18] Speaker 2: Bloomberg Senior Tech editor Mike Shephard, and let's talk about
[2:21] Speaker 2: China and AI. I mean, let's start with Moonshot and
[2:24] Speaker 2: Kimmy K three. I will try to get as much
[2:26] Speaker 2: of the detail of this release as possible and the
[2:29] Speaker 2: market reactions clear. But what else do we know?
[2:33] Speaker 3: Well, what we know is that this really does put
[2:36] Speaker 3: China on the map once again with the model that
[2:39] Speaker 3: rivals some of the best coming from Silicon Valley, and
[2:42] Speaker 3: it also echoes in a way if you ask around
[2:45] Speaker 3: town here in Washington, certainly among some of the companies
[2:48] Speaker 3: here in the US, that some of these models that
[2:51] Speaker 3: China is putting forth may have been built on the
[2:54] Speaker 3: backs of this practice known as distillation, where results are
[2:58] Speaker 3: harvested from the outs of American AI models to build
[3:03] Speaker 3: this rival generation of chatbots in China. On the chief,
[3:06] Speaker 3: China has rejected a lot of those claims, and yet
[3:09] Speaker 3: it certainly is a friction point between the US and
[3:13] Speaker 3: China in this broader global race to dominate the emerging
[3:17] Speaker 3: technology of AI.
[3:19] Speaker 2: Chinese President Jujingpang has been out staking his claim of
[3:25] Speaker 2: the leadership in the field of AI, talking about China's
[3:29] Speaker 2: domestic prowess and AI, but also kind of taking policy
[3:32] Speaker 2: action to support what we see from Moonshalt. What do
[3:36] Speaker 2: we need to know about what President g has been
[3:38] Speaker 2: saying this week.
[3:41] Speaker 3: Well, really earlier today, he made a debut appearance, his
[3:45] Speaker 3: first ever at the Shanghai World AI Forum. This is
[3:50] Speaker 3: an event held annually there, but this really signified his
[3:54] Speaker 3: personal stamp on China's push to lead in AI. Earlier
[3:59] Speaker 3: this year, the government there had agreed to put as
[4:02] Speaker 3: much as two hundred and ninety five billion dollars toward
[4:05] Speaker 3: building a nationwide interconnected network of computing hubs data centers
[4:11] Speaker 3: to really power the nationwide push to ensure that there
[4:15] Speaker 3: is enough compute to go around to support all of
[4:18] Speaker 3: these startups that are developing models like Kimmy and others
[4:22] Speaker 3: that we have been talking about on this program. And
[4:25] Speaker 3: then she is also making this not just a message
[4:28] Speaker 3: about China, but about the world. He is calling for
[4:31] Speaker 3: AI to be something distributed more equitably around the world.
[4:35] Speaker 3: He pitched this message to the Global South, saying he
[4:39] Speaker 3: wants to see AI for all countries, not just the
[4:42] Speaker 3: richest and most powerful, and that really played to concerns
[4:45] Speaker 3: that we have heard from decades from developing nations that
[4:48] Speaker 3: they are the last to bear the fruits of any
[4:51] Speaker 3: big technological advance and AI. Many countries in the Global
[4:55] Speaker 3: South have been watching advances by the US and by
[4:58] Speaker 3: China wondering when they would get a p the action,
[5:00] Speaker 3: and she is making an offer to them that look,
[5:02] Speaker 3: if you sign up with us, you will you will
[5:06] Speaker 3: get your fair share. He is also calling for some
[5:09] Speaker 3: guardrails on the technology. He sees just like officials here
[5:13] Speaker 3: in the US do some of the security risks potentially
[5:16] Speaker 3: not only to China's national security, but the global security.
[5:19] Speaker 2: But he is also calling for cooperation with.
[5:22] Speaker 3: The United States and other countries when it comes to
[5:26] Speaker 3: development of AI and promoting it worldwide.
[5:28] Speaker 2: Ed BLUEBGGS, Mike Shepherd in DC, thank you very much.
[5:32] Speaker 2: The surprise release of Kimmi K three. It is a
[5:35] Speaker 2: massive market story and throughout the hour we're going to
[5:38] Speaker 2: get different perspectives on what this means for American AI.
[5:42] Speaker 2: Will continue to track the market moves, but right now
[5:44] Speaker 2: you see the Nazak one hundred and in particular semiconductors
[5:48] Speaker 2: under pressure. That's in the equity markets. The AI trade
[5:51] Speaker 2: isn't just hitting stocks, it's weighing on bonds too. Debt
[5:55] Speaker 2: sold by tech giants to fund massive AI investments is
[5:59] Speaker 2: under performing as investors grow more cautious about the sectors
[6:03] Speaker 2: spending spree. Bloomberg's Tasos Vossos joins us for more like
[6:07] Speaker 2: look the bond market. This predates the headline overnight right
[6:11] Speaker 2: on Kimik three. But if you look at portfolio performance,
[6:16] Speaker 2: we made a lot of news headlines about the issuance
[6:19] Speaker 2: of the debt, the need for capital. Tell us how
[6:22] Speaker 2: those that's gone for those bonds in the interim period.
[6:27] Speaker 4: And that's the thing. A few months ago people were
[6:29] Speaker 4: already warning the debt is building too fast within portfolios,
[6:34] Speaker 4: within indices. What if something actually goes wrong, And as
[6:37] Speaker 4: you said before, it's not driven by the same thing,
[6:39] Speaker 4: which actually makes it rather confusing for investors who may
[6:43] Speaker 4: have both equities and bonds in their portfolio, some of
[6:46] Speaker 4: them issued by the same AI company at Hyperscala, for example.
[6:50] Speaker 2: Because of the credit market.
[6:52] Speaker 4: The main concern now is whether the debt is being
[6:55] Speaker 4: built up too fast and by too much more than
[6:57] Speaker 4: the market can actually absorb, because obviously the AI races
[7:01] Speaker 4: is heating up and you cannot possibly lose out. You
[7:05] Speaker 4: have to spend a load of money and you have
[7:07] Speaker 4: to actually raise it from every corner of the all
[7:09] Speaker 4: you can find, which is why now that we're seeing
[7:11] Speaker 4: a downturn in bonds that are leading losses is actually
[7:15] Speaker 4: happening across the world and not just in the United States.
[7:19] Speaker 2: Tells us, I'm going to read you some of the
[7:20] Speaker 2: data from your own report, if that's okay. Seventy nine
[7:24] Speaker 2: percent of the sector's bonds sold since early twenty twenty
[7:28] Speaker 2: five are now currently indicated at a wider spread compared
[7:32] Speaker 2: to the first day of trading. Explain that to us.
[7:35] Speaker 2: But the other point being that price moving inversity yield,
[7:40] Speaker 2: the price have fallen from the issue price on average
[7:43] Speaker 2: across that that bucket. What's going on?
