Loading summary
Transcript103 lines
- [00:03]
A
Hi, I'm Lisa Mateo introducing you to the new Stock Movers Report from Bloomberg. These are short audio reports, five minutes or less, delivered right to your podcast feed. Throughout the day, Stock Movers fills you in on the day's winners and losers on Wall street and tells you about the news and data that's driving those gains and losses. If you want to stay plugged into the stock market but don't want to spend all day watching tickers scroll across your screen, then Stock Movers is a place for to get informed. Listen a couple times throughout the day to find out what's moving equities and why. Search for Stock Movers on Apple podcasts, Spotify or anywhere else you listen. Get the latest stock news and data backed by reporting from Bloomberg's 3,000 journalists and analysts across the globe. Subscribe to Stock Movers wherever you get your podcasts from the heart of where innovation, money and power collide in Silicon Valley and beyond. This is Bloomberg Technology with Caroline Hyde and Ed Ludlow. Live from New York, I'm Caroline Hyde.
- [01:30]
B
And I'm Tim Stanback in San Francisco. This is Bloomberg Technology.
- [01:34]
A
Coming up, Alphabet is set to acquire cloud security firm whiz for $32 billion in cash. The acquisition is Alphabet's largest to date. Plus all eyes on Nvidia. CEO Jensen Huang is set to take the stage in a few hours as investors seek clarity on Blackwell Ultra and the next AI chip breakthrough. And Chinese carmakers are gaining market share globally. This is BYD soars on the news of its 5 minute charging system. Alphabet off by more than 3% as the market falls, but also after a huge acquisition is announced, its biggest ever and it's going to be leaning into cloud security with the help of Wiz. This is just a five year old company, Tim, and it's backed by VCs. It's going to be a bit of a payday if they can get it secured by 2026, but phenomenal ride for these group of four founders who built once before and sold out to Microsoft back in 2015. This time it's Google's chance.
- [02:24]
B
Yeah, incredible track record and incredible what they've done in five years. For more on Google's Wiz acquisition, Bloomberg's Katie Roof joins us now. Katie, you've been following this for the last 10 months. Whiz spurned the offer of $23 billion over the summer. Apart from the additional $9 billion, what changed?
- [02:41]
C
That's right. Well, yes, obviously they had 9 billion more reasons to say yes, but also one thing that's changed is the reg regulatory environment and at least the perception that it will be better under this regulatory environment. Yes, we've been following this for 10 months. We started to hear rumors even two months before things leaked last year. The talks first began at rsa, this big security convention last year. So it's been going on for a while.
- [03:08]
A
What's so interesting? Yes, hours have increased significantly after the last eight months, we understand and that vindicates the price point. But for a $32 billion price tag and a company that's going to remain independent, there's actually still going to be multi cloud provision. It's not just going to be selling into Alphabet customers when it comes to cloud. It's still going to be servicing us. And as you. It's a pretty phenomenal deal.
- [03:33]
C
Absolutely. It's definitely a phenomenal deal for whiz. But you know, it just shows how threatened Google feels right now about security and how worried they are about competing with Microsoft and Amazon grow AI related security threats and you know, they're willing to pay the price. I mean granted you know Google can easily afford this, it's a high number for whiz. But you know these trillion dollar companies can, can make these multi billion dollar deals happen quite easily.
- [04:03]
B
Does it move the needle with competition in the cloud space? As we just talked about, you have us number one, you have Microsoft Azure number two. Google Cloud is in third place. Can it compete now that it has or if the deal goes through?
- [04:17]
C
I mean that will be the question. That remains to be seen. But you know, certainly that's their goal here. They really are hoping that you know, look, Wiz that's only been around just a few years quickly gained a significant percentage of the Fortune 100 customers. And you know they, they want to be able to sell this as part of their, their offerings. You know, certainly, you know, as Caroline noted it, it can still work with the competitors but Google clearly has this, this on their team right now. Whiz will maybe continue to work with them and innovate, you know, under the umbrella of Google.
- [04:52]
A
What's interesting is they could have fundraised in the private market still. They could have eyed an ipo, but they've gone this particular route. Who wins out? Remind us of the VCs that are back this from day one.