[7:46] Speaker 4: Well, the reason why we're using so many metrics is
[7:49] Speaker 4: because different investors may be using different metrics to benchmark
[7:52] Speaker 4: themselves and explain to their you know, their investors whether
[7:56] Speaker 4: they are actually performing well or not. But whichever metric
[7:59] Speaker 4: you use on the air AI bonds now, whether it's spread,
[8:01] Speaker 4: whether it's surprized, whether it's total return that looks into
[8:04] Speaker 4: interest income as well, they're all negative. And what that
[8:07] Speaker 4: shows is that whichever your approach has been towards that
[8:10] Speaker 4: sector actually hasn't worked out. Whether you try to eliminate
[8:13] Speaker 4: one one risk or another, either way you're in the red.
[8:17] Speaker 4: So what the number of active investors have been telling
[8:19] Speaker 4: us is that, actually, do you know what, I don't
[8:21] Speaker 4: even want to hold them at all, because unlike stocks,
[8:24] Speaker 4: where if the AI future is as bright as some
[8:28] Speaker 4: people pull tend to be and you end up holding
[8:30] Speaker 4: an AI winner, then the stock can go through the roof.
[8:34] Speaker 4: The problem is with the bonds because of the mechanics
[8:36] Speaker 4: getting paid face value at part that's not going to happen,
[8:39] Speaker 4: So the upside is actually rather limited by the downside
[8:42] Speaker 4: could be really significant.
[8:45] Speaker 2: Bloomberg tass rs us with the credit market some of
[8:48] Speaker 2: what happening in AI and hyperscalers, thank you very much. Indeed,
[8:52] Speaker 2: coming up another top story, Netflix is under pressure. The
[8:55] Speaker 2: streaming giant is warning of slower sales growth and we're
[8:58] Speaker 2: going to speak with hell Wang of Philip Securities to
[9:01] Speaker 2: try and work out what the story is here with Netflix.
[9:04] Speaker 2: That's next. This is Bloomberg Tech.
[9:13] Speaker 5: We believe it takes a great artists to make something great,
[9:16] Speaker 5: and AI is not changing that. AI will give creators
[9:19] Speaker 5: better tools to bring their visions to life. Movies are
[9:22] Speaker 5: being made by people who make movies. AI provides them
[9:25] Speaker 5: with better tools to make them even better.
[9:29] Speaker 2: That was Netflix co CEO Ted Serrandos, speaking on yesterday's
[9:33] Speaker 2: earnings call, look at Netflix shares at one point in
[9:37] Speaker 2: the session, We're on track for our biggest drop in
[9:39] Speaker 2: four years, down more than eight percent. The company outlined
[9:43] Speaker 2: a bigger role for AI and pointed to advertising as
[9:47] Speaker 2: its next major growth engine, But investors are beginning to
[9:50] Speaker 2: worry that the company's breakneck growth is starting to slow. Specifically,
[9:54] Speaker 2: it's slowing for a second straight quarter with decelerating revenue gains.
[9:59] Speaker 2: Joining us now is Helena Wong. She has a by
[10:01] Speaker 2: rating on Netflix. They're looking for a story beyond subscriber growth, right.
[10:08] Speaker 2: It was so interesting to go through how they'll use AI, advertising,
[10:13] Speaker 2: live content. We'll go over all three. But for you, Helena,
[10:16] Speaker 2: what was the main takeaway from this Netflix print and
[10:19] Speaker 2: what was said on the call?
[10:21] Speaker 6: Well, I guess if you just look at a quarter
[10:23] Speaker 6: results itself, I will say it's still overall broadly in line.
[10:28] Speaker 6: So membership trends has been really healthy. They recently increased
[10:32] Speaker 6: their price again earlier this year. In the markets like
[10:35] Speaker 6: the US, Mexico and Spain. It continues to be well accepted.
[10:39] Speaker 6: The air revenue continues to suspend. It's on track to
[10:41] Speaker 6: double its revenue this year, so that helps us the
[10:44] Speaker 6: money tandation. So margins are holding up fairly well as well,
[10:48] Speaker 6: so I would say the results is actually pretty healthy.
[10:51] Speaker 2: Yeah, you know how long did Netflix say, judge us
[10:54] Speaker 2: by traditional financial metrics and you look at those metrics
[10:57] Speaker 2: pretty good. They're talking about how Netflix will be different
[11:01] Speaker 2: as well, right the platform, I mean more live content,
[11:04] Speaker 2: live sports in particular, video podcasts, creators, cloud gaming. I'm
[11:10] Speaker 2: looking at that and saying, I watch Netflix in the evening,
[11:14] Speaker 2: sit down, stream a show, a film. What about the
[11:17] Speaker 2: rest of the day. To me, it seems like Netflix
[11:19] Speaker 2: is thinking about how do we get into that dressable
[11:22] Speaker 2: market for people's eyeballs morning through evening. Is that what
[11:25] Speaker 2: you see?
[11:27] Speaker 7: Yeah?
[11:27] Speaker 6: I think another challenge that they're seeing is they're facing
[11:31] Speaker 6: a lot of competition that is not just from the
[11:33] Speaker 6: traditional streaming platform, So they are competing a lot of
[11:37] Speaker 6: the screen time with short video subscribers, so services like
[11:42] Speaker 6: YouTube shorts and TikTok videos, because consumer behavior is now changing,
[11:47] Speaker 6: so now every time you go home, you're not just
[11:50] Speaker 6: oning a TV, You're spending a lot of time on
[11:52] Speaker 6: your phone as well. So they're definitely competing with this
[11:56] Speaker 6: limited screen time that people have, so for them, they
[11:59] Speaker 6: are also they're trying to strengthen their engagement. They're trying
[12:03] Speaker 6: to do a lot of the life events, a lot
[12:05] Speaker 6: of the podcasts, and a lot of the mobile features.
[12:07] Speaker 6: But it's definitely going to be a long term challenge
[12:09] Speaker 6: for them.
[12:11] Speaker 2: They're talking about using AI in three different ways, improving
[12:16] Speaker 2: the product, lowering costs, and expanding margins. You talked about
[12:21] Speaker 2: the margins point earlier for the stock. What's most meaningful
[12:25] Speaker 2: for Netflix to communicate on the on their use of AI.
[12:30] Speaker 6: I think the use of AI is inevidable because it's
[12:33] Speaker 6: now just expanding to all aspects of the industry. I
[12:38] Speaker 6: think the reason that the market is reacting very negatively
[12:41] Speaker 6: is mainly still because that they're giving a relatively softer
[12:45] Speaker 6: guidance for their startcolder revenue, and I would say it's
[12:49] Speaker 6: still considered pretty healthy growth. It just compared to the
[12:52] Speaker 6: growth that we haven't seen earlier, it is considered slightly softer.
[12:56] Speaker 6: Some Netflix is a business where everything is going right
[13:00] Speaker 6: lay the investor's expectation for it is just relatively high,
[13:04] Speaker 6: so they're really not tolerating any of the minor disappointment
[13:08] Speaker 6: that coms it. I think another reason that it's not
[13:11] Speaker 6: doing well is because the content slate is not as
[13:14] Speaker 6: compelling compared to last year. So this year they have
[13:17] Speaker 6: some exciting.
[13:19] Speaker 2: Sorry to interrupt you, Helena, I'm so glad you went
[13:22] Speaker 2: to the content slate, right. I get you have to
[13:24] Speaker 2: model on certain metrics the growth of this business, but
[13:28] Speaker 2: like what are you watching right now? It still comes
[13:31] Speaker 2: down to content, is king? How's Netflix doing on that front?
[13:36] Speaker 6: Yes, I would say their content slate is definitely not
[13:39] Speaker 6: exciting when you're compared to last year. So this year
[13:42] Speaker 6: they do have some exciting coming up, like seventy two
[13:44] Speaker 6: Hours with Kevin Hart, and they have some NFL games
[13:48] Speaker 6: lining up, But compared to last year, they had Wednesday,
[13:52] Speaker 6: they had Strangers since they had all those exciting contents,
[13:56] Speaker 6: they had Squeaking, so that was all. The are very
[14:00] Speaker 6: popular franchise with a lot of hardcore fans, so they
[14:04] Speaker 6: actually managed to bring a lot of subscribers onto the
[14:06] Speaker 6: platform because of that. So this year, compared to that,
[14:09] Speaker 6: it is just not very exciting. And for Netflix, it's
[14:12] Speaker 6: a company that has a lot of high expectations, so
[14:16] Speaker 6: investors are constantly looking for science that there's going to
[14:20] Speaker 6: be reacceleration in the business. So for this sector, we're
[14:23] Speaker 6: not really.
[14:23] Speaker 2: Seeing that, Helena. We just showed Lucas Shaw, who leads
[14:27] Speaker 2: our coverage of screen time the Entertainment Industry's column prior
[14:31] Speaker 2: to earnings, and he was writing about how Netflix can't
[14:35] Speaker 2: get its audience to stick with the show, so they
[14:38] Speaker 2: start and then they don't stick with it. Why does
[14:40] Speaker 2: that matter?
[14:42] Speaker 6: Well, it does matter because at the end of the day,
[14:45] Speaker 6: it really comes down to your ability to persuade your
[14:50] Speaker 6: customer to stay on your platform. But I will say
[14:53] Speaker 6: right now, I don't see very clear evidence that their
[14:56] Speaker 6: engagement is deteriorating. Actually, their engagement is improving compared to
[15:01] Speaker 6: twenty twenty five according to their Engagement report, So I
[15:05] Speaker 6: would say right now, there's no clear sign that they're
[15:08] Speaker 6: losing on that, and there's still very strong pricing power.
[15:11] Speaker 6: So overall, I would still say we're very very positive
[15:14] Speaker 6: on Netflix, especially with the reason price correction. So previously
[15:18] Speaker 6: the premium valuation largely limit the upside that we have
[15:21] Speaker 6: in our vieuse, So right now is the stock foolback.
[15:23] Speaker 6: We do think it is a good opportunity to actually
[15:26] Speaker 6: start runing a.
[15:27] Speaker 2: Pollution hell and along with Philip Securities back on Bloomberg Tech.
[15:31] Speaker 2: Thank you very much, all things Netflix coming up. There's
[15:34] Speaker 2: a setback for Google's AI ambitions. Why Gemini's latest delay
[15:39] Speaker 2: has invested this question in whether the company is losing
[15:41] Speaker 2: ground in the AI race. This is Bloomberg Tech.
[15:56] Speaker 8: It's time now for talking tech. I'm you Hira on
[15:59] Speaker 8: on first step. Ze to Ai, also known as Jipu,
[16:02] Speaker 8: is on track to be China's first AI firm, hitting
[16:05] Speaker 8: one billion dollars in annual sales. The AI startup achieved
[16:08] Speaker 8: its full year sales target in July, marketing a milestone
[16:11] Speaker 8: for a company aiming to distinguish itself as a provider
[16:15] Speaker 8: of AI to companies and enterprises. Plus, Data Bricks is
[16:19] Speaker 8: in its second round of financing this year and is
[16:22] Speaker 8: seeking capital from investors led by Coaching Managements at a
[16:25] Speaker 8: one hundred and eighty eight billion dollar valuation that would
[16:28] Speaker 8: mark a forty percent jump from the one hundred and
[16:31] Speaker 8: thirty four billion it achieved earlier this year. Data Bricks,
[16:34] Speaker 8: of course, has long been regarded as an attractive candidate
[16:37] Speaker 8: to IPO and Chinese Sorry Japanese memory chip maker Kiyoxia
[16:43] Speaker 8: has now lost more than half its market value from
[16:46] Speaker 8: its June peak. At that point, the company was Japan's
[16:50] Speaker 8: highest valued firm. The sell off coming on growing concerns
[16:53] Speaker 8: AI driven rally has gone too far, something we are
[16:56] Speaker 8: seeing again today.
[16:58] Speaker 2: Ed okay, thank you. The AI race is moving so
[17:01] Speaker 2: fast that even a few months can make a big difference.
[17:05] Speaker 2: Bloomberg's reporting Google is months behind on the release of
[17:08] Speaker 2: its next flagship AI model, Gemini three point five, pro
[17:12] Speaker 2: raising new questions about whether it's falling behind rivals Open
[17:16] Speaker 2: Ai and Amthropic. Bloomberg Intelligence analyst man Deep Singh, writing
[17:20] Speaker 2: a reaction piece to the report that this could be
[17:22] Speaker 2: a sign that recent high profile departures have hurt frontier
[17:26] Speaker 2: model development, and Mandeep joins us Now, it was an
[17:30] Speaker 2: interesting and detailed report from Bloomberg News, but I like
[17:33] Speaker 2: the thesis presented in the react go a little bit
[17:36] Speaker 2: deeper into it. Well, we've seen you.
[17:40] Speaker 7: Fabrile five and the latest OpenAI model clearly have had
[17:45] Speaker 7: a step up in terms of functionality and coding agencies,
[17:49] Speaker 7: where Google seems to have missed the boat with Gemini.
[17:53] Speaker 7: So I would have expected Gemini three to be released sooner,
[17:57] Speaker 7: and now we're talking about Gimmy three and how you
[18:00] Speaker 7: know that head.
[18:08] Speaker 2: And a step. I'm going to jump in here, guys,
[18:10] Speaker 2: because I think the man Deep's zoom is hitting some
[18:14] Speaker 2: technical issues and maybe the control room just hit mute
[18:17] Speaker 2: on him for a second. Sorry about that, Man Deep.
[18:19] Speaker 2: We'll come back to the story that the race and
[18:21] Speaker 2: model development. Obviously today with Kimmy K three, it's our
[18:25] Speaker 2: top story. Let's get to Apple. Apple's product refresh is
[18:28] Speaker 2: far from over. The company is now preparing a sweeping
[18:32] Speaker 2: update to its iPad lineup. Bloomberg's learned Apples planning its
[18:35] Speaker 2: biggest iPad Mini overhaul in five years, complete with an
[18:39] Speaker 2: O lead display. Bloomberg's Apple and Consumer Tech Managing editor
[18:43] Speaker 2: Mark Duman joins us with the details spare of forought
[18:46] Speaker 2: for iPad Mini. It's been a while since I think
[18:48] Speaker 2: you and I have talked about that product line a
[18:51] Speaker 2: lot of detail in the new strategy. What do we
[18:53] Speaker 2: need to know?