- [05:05]
D
Sure.
- [05:05]
C
So you've got Index, you've got Sequoia, you've got Cyber Starts which is a high Israeli security firm. Those are all on the board. You also have Insight Partners on the board as well. But they've also been backed by Green Oaks and Lightspeed and Andresa. They have such a long list of, you know, top tier Silicon Valley investors that were throwing money at this thing. It was practically born a unicorn. It was, it was so highly valued when it first launched several years ago. And so there have been big believers in this company for a while. They felt like, you know, just, you know, hit the ground running.
- [05:41]
A
Katie Ruth, thanks so much on all angles on Whiz and Google. Meanwhile we turn our attention to in video. The Global Artificial Intelligence Conference is upon us. GTC expected to give developers and investors insight into the company's future chip plans as well as of course we're awaiting CEO Jensen Huang's keynote speech. Let's bring in Bloomberg Intelligence is Mandeep Singh. Look, earnings. Jensen couldn't quite get the investor base to get excited about the new scaling law, which is reasoning models. What is he going to be able to convince us this time?
- [06:12]
E
Well, I assume the focus will be on agents and inferencing given the market has clearly made that pivot post deep seek. Everyone is focused more on, you know, reasoning and inference time scaling. And even though we are talking about scaling laws, I mean it's fair to say everyone isn't very excited about pre training anymore where you are spending, you know, $10 billion upfront in training, just the next version of your model. So, so I would assume Jensen will try and convey that, you know, reasoning models do require a lot of GPUs and Nvidia GPUs are still the most popular when it comes to inference time reasoning. But clearly, you know, Google spending $32 billion on an acquisition takes away some of the capex they could have spent acquiring Nvidia GPUs. So you have to remember four customers contribute 45% of Nvidia's revenue. Now if they are doing more deals and deals of this magnitude, that will play into the capex expectations for next year.
- [07:14]
B
So Mandeep, is that your view, that reasoning, that inference still does use as much compute as training these labs so.
- [07:23]
E
It changes the nature of compute. Remember you know when you were training models with bigger clusters as Grok3 did you know with 200,000 GPUs you required those all as a cluster. Now if you're doing more inference time reasoning, you don't require one data center with, you know, giant cluster, you can have it distributed for latency purposes. And that's where you know if you are a hyperscaler you are looking at different Options, different custom ASICs because these AI agents are performing different tasks. It's not all text based chatbots. It could be, you know, audio based image generation. So custom ASIC solutions could work. And Broadcom did talk about, you know, adding four new customers when it comes to their Asics. So clearly there are a lot of moving parts when it comes to inferencing. But that's what I think Jensen would focus on during his keynote is interesting.
- [08:16]
A
Actually reports today of how much Amazon is thinking about the price point of its homegrown chips at the moment and where it can serve an inference and inferentia. But when you've got that sort of level of competition, they have to keep saying we've got the next thing, got the next thing. So talk about the next thing. We've got Blackwell Ultra and then the next iteration of the chip.
- [08:34]
E
I mean, look, Nvidia's value proposition is they can do so much more with one chip in terms of power efficiency and the compute they give versus the alternative. So Amazon may have a chip, but it's not the same price, power, performance that Nvidia is giving you with one chip. And that's their value proposition that you have the CUDA layer when it comes to, you know, the cluster that you are running on for Nvidia. So all the standardization does make a difference. But hyperscalers worry about, you know, costs, not their what they are paying this year, but five years down line, 10 years down line. So they're thinking long term and you want to move away from Nvidia given the size of, you know, the investments they are making with the compute right now.
- [09:15]
B
That was Bloomberg Intelligence. Is Mandeep Singh joining us in New York. Mandeep, thanks so much. Well, let's talk more about what investors will be looking for from GTC and Jensen Huang's keynote later today. We're joined by Daniel Pilling, portfolio manager and senior research analyst at Sands Capital. Daniel, good to see you. What do you want to hear from Jensen today that will convince you, that will convince other investors that earnings growth and indeed growth of the company has not peaked.
- [09:41]
F
Good morning. Yes. So I think first of all I agree with the reasoning, thought process and the inference. So inference is becoming much more computer intensive. It can be up to 102,000 times more computer intensive. Nvidia is the biggest player in that by far. And you know, the hope would be that Jensen will share some of his long term strategy on sort of how does he capture and continue to gain market share with an inference. Now the second thing though, I would say is I actually think that the training, the scaling laws within training are not entirely dead yet either, both in pre training and post training. And my guess would be that the company in video will confirm that as well. Now, this year we're going to see the big Blackwell cluster scaling. That's going to be a 10x improvement versus last year for sure. And then secondly also we have this reinforcement learning without humans in the loop that Deep Seq has shown. That seems to describe another scaling law in post training. And as far as we can see, that has driven a lot of a ton of training demand and which you can see actually frankly on GitHub and hugging face in terms of the number of Deep SEQ models and then the three. And the third thing, if I may say, there's going to be a lot of talk about physically that's going to its S curve as we speak. You can see that in Waymo, right? In San FRANCISCO they have 20% market share. And I think there's going to be a lot of talk about agents, but frankly that's a little bit further out. So it's nice if they talk about that, but I think it's more important to talk about inference, scaling and trading scaling.
- [11:12]
A
Really interesting points. He's often paid lip service to Elon Musk and Tesla, for example. Maybe a bit with Mamo, maybe the future of robotics as well. Daniel, I'm interested in the rampant pace at which now we see Jensen put out new products. We're going to hear a lot about Blackwell Ultra, the next iteration. We're also going to hear about Rubin, I'm sure, which is in the next AI breakthrough. Can they really develop at this pace? Because we saw the sort of supply chain hiccups with Backwell as it rolled out.
- [11:40]
F
Yes. So I think the answer is yes, and I'll tell you why. They have the best people. They have the most people focusing this product and they have the biggest R and D pools and they have the most capacity to Taiwan say. I think the interesting counter question to this though is to say, can the others even follow? Right? So, for example, big hyperscalers, they all try to develop their own chips. They're trying to do this with a partner. They're not doing it all alone, which means that they're trying to optimize the system by themselves. Nvidia is optimizing the systems by themselves, whereas the others are doing with partners, which is going to make it much, much, much more difficult to compete with Nvidia. So I think the answer is yes. And I think that creates a lot of competitive differentiation versus anybody else.
- [12:20]
B
Daniel, you mentioned physical AI and I'm curious what you think is the realistic form factor when it comes to physical AI and just how far out we are from that.
- [12:31]
F
Yeah, so, so we thought about this a lot as well. So it's clear that the big, big, big physical AI, the revenue growth driver in the next one to two years is going to be self driving decent in our opinion. Why? Because you can see it in San Francisco already. The experience is amazing. When you drive away more people, they have 20% market share, they're moving to many other cities and this is just the beginning. Right. They have a tiny, tiny market share of the total mileage. Now the human rights, they're pretty interesting as well, but arguably a little bit further away. So in our opinion it's going to be self driving vehicles within big cities in the next one to two years. That should drive a massive explosion in terms of inference demand, which benefits the likes of Nvidia but many, many other players on the supply chain as well.
- [13:15]
A
Daniel, how far away is Quantum? Because we actually heard a sort of a long time rage, maybe up to two decades coming from Jensen, but now he's actually sitting on a panel around Quantum and trying to make it a quantum day.
- [13:28]
F
Yes. So we think quantum computing is sort of maybe in the 80s in terms of where semiconductors were like at that time period. Right. So it's a long time out. The reasoning for that is first of all there's no CUDA equivalent as of today, at least for quantum computing, so you can't actually use it easily. And secondly, the scaling of the qubits has proven to be quite difficult. So it's at least 10, 15, 20 years out in terms of a use case potentially. And I would actually argue if Nvidia plays their cards well, then quantum computing could be one of the parts of computation they offer in the very, very long term alongside the GPU, CPUs and whatever else you might have.
- [14:09]
A
We want to thank you Daniel Pulling always fascinating catching up with the portfolio manager and senior research analyst at Sans Capital. Meanwhile, we have some breaking news for you because of course the phone call has begun. Russian President Vladimir Putin and Donald Trump have been speaking on the phone today. We understand that the White House confirmed Talk started at 10am Eastern Time. Discussions are said to be related, of course, to the end of the war with Ukraine. A key objective being a 30 day truce with Kiev has already, we understand Kiev has accepted that we'll bring you any more Breaking news on that.