[18:54] Speaker 9: Yeah, Apple's got a jam packed product portfolio over the
[18:57] Speaker 9: next few years, and part of that is a big
[19:00] Speaker 9: overhaul of the iPad lineup. And so this fall you're
[19:03] Speaker 9: going to see one of the biggest ever updates to
[19:06] Speaker 9: the iPad Mini. It's getting an OLED screen, which means
[19:09] Speaker 9: the display quality is now going to be up there
[19:11] Speaker 9: with an iPhone or an iPad pro. This has been
[19:14] Speaker 9: a long time coming. This is something that tech enthusiasts
[19:17] Speaker 9: have been clamoring for.
[19:18] Speaker 2: You're also going to.
[19:19] Speaker 9: See updates to all the other iPads next year that
[19:22] Speaker 9: includes faster iPad pros, new iPad airs in a new
[19:26] Speaker 9: entry level iPad. In the backdrop here is that Apple
[19:30] Speaker 9: has made all of the iPads more expensive in recent
[19:32] Speaker 9: weeks because of the memory shortage. Right, some of them
[19:35] Speaker 9: went up by one hundred dollars, some of them went
[19:37] Speaker 9: up one hundred and fifty, some of them went up
[19:39] Speaker 9: by two hundred dollars. So now you need upgrades to
[19:42] Speaker 9: the product to make them more worthwhile in this competitive
[19:46] Speaker 9: environment where you have great tablets from other companies like Amazon,
[19:50] Speaker 9: Samsung and the.
[19:51] Speaker 2: Like marks this is the first opportunity to you. And
[19:54] Speaker 2: I've had to discuss Apple's AI plan for China. There
[20:00] Speaker 2: was some developments this week in how they're able to
[20:03] Speaker 2: put an AI product into that market. Could you explain
[20:06] Speaker 2: the basics of it to us?
[20:08] Speaker 9: So right now, Apple Intelligence inside of the United States
[20:13] Speaker 9: and the rest of the world for the most part,
[20:15] Speaker 9: uses models from Apple on device, models from Apple in
[20:19] Speaker 9: the cloud, and then there's partners for some features that
[20:22] Speaker 9: Apple doesn't do so well, so search they work with Google,
[20:26] Speaker 9: and then they work with open Ai as sort of
[20:28] Speaker 9: a fallback for Siri. If you ask it's something it
[20:31] Speaker 9: doesn't know the answer to, it'll query chat GBT if
[20:35] Speaker 9: you allow it to. So for China, there's a few nuances.
[20:38] Speaker 9: For search, they're working with Baidu for that chatbot, the
[20:41] Speaker 9: open Ai replacement, they're working with Baidu. Obviously open Ai
[20:45] Speaker 9: Chat gybt is not available in that part of the world.
[20:49] Speaker 9: But then there's the Ali baba component, and so the
[20:52] Speaker 9: phones in China will continue to use the on device
[20:55] Speaker 9: Apple models, but there will be a layer on top
[20:58] Speaker 9: using Ali Baba ai technology that I call a censorship filter.
[21:03] Speaker 9: And so what this will do is we'll keep the
[21:05] Speaker 9: phone in constant connection with the Chinese government, so as
[21:09] Speaker 9: there's updated models that are hitting the iPhone, Apple of
[21:12] Speaker 9: course is always updating its models in the background. Those
[21:14] Speaker 9: will need to be approved every so often with the
[21:17] Speaker 9: Chinese government. And this type of censorship filter is what
[21:20] Speaker 9: other companies in China who operate have to do show
[21:24] Speaker 9: me Huiwei the other phone providers with AI technology on
[21:27] Speaker 9: their phones.
[21:29] Speaker 2: All right, most mark German with everything that's happened this week,
[21:32] Speaker 2: Actually not everything. There was a lot more with Apple too,
[21:34] Speaker 2: but thank you very much. Coming up, just a little
[21:37] Speaker 2: over a month after its blockbuster IPOs, SpaceX is experiencing
[21:41] Speaker 2: a few setbacks. There was a scrub start ship launch
[21:44] Speaker 2: last night. Stock reacted, but the slock generally is under pressure.
[21:48] Speaker 2: We can have more in that story next. A reminder. Overnight,
[21:52] Speaker 2: China's Moonshot released Kimmi K three, an open weighted model
[21:56] Speaker 2: that has top benchmarks. But there's a big focus on
[22:00] Speaker 2: training costs and token economics with this as well. Right now,
[22:05] Speaker 2: the market's asking itself, do we need to reconsider all
[22:09] Speaker 2: of the money that is going into AI from an
[22:12] Speaker 2: American perspective? And do we need to think again about
[22:17] Speaker 2: what the MOTI is for some of these American companies.
[22:21] Speaker 2: We're going to have that conversation next. Now, if I'm lucky,
[22:25] Speaker 2: We're going to cut some beautiful pictures of San Francisco.
[22:28] Speaker 2: It's half time. We'll be right back. This is Bloomberg
[22:30] Speaker 2: Tech Welcome back to Bloomberg Tech. All across the AI trade.
[22:56] Speaker 2: US technology stocks are under pressure then, as that one
[22:59] Speaker 2: hundred and Socks both down in the session significantly, but
[23:03] Speaker 2: on the week the Socks is headed for its biggest
[23:05] Speaker 2: weekly drop since April of twenty twenty five. The NATSAK
[23:09] Speaker 2: one hundred is not far off that about a percentage
[23:12] Speaker 2: point from being in the realm of the same milestone
[23:15] Speaker 2: China's moonshot with a surprise release of Kimmy K three.
[23:19] Speaker 2: The market recalculating a lot of the spending that has
[23:23] Speaker 2: is and will happen on AI development and the infrastructure
[23:28] Speaker 2: side in video biggest points drag Intel down significantly. One
[23:33] Speaker 2: story which is kind of chopped and changed is that
[23:36] Speaker 2: at one moment in the session Apple overtook in video
[23:40] Speaker 2: as the world's most valuable company. We're going to do
[23:42] Speaker 2: the math on that a little bits time. It's close,
[23:45] Speaker 2: but again we're a few hours from the session being
[23:47] Speaker 2: closed overnight. Space SpaceX's Starship rocket mission failed to launch yesterday.
[23:54] Speaker 2: It would have been its thirteenth Starship test flight. Right
[23:57] Speaker 2: at liftoff, blooms of smoke or steam from the launch tower.
[24:03] Speaker 2: There was an engine failure. This isn't the first time.
[24:05] Speaker 2: Starship previously had delays and malfunctions. But what CEO Elam
[24:09] Speaker 2: Musk said on X is that the next launch attempt
[24:12] Speaker 2: would hopefully come in the next few days. What was
[24:14] Speaker 2: interesting about it in the moment in after ours trading,
[24:18] Speaker 2: the stock immediately reacted and there was a decline of
[24:20] Speaker 2: about four percent already in the session yesterday, SpaceX's shares
[24:26] Speaker 2: a dropped below its IPO price. The company's facing turbulence
[24:30] Speaker 2: just a little over a month after its historic IPO
[24:33] Speaker 2: and dampening euphoria potentially for newly public companies joining US now.