- [14:50]
B
Well, let's take a look at some China tech names, starting with the ride hailing app Didi Global. Think of it as China's answer to Uber. It did swing back to a quarterly loss and it's a blow to the company that's exploring a Hong Kong listing this year. You can see shares on the day lower by about 3%. Didi has had an inconsistent recovery since that regulatory crackdown a few years ago. That was when it was delisted from the New York Stock Exchange and lost much of its market value. This after Beijing cracked down on data sharing practices among big tech companies. And show me shares are higher right now. Revenue in the fourth quarter beat the average analyst estimate. The company's CEO also said it's delivered 200,000 vehicles and it's raised its target this year from 300,000 to 350,000 vehicles. Shares higher right now by about 3.3%.
- [15:37]
A
Caroline, isn't that extraordinary that we know that Show Me first got into the EV market only in the summer of last year and now they're pumping out 200,000 and likely to increase it to 350. All the while that we're seeing just the Chinese EV sector go from strength to strength also in innovation. I mean, what did you make of BYD's pop today?
- [15:56]
B
Yeah, I mean it's pretty incredible. Five minute charging. I mean, that's really what led to that pop today. And if you think about it from the perspective of somebody who has an EV and maybe has range anxiety, that is a game changer. If it takes just the same amount of time for you to charge your vehicle as it does to go to a gas station, then boom. That's kind of the golden goose when it comes to this tech. Caroline.
- [16:17]
A
But not everyone can get access to those golden goose or at least those eggs because look, European countries can buy the Hanao or the Tang L. The latest that's going to be coming from BYD with this new electric charging technology, but not in the U.S. let's just talk about that because of course President Trump is hoping his tariffs are actually going to continue to combat cheaper Chinese cars and stop them entering the US Market. Already. You can't access byd, for example, but outside of the US Chinese automakers have really established a global presence. It's today's big take story. He discusses one of the authors of the piece, Chester Dawson. And let's just take BYD free as a case study here. Chester. They have just come from behind to end up being basically the World's biggest EV maker. And they're doing it by accessing the emerging markets, not, of course, the us.
- [17:05]
B
That's right, yeah. I mean, it is somewhat unusual. You know, typically when you have a fast rising new entrant in the automotive business, the first thing they want to do is get to the US because it's one of the richest and biggest markets around the world. But as you mentioned, you know, it's been kind of core to US economic policy and national security policy to start, you know, drawing up that drawbridge and protecting manufacture of cars. You know, obviously the Detroit three, but also companies like Toyota that make a lot of cars in the US are not facing that competition. But as you say, technology doesn't stand still. And they're introducing some pretty impressive new developments. They're going to show up in places like maybe Johannesburg and, you know, Bangkok before they get to the U.S. do they ever get to the U.S. i mean, Chester, people love these vehicles. Ford CEO Jim Farley a few months ago was quoted as really loving the show Me vehicle that he had been driving. That was a sort of big news in the US Auto world. People love these cars. What do they love about them? Well, I think the one thing they love is the price. I mean, they're very competitive, which is not to say they're, I mean, they're cheap in terms of the price, but they're not junk. They come with some pretty impressive features and creature comforts. You know, as you noted at the outset, you know, Shami, which used to make cell phones, kind of like, you know, the Apple on the iPhone. But in China, they quickly moved into cars and, you know, they're now projected to be making seven times what Rivian, a US electronic electric vehicle startup, wants to make. And that's in large part due to the fact that, you know, they're very user friendly. So yeah, it's, it's no surprise. They're good price and pretty good quality as well. Bloomberg's Chester Dawson. Chester, always good to see you. Thanks so much for joining us. Coming up, we're going to look at Google's impact in the health sector, all with AI. This is Bloomberg.
- [19:15]
A
Hello, I'm Caroline Hoepke.
- [19:16]
B
And I'm Stephen Carroll.