[24:39] Speaker 2: Is Bloomberg's Anthony Hughes. This is where we stand. We're
[24:42] Speaker 2: showing it on the screen June twelfth to present day.
[24:46] Speaker 2: What has been the market's digestion of the biggest IPO
[24:51] Speaker 2: of all time?
[24:52] Speaker 10: Yes, I mean, after all, the hyperspace is going public
[24:55] Speaker 10: last month. I think we've seen a bit of a
[24:57] Speaker 10: hangover in the IPO market really, with the momentum trade
[25:01] Speaker 10: also coming off here at the same time the AI
[25:03] Speaker 10: trade coming off, you know, but it's not unusual that
[25:06] Speaker 10: a big IPO might sort of pull back here given that,
[25:11] Speaker 10: you know, some of the technical dynamics. You know, once
[25:17] Speaker 10: the price price sort of settles down and the market
[25:21] Speaker 10: settles down, that the stocks come back. So I mean,
[25:24] Speaker 10: obviously it's a bit concerning for investors that SpaceX is
[25:28] Speaker 10: already trading at a discount.
[25:30] Speaker 2: I think it's about three percent.
[25:31] Speaker 10: To the IPO price, because we still have all this
[25:33] Speaker 10: stock to come out from out of lock up as well.
[25:35] Speaker 10: So you know, but overall, the IPO market has lost
[25:39] Speaker 10: a lot of steam here because of mainly because of SpaceX,
[25:42] Speaker 10: and you know, the average return from IPOs this year
[25:45] Speaker 10: has really evaporated to zero.
[25:48] Speaker 2: I would just again meet the point that SpaceX being
[25:50] Speaker 2: down five percent, and this Friday session is in part
[25:53] Speaker 2: because of the scrubbed Starship test thirteen last night. You know,
[25:58] Speaker 2: we're measuring the success from IPO based on the first
[26:00] Speaker 2: thirty days of trade. But for the bankers, you know,
[26:04] Speaker 2: and and the investors that participated, this was a success, right, Anthony,
[26:09] Speaker 2: that they would say that they pulled this off.
[26:11] Speaker 10: Well, yeah, I think at the time when SpaceX went public,
[26:15] Speaker 10: I think it was it was it was a very
[26:17] Speaker 10: well executed process. I think that you know, as as
[26:22] Speaker 10: time goes on, the stock's going to trade more on
[26:24] Speaker 10: the you know, the news, the fundamentals and news news
[26:28] Speaker 10: that the news flow here, and obviously there's been some
[26:30] Speaker 10: bad news flow initially here. But you know, it's a
[26:33] Speaker 10: long for a lot of the a lot of investors
[26:36] Speaker 10: that got behind SpaceX. It's a long term proposition, and
[26:39] Speaker 10: you know, anyone who's familiar with what happened with Tesla
[26:42] Speaker 10: knows that there's plenty of ups and downs along along
[26:44] Speaker 10: the way. But obviously the investors will be hoping that
[26:46] Speaker 10: longer term the prices performs.
[26:49] Speaker 2: A lot better. Bloomberg's Anthony Hughes on SpaceX. Thank you
[26:53] Speaker 2: very much from SpaceX. So the next generation of space startups.
[26:57] Speaker 2: Texas is fast becoming the capitol of the commercial space industry.
[27:02] Speaker 2: So what's driving the boom and what does it mean
[27:05] Speaker 2: for the state's economy. Bloomberg originals has the story.
[27:10] Speaker 11: For most of space white history, NASA and government or
[27:15] Speaker 11: nation states were really the biggest players in charge. That
[27:20] Speaker 11: has recently shifted.
[27:22] Speaker 2: This was in front here.
[27:23] Speaker 12: We really believed for a long time it was only
[27:25] Speaker 12: a place for governments to build and innovate.
[27:28] Speaker 6: What we're seeing is a democratized space sector with a
[27:31] Speaker 6: lot of smaller players and a lot of larger commercial
[27:34] Speaker 6: players as well.
[27:36] Speaker 13: With the global space economy currently valued it over six
[27:39] Speaker 13: hundred and fifty billion dollars, one American state in particular
[27:43] Speaker 13: is betting big on this rapidly growing sector.
[27:46] Speaker 3: Governor Abbat, coming to you from the great Data Technis Texas.
[27:50] Speaker 11: Is a really good snapshot of the new direction that
[27:54] Speaker 11: the space industry is taking. The biggest tenant in Texas
[27:59] Speaker 11: is Space, so they have set up shop with their
[28:02] Speaker 11: Star based facility on the southern tip of Texas. Another
[28:05] Speaker 11: big tenant is Blue Origin in West Texas outside of
[28:09] Speaker 11: Van Horn. And then you also have a big cluster
[28:13] Speaker 11: of companies in Houston, the home of NASAs Johnson Space Center,
[28:16] Speaker 11: so you have Axiom Space, you have Intuitive Machines. And
[28:20] Speaker 11: then in Central Texas, Firefly Aerospace has its rocket facility
[28:24] Speaker 11: and development center.
[28:28] Speaker 13: Last year, the Texas Space Commission gave out one hundred
[28:31] Speaker 13: and fifty million dollars in grants to support space business
[28:34] Speaker 13: in the state. It's due to double that by the
[28:37] Speaker 13: end of this year.
[28:40] Speaker 14: What we wanted to do is make it well known
[28:43] Speaker 14: that those companies who had already decided to be set
[28:46] Speaker 14: up here in Texas that they didn't leave, and that
[28:49] Speaker 14: they are also helped to expand their existing footprint and.
[28:52] Speaker 9: Lift off the crew of Artimists two now bound for
[28:56] Speaker 9: the room.
[28:58] Speaker 14: From a NASA standpoint, because of that competition, they're lowering
[29:02] Speaker 14: the overall costs that allows us to be able to
[29:05] Speaker 14: do more.
[29:06] Speaker 5: Humanity's next great voyage begins.
[29:11] Speaker 2: Or you can watch the full Bloomberg original story online
[29:14] Speaker 2: and of course on the Bloomberg terminal. Let's get back
[29:16] Speaker 2: to Apple, who at one point surpassed and Video this
[29:20] Speaker 2: morning is the world's most valuable company, likely benefiting from
[29:23] Speaker 2: this broader rotation in the tech sector. Out of the
[29:27] Speaker 2: big spenders on AI joining us as Bloombos, Ryan Verslica.
[29:31] Speaker 2: I keep glancing down at my terminal. I think, you know,
[29:34] Speaker 2: we're changing hands a few times here, but I think
[29:36] Speaker 2: in Video is now slightly back ahead. But that's been
[29:39] Speaker 2: the story, right. Apple doesn't have the big capital expenditure story.
[29:44] Speaker 2: It doesn't have the big dollar story around AI. Video
[29:47] Speaker 2: has been the principal beneficiary of everything in the last
[29:50] Speaker 2: four years.
[29:52] Speaker 12: That's what's very interesting about this trade rate. Now what
[29:55] Speaker 12: used to be seen as a real risk for Apple
[29:58] Speaker 12: that it wasn't aggressively pursuing AI and the way that
[30:01] Speaker 12: some of the other big cap tech companies are. That's
[30:03] Speaker 12: now become a real asset. It doesn't have the sort
[30:06] Speaker 12: of capex risk that's being priced into Microsoft.
[30:09] Speaker 2: To give sort of the.
[30:12] Speaker 12: Typical example there, it's really been benefiting sort of as
[30:15] Speaker 12: a safe haven as we see the chips face unwind
[30:18] Speaker 12: as we see more questions right the hyperscaler side of
[30:21] Speaker 12: the market, it's really been pushing higher. I think it's
[30:23] Speaker 12: up more than twenty percent in a couple of weeks.
[30:26] Speaker 12: That's a huge gain for a company of this size,
[30:28] Speaker 12: and it's really because of its own unique path that
[30:31] Speaker 12: it's charting within AI. I would also add that it
[30:34] Speaker 12: does have the foldable iPhone that's expected to come out
[30:37] Speaker 12: later on this year. It does have the agentic AI
[30:40] Speaker 12: SERI Digital Assistant. There's been some disappointments with that, but
[30:44] Speaker 12: there is optimism that we are going to start to
[30:46] Speaker 12: see a big iPhone upgrade cycle going up that's going
[30:49] Speaker 12: to help overshadow the increase in high memory chip prices.
[30:53] Speaker 12: But in general, it is seem like the fact that
[30:55] Speaker 12: Apple isn't participating in AI in the same way that
[30:58] Speaker 12: some of its peers are a real positive and thing
[31:01] Speaker 12: that is distinguishing the company.
[31:04] Speaker 2: In the here and now. I see the Philadelphia Semiconductor
[31:07] Speaker 2: Index on the week, on track for its biggest weekly
[31:09] Speaker 2: drops since April twenty twenty five. We've talked a lot
[31:13] Speaker 2: about Moonshot and Kimmy K three, A lot more to
[31:15] Speaker 2: come in the show on that, But you and the
[31:17] Speaker 2: team writing that the SOCKS is set to fall into
[31:20] Speaker 2: a bear market. That's big.
[31:23] Speaker 12: Yeah, there's been a real reversal in momentum in the
[31:26] Speaker 12: chip space. I think there are some concerns about you
[31:28] Speaker 12: know this. The index is up. I think at some
[31:30] Speaker 12: point that was nearly double year today, so a huge
[31:33] Speaker 12: gain there and still up quite significantly for this year,
[31:36] Speaker 12: even though it has pulled back some. But I think
[31:38] Speaker 12: there is a little bit of profit taking, a little
[31:39] Speaker 12: bit of reversal in momentum. And as we are discussing
[31:43] Speaker 12: you know, this whole question about hyperscale or ROI, that
[31:47] Speaker 12: is really feeding into the chip space as well, because
[31:49] Speaker 12: if the companies that are doing all of the spending
[31:51] Speaker 12: on AI, if they're not seeing a dramatic return on that,
[31:55] Speaker 12: if it's not flowing through to their revenue. The issue
[31:58] Speaker 12: is are they going to start pulling back on spin?
[32:00] Speaker 12: Is the peak earnings being priced into the chip space
[32:03] Speaker 12: right now? And where do they go from there? If
[32:06] Speaker 12: the hyperscervis are pulling back chip stocks, likely we'll see
[32:09] Speaker 12: weaker growth going forward, is the concern.
[32:12] Speaker 2: Bloomberg's Ryan Fastelica on it. Thank you very much. Indeed,
[32:16] Speaker 2: coming up, let's talk a little bit more about private markets.
[32:18] Speaker 2: One of the oldest venture firms in the world just
[32:21] Speaker 2: raised one point five billion dollars for its eighteenth fund.
[32:25] Speaker 2: There's also a lot of relevance here to what's happening
[32:28] Speaker 2: around the world. Some modern EDI general partner at Greylock
[32:31] Speaker 2: here on Bloomberg Tech, stay with us. This is Bloomberg Tech.
[32:45] Speaker 2: Greylock Partners just raised one point five billion dollars for
[32:48] Speaker 2: its eighteenth fund, dedicated to partnering with founders at the
[32:51] Speaker 2: earliest stages of AI across sectors like infrastructure, cybersecurity, and fintech.
[32:56] Speaker 2: Joining us now is Greylock General Partner. So modermedian. It's
[33:00] Speaker 2: good and time need to have a deep private markets
[33:03] Speaker 2: and venure caps with conversation, It's so interesting because I
[33:08] Speaker 2: think in aggregate we see a lot of the most
[33:11] Speaker 2: established firms raising funds with some regularity. This is the
[33:16] Speaker 2: eighteenth fund at some scale, So it's an early stage
[33:19] Speaker 2: fund one point five billion dollars. What do we infer
[33:23] Speaker 2: from that?
[33:23] Speaker 15: Well, thanks for having us on the show. And as
[33:26] Speaker 15: you mentioned, we just launched our eighteenth fund. It's a
[33:28] Speaker 15: billion and a half dollars dedicated to early stage AI entrepreneurs.
[33:31] Speaker 15: Greylock over the last six decades has been partnering with
[33:34] Speaker 15: entrepreneurs when companies get started, companies like Airbnb, Facebook, palad
[33:38] Speaker 15: Alto Networks. And we're excited with this new fund to
[33:41] Speaker 15: back a new generation of AI entrepreneurs. And as you know,
[33:43] Speaker 15: we've partnered with many of the early AI leaders dating
[33:47] Speaker 15: back many years before AI was obvious.
[33:48] Speaker 2: So we rewind the clock to twenty nineteen.
[33:50] Speaker 15: We backed Base ten based ten today as the leader
[33:53] Speaker 15: in AI in prints with you know, serving some of
[33:55] Speaker 15: the leading application companies like Cursor, a Bridge and others.
[33:59] Speaker 15: We backed AI, one of the leaders in customer service AI.
[34:02] Speaker 2: If you call United or Maria today.
[34:04] Speaker 15: You're talking to a crust AI agent. And of course
[34:07] Speaker 15: we're partners with the companies like Anthropic and Open AI,
[34:09] Speaker 15: powering the foundation model layer that's driving this whole economy.
[34:12] Speaker 15: But at the same time, I think one of the
[34:13] Speaker 15: benefits of our history is we've seen many of these
[34:16] Speaker 15: technology waves, and when we think about the iwave and
[34:19] Speaker 15: you connect it back to let's say, the mobile wave,
[34:21] Speaker 15: we're in the equivalent of two thousand and eight or
[34:22] Speaker 15: two thousand and nine. We're one to two years after
[34:24] Speaker 15: the iPhone moment, and I think we're going to look
[34:27] Speaker 15: forward and many of the defining AI companies of this
[34:29] Speaker 15: generation have yet to get started, and this new fund
[34:32] Speaker 15: is all about finding those entrepreneurs and helping them build
[34:34] Speaker 15: those compies.
[34:34] Speaker 2: And I think it's interesting what those entrepreneurs are doing
[34:37] Speaker 2: and what kind of companies that they're forming. You as
[34:40] Speaker 2: an example, so you're investors in open AI and Anthropic.