- [19:17]
A
Every weekday morning, we bring you the Bloomberg Daybreak Europe podcast. It's 15 minutes to get your day started.
- [19:24]
B
All the top global news stories, how they affect the economy and your finances. We have to understand the changing world when it comes to trade. Wage growth is still really, really high.
- [19:34]
A
The disinfl process is well on track these days a lot can change overnight, so we're here first thing every morning to bring you up to date.
- [19:43]
B
We have 3,000 journalists around the world.
- [19:45]
A
To tell you what happened and why it matters.
- [19:48]
D
US exceptionalism is being questioned heavily.
- [19:51]
A
Some damage is done. Tariffs are bad for growth. Economic stability is non negotiable. At Bloomberg we say that context changes everything. That means the data and expertise to help you understand a world that's changing at breakneck speed.
- [20:05]
B
So join us every weekday morning for.
- [20:07]
A
The Bloomberg Daybreak Europe podcast. Subscribe today to get us in your feed by 7am.
- [20:23]
B
Google is expanding its health related AI summaries in the search to improve its influence in the health sector. Answers will now cover thousands more health topics and expand to more countries and languages. The company is also adding a separate feature in search called what People Suggest, which it said aims to provide users with information from people with similar lived medical experiences.
- [20:45]
A
Tim yeah, Google also said it has been updating a new AI system to help researchers speed up the scientific and biomedical discoveries. For example, Google's a co scientist tool. It's an AI system meant to act basically as a virtual collaborator for biomedical scientists. I actually got to sit down with Annalise Palowski, founder and principal investigator for the Accelerated Science, Biochemistry and Molecular Biology lab over at Google. I asked her to walk us through what this looks like.
- [21:12]
G
Take a listen.
- [21:14]
D
The idea is we built a system that works together with scientists. So for instance, if it's a Friday afternoon, I'm in my lab, I have all these ideas I want to explain floor, but I have my two kids at home and so I can actually plug in a research goal into the system, check the ideas it's creating and the evolution over time. And by Monday morning I have a bunch of different approaches I can use for my research.
- [21:34]
A
Ultimately, there's been much said analysis about how generative AI can bring fruits to bear when it comes to health care, when it comes to research, when it comes to tackling disease. Is this the reality that you're currently thinking can be the help with a co scientist, an AI co scientist, scientists?
- [21:50]
D
We're at the early stages of the project, so in our paper we show these three examples or we've seen that there is evidence in this direction and we're looking forward to leaning in more and showing more capabilities as we extend this work.
- [22:02]
A
Has there been any friction, or should I say just reticence by the scientific community to allow models to duke it out on their areas of focus over a weekend while they sort of down tools.
- [22:15]
D
There can be skepticism, but as a scientist, skepticism is good, right? It encourages that we push ourselves, that we create a system that's useful and something that's impactful for the community.
- [22:24]
A
How have you stress tested it? For many the limitations of generative AI and for agents is, well, the fact that they do sometimes get things wrong. When have you seen that occur and how have you managed to counterbalance it? Right.
- [22:36]
D
So no large language model is perfect, but we try to supplement this with novelty and checkness reviews. We both use the literature that's available and web search. We add in tools and databases and we're learning to extend our work into knowledge graphs and other strategies where we can improve and refine our techniques. The other thing is we're looking to the community for feedback. So we created a trusted tester program and we welcome and open those different ideas and approaches. So how we can improve AI Co.
- [23:02]
A
Scientists Google Research analyst and scientist Annalisa Palsy ski there ahead of the key Google Health event in New York. Coming up, Nvidia's GTC has investors focused on the outlook. Luxe Capital's grace is for this with us and to join to talk all about the startups and where she's finding opportunities for the technology. This is Bloomberg Technology. Welcome back to Bloomberg Technology. I'm Caroline Heiden, New York and I'm.
- [23:34]
B
Tim Stanback in San Francisco.