[34:43] Speaker 2: You got into Anthropic at Series F in the later rounds,
[34:47] Speaker 2: those are the leading frontier labs. With this new eighteenth fund,
[34:53] Speaker 2: you are not necessarily going after the model layer. What
[34:57] Speaker 2: is it you're focused on.
[34:58] Speaker 15: We think of it as three layers, and we will
[35:00] Speaker 15: continue to invest in all three. So there's the model layer.
[35:02] Speaker 15: As you mentioned, we're investors in both Anthropic and OpenAI,
[35:05] Speaker 15: we're seeing new companies get built around new data domains
[35:08] Speaker 15: or new approaches, and we continue to look for and
[35:11] Speaker 15: we'll back new companies at that layer. There's a very
[35:13] Speaker 15: big opportunity and infrastructure, the entire stock's going to get
[35:17] Speaker 15: rewritten around agents. Today we have companies like base ten
[35:20] Speaker 15: and Inference, brain Trust and Observability, Snorkel and Data but
[35:24] Speaker 15: we believe we're going to see a whole new agent
[35:26] Speaker 15: cloud get created. And just like we saw happen with
[35:29] Speaker 15: the rise of the cloud and the hyper scalers and
[35:31] Speaker 15: companies like data Dog and Mango, deb and Snowflake, we're
[35:34] Speaker 15: going to see similar purpose build services for these new
[35:37] Speaker 15: workloads and that's an amazing opportunity for entrepreneurs. And then,
[35:40] Speaker 15: of course, on top of that are applications, and within applications,
[35:43] Speaker 15: we already have companies like Cresstown Customer Service or Abnormal
[35:46] Speaker 15: and Cybersecurity, but we're going to see a proliferation of
[35:49] Speaker 15: application companies, especially now that agents actually work. We're six
[35:53] Speaker 15: months into the models that can power long running agents
[35:57] Speaker 15: and we're seeing amazing results. For example, you know, we're
[35:59] Speaker 15: investor in a company called RESOLVEI. They built agentic on
[36:02] Speaker 15: call engineers in the last six months at companies like Coinbase,
[36:06] Speaker 15: Door dash Fireworks, engineers don't have to wake up in
[36:09] Speaker 15: the middle of the night when there's an incident because
[36:11] Speaker 15: the resolve agent can autonomously handle that incident. That's an
[36:14] Speaker 15: example of agentic AI having huge impact. And we're just
[36:17] Speaker 15: getting started in that era of technology.
[36:20] Speaker 2: We need to talk about Moonshot and Kimmy K three.
[36:24] Speaker 2: You know, the public markets is where you see the drama.
[36:28] Speaker 2: But what is your interpretation of that and why a
[36:33] Speaker 2: twenty billion dollar startup from China topping a benchmark and
[36:38] Speaker 2: people looking at their economics has caused this reaction.
[36:42] Speaker 15: Well, I'd start by saying it was an amazing release yesterday,
[36:44] Speaker 15: largest open weights model to ever be released, two point
[36:47] Speaker 15: eight trillion parameters. As you said, really strong initial results,
[36:50] Speaker 15: and they also showed a lot of interesting techniques from
[36:52] Speaker 15: an efficiency perspective and how they built that model. That's
[36:55] Speaker 15: really impressive and I think it's great for that. Models
[36:58] Speaker 15: like that, models like the one that came out from
[37:00] Speaker 15: Thinking Machines the day prior, continue to give vibrancy to
[37:03] Speaker 15: the open source ecosystem, which is an important part of
[37:06] Speaker 15: our overall AI economy. At the same time, I think
[37:09] Speaker 15: the reaction over the last twenty four hours is perhaps
[37:11] Speaker 15: a little bit premature, and I would say, you know,
[37:13] Speaker 15: a couple of points. It reminds me a little bit
[37:15] Speaker 15: of when the deep Seek moment happened last year, and
[37:18] Speaker 15: when you look at this new Kimmi model, there's maybe
[37:19] Speaker 15: a few things, you know, the audience should consider. The
[37:21] Speaker 15: first is benchmarks are imperfect, right. You know, you can
[37:25] Speaker 15: take a model and make it very very strong at
[37:27] Speaker 15: a particular set of benchmarks. I don't think it's until
[37:29] Speaker 15: that model has had time to percolate in the real
[37:31] Speaker 15: world that we can get a true sense for the
[37:33] Speaker 15: trade offs that the models incurred and what its real
[37:35] Speaker 15: world performance looks like. The second is this whole discussion
[37:38] Speaker 15: around cost. I think the discussion misses the point. It's
[37:42] Speaker 15: very focused on token cost. Yes, but we think about
[37:45] Speaker 15: things more in terms of task cost. Not every token
[37:49] Speaker 15: is equal, and so I'd make two points as it
[37:50] Speaker 15: relates to the cost profile of Kimmy. The first that's
[37:53] Speaker 15: actually interesting is Kimmy K three is much more expensive
[37:56] Speaker 15: on a per token basis than KIMMYK two, which is
[37:59] Speaker 15: sort of contra to the narrative we.
[38:00] Speaker 2: Do not know what it costs to train can be
[38:03] Speaker 2: K three. We have an idea on K two. I
[38:05] Speaker 2: just want to put that out there. Absolutely absolutely correct.
[38:08] Speaker 15: So the per token cost is more, but even more importantly,
[38:10] Speaker 15: it's not particularly token efficient. And so what that means
[38:13] Speaker 15: is for a given task, it actually uses many more
[38:16] Speaker 15: tokens than an open AI or anthropic model. And you're
[38:19] Speaker 15: seeing that already in the early cost beunchmarking. And so
[38:22] Speaker 15: I think this conclusion that it's going to lead to
[38:23] Speaker 15: price erosion for the frontier is perhaps a bit of
[38:26] Speaker 15: pretty in this case, the frontier, let's say it's anthropic
[38:29] Speaker 15: and open AI.
[38:30] Speaker 2: Right. You know some people are making the argument, well,
[38:32] Speaker 2: hold on a minute, if a twenty billion dollar valuation
[38:35] Speaker 2: Chinese startup, we don't know the training costs, but they're
[38:38] Speaker 2: basically saying three million, three dollars per million tokens on
[38:42] Speaker 2: the input side fifteen dollars per million tokens on the
[38:45] Speaker 2: output side. If they can do that, why are we
[38:49] Speaker 2: valuing anthropic at nearly a trillion dollars? Like, what's the
[38:51] Speaker 2: moat that anthropic has?
[38:54] Speaker 15: I think both in thropic opening, I have multiple modes.
[38:57] Speaker 2: The firs is.
[38:57] Speaker 15: They have very significant revenues. These are the fact to
[39:00] Speaker 15: growing companies in human history, and those revenues are not
[39:03] Speaker 15: just on the back of their models and their amazing
[39:04] Speaker 15: API businesses, but their first party products. Right and I
[39:08] Speaker 15: think relative the last time I was on the show.
[39:10] Speaker 15: You look at the success that Anthropic has had with
[39:12] Speaker 15: cloud Code, Cloud Cowork most recently, claud tag It and
[39:16] Speaker 15: open Ai are full stack AI company.
[39:19] Speaker 2: Can I just say one thing? I think for two
[39:20] Speaker 2: years and it's been too long since you've been on
[39:22] Speaker 2: the show, but we've basically assumed the best model wins.
[39:26] Speaker 2: Is it as simple as that you seem to be
[39:28] Speaker 2: saying it's.
[39:28] Speaker 15: Not, Well, it depends on how you define the best model, right,
[39:32] Speaker 15: and I think I would be again, I'd be careful
[39:34] Speaker 15: to jump to the conclusion that Kimmy is now competitive
[39:36] Speaker 15: with the best, certainly on some of the benchmarks that
[39:39] Speaker 15: they release, it's competitive with i'd say one generation prior.
[39:43] Speaker 15: And of course we don't know what's coming out from
[39:46] Speaker 15: Anthropic and OPENINGI in the coming weeks and coming months,
[39:48] Speaker 15: but it's been rumored and reported that they're significant new
[39:51] Speaker 15: releases coming out from those models, So we have a
[39:53] Speaker 15: particular checkpoint in time that we're comparing it to. I
[39:56] Speaker 15: think we're going to see a lot of really exciting
[39:58] Speaker 15: releases in the coming months, but maybe, and if I can,
[40:00] Speaker 15: I want to make one broader point, which is if
[40:03] Speaker 15: you just think about the overall size of the token
[40:05] Speaker 15: economy and where we're going. Right in twenty twenty three,
[40:08] Speaker 15: Opening Eyes API was processing per day about thirty million tokens.
[40:12] Speaker 15: In March of twenty twenty six, they announced they were
[40:14] Speaker 15: doing fifteen billion, so just in less than three years,
[40:18] Speaker 15: a growth of fifty times. I think when we are
[40:21] Speaker 15: talking in twenty thirty, the overall token economy will be
[40:24] Speaker 15: two orders of magnitude larger than it is today, and
[40:26] Speaker 15: so there's going to be plenty of opportunity for the
[40:29] Speaker 15: best leading closed frontier models to grow, for the open
[40:32] Speaker 15: source economy to grow, and for the application layer to grow.
[40:35] Speaker 15: I think the only mistake one can make is underestimating
[40:38] Speaker 15: the size of this overall wave.
[40:40] Speaker 2: Gray Luck, General partner, some monomenty back on Bloomberg Tech.
[40:43] Speaker 2: I hope to have you back with some regularity. Congratulations
[40:46] Speaker 2: on the fund and then your success at the firm
[40:49] Speaker 2: coming up forgets Superheroes director chrispher Nolan is becoming a
[40:54] Speaker 2: franchise all on his own. We're going to take a
[40:56] Speaker 2: very quick look at the blockbuster business behind the opposite
[40:59] Speaker 2: odd that's next. This is Bloomberg Tech. Let's go back
[41:10] Speaker 2: to Netflix. Bloomberg Intelligence is out with the react saying
[41:13] Speaker 2: the results quote reinforce the bearish narrative with a muted
[41:16] Speaker 2: third quarter in twenty twenty six, guide singingly a slowdown
[41:20] Speaker 2: in revenue growth while plateauing engagement keeps sentiments subdued. Keita
[41:24] Speaker 2: ranging up and of Bloomberg Intelligence, the author of that research.
[41:27] Speaker 2: You know, you and I in real time went over
[41:29] Speaker 2: Netflix yesterday. But the main point you want to hammer.
[41:32] Speaker 16: Home, Yeah, the main point is they somehow have to
[41:36] Speaker 16: turn the narrative around ed on this whole deceleration in
[41:40] Speaker 16: revenue growth. I mean, we've seen it significantly come down,
[41:43] Speaker 16: you know, from sixteen percent in the first quarter to
[41:46] Speaker 16: fourteen percent in the second quarter and now eleven point
[41:48] Speaker 16: seven percent guidance for the third quarter. And really the
[41:52] Speaker 16: whole of you know, narrative here is anchored by those
[41:55] Speaker 16: double digit, solid double digit mid teens gains in revenue
[41:59] Speaker 16: growth until and unless we see that picking up along
[42:03] Speaker 16: with a pickup in engagement and subscriber growth. I think
[42:06] Speaker 16: it's going to be really hard to get behind this,
[42:09] Speaker 16: you know, the whole Netflix thesis.
[42:12] Speaker 2: Straight off to the show, Bloomberg Tech producer Justin Lowes
[42:14] Speaker 2: running to an Imax somewhere in New York City to
[42:17] Speaker 2: watch The Odyssey. Is this going to be the big
[42:20] Speaker 2: one of the year.
[42:21] Speaker 16: It absolutely is. I mean we've you know, believe it
[42:24] Speaker 16: or not, the tickets went on sale for this ed
[42:28] Speaker 16: over a year ago and they sold out within minutes.
[42:31] Speaker 16: For you know, all the Imax and the premium formats.
[42:33] Speaker 16: This one is going to be huge. We just got
[42:35] Speaker 16: the numbers for the Thursday preview releases highest ever this year,
[42:40] Speaker 16: beating even Toy Story five. So I think this, you know,
[42:43] Speaker 16: combined with all of the buzz, the fact that it
[42:45] Speaker 16: was shot in an Imax camera, and that you know,
[42:47] Speaker 16: everybody wants to I mean, Christopher Nolan is always great
[42:50] Speaker 16: at eventizing a film, and I think we're going to
[42:53] Speaker 16: see that really translate into some very very strong numbers
[42:55] Speaker 16: in the box office.
[42:58] Speaker 2: I really want to see this one too. Media and
[43:00] Speaker 2: Entertainment ANALYSISKI that Arang an athen of Bloomberg Intelligence. Thank
[43:03] Speaker 2: you very much. That does it for this edition of
[43:06] Speaker 2: Bloomberg Tech. Again, the picture in financial markets, in equity
[43:11] Speaker 2: markets is there's a lot of red. We're off session lows,
[43:14] Speaker 2: significantly off session lows. But the catalyst was China's moonshot
[43:19] Speaker 2: releasing Kimmy K three, and that causing concerns that, well,
[43:23] Speaker 2: why are we spending all this money in America on
[43:25] Speaker 2: developing AI if a startup in China can do it
[43:29] Speaker 2: at a much cheaper level with different economics. There is
[43:32] Speaker 2: some dispute of that. We went over in the show.
[43:34] Speaker 2: In video and Intel down Apple lower. At one point,
[43:37] Speaker 2: Apple was the world's most valuable company, pipping in video.
[43:40] Speaker 2: Right now, it's not We're going up for more tech
[43:42] Speaker 2: earnings next week. This is your calendar, this is what
[43:45] Speaker 2: we're bracing for. It really starts in earnest over the
[43:48] Speaker 2: next seven days. Don't forget to recap the show on
[43:51] Speaker 2: the podcast. You know where to find it on the
[43:53] Speaker 2: Bloomberg platforms, but also online on Apple, Spotify, and iHeart.
[43:57] Speaker 2: Have a great weekend from San Francisco. This isl and
[44:00] Speaker 2: bag Tack

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.