- [23:36]
A
Quick check on these markets Tim, because risk off at the moment. Many reasons for investors to be anxious got the Fed later today. What do tariffs actually mean for the US economy? What does Middle Eastern instability once again mean in terms of a flight to Havens? Golds at a record but the Nasdaq's off by 1.8%. We're seeing Bitcoin, another key risk asset of Choice, off by 2.5%. We're only at 81,000. Now move on. Some of the individual names that have tried to keep us above water but fail Nvidia is one of the key points drags on the day even as we look ahead to gtc. And the keynote coming of course from Jensen Huang at 1pm Eastern time. Can he re inject optimism into the need for his GPUs for inference as well as training? I'm looking at Alphabet making a big purchase $32 billion for Whiz down three and a half percent as indeed the entire market sinks. I'm looking at matter though off by almost 5% percent. This was the one Magnificent Seven name that actually was in the green for the year. Not anymore. When out the lowest since 11-29-10.
- [24:33]
B
Well, let's see what Catherine Rooney Vera has to say. She's Stonex Group chief market strategist and she joins us now. Catherine, you saw what Caroline was talking about there. With tech under pressure, the NASDAQ 100, only 8 of the 100 stocks in the green right now. What's your outlook for the sector?
- [24:50]
H
Well, it depends on what happens with rates. This is a very rate sensitive sector. So with treasury yields high 3 1/2 to sorry for 430 to 450 that squeezes long duration growth stocks and that's unwinding some of these more crowded trades. I think that's probably one of the biggest drivers of tech's move to the downside. Then of course we have any type of broad market correction. Given that tech is about 30% of the S and P market cap, if it is vulnerable as well to any risk off trade which could be precipitated by trade tensions or, or anything of the like. I think it's also entering a mature phase so that anything related no longer gets this immediate bump up. But now the market is starting to be more discerning and more aware of massive capital spending and the potential for, you know, less capital spending going forward. Especially with the American introduction of lower cost chips and competition within Nvidia though.
- [25:52]
A
For example, let's just use that as the pin up now trading at 25 times future earnings. A lot of froth has come out of the valuations. At what point do we start to get dip buying, do you think, Catherine?
- [26:03]
H
I suspect we soon will. 10% correction is, is something but it certainly doesn't unwind the frothiness that we've seen in the run up for a second stocks in general over the past several years. And I'll add one point which I think is that what you're seeing is a resetting of value expectations and a move out of tech and into second derivative AI plays such as industrials and even energy. You're seeing financials catching a bit. My top pick for this year is health care and the reason for that is both because it's defensive, so it's a nice hedge of your bets, but also because it was the worst performing sector last year year. And if effectively we do get a purchase of the dip and a resumption to the upside of the S and P S and P by year end, then my suspicion is that the most beaten up sector from last year will become one of the better performing sectors as those value investors come in.
- [26:58]
A
I mean You've been right thus far this year. Health care has been the winning formula and financials techs languishing near the bottom. I use another case study in Tesla because last year but since the election results, Tesla didn't trade on fundamentals, it traded on vibes and an association with the White House. When we go back to vibes again, if ever, I mean, does it, is it case in point, once yields stop, we do start to see the upside and risk hunger just returns like that or are we in a totally different paradigm right now?
- [27:28]
H
I think you make a really good point. And I think the vibe shift now is less focused on inflation and the Fed and it's more focused on, on tariffs. So we have to ask ourselves the biggest component of the S and P, how vulnerable is it to tariffs? And are is this tariff fear a real threat to economic growth? And what we did was we looked back at President Trump 1.0 and even the Federal Reserve came out with a recent analysis quantifying Caroline, the impact of Core PCE from Donald Trump 1.0, 27, 2018, 2019 tariffs and they quantified it at 0.5 percentage point points addition to Core PCE. Now the impact on final goods back then was a one year effect before it rolled off. And then we saw intermediate goods have a longer lasting impact to the upside on prices. This time is slightly different because supply chains have already adjusted. So that means that new tariffs would hit margins and earnings rather than spark inflation. So my concern is that, you know, additional tariffs will directly impact market sentiment and that will bring additional downside to tech. So my call has paid off pretty well which is say long tech buy puts on tech. Now that trade is looking really awesome. I put it in when tech was at all, all time highs but now what we see is that put volumes are at all time record highs. So now it's expensive to do. But it has been, it has proven to be a very good trade.
- [29:01]
B
Catherine, up to now we've seen the main beneficiaries of the AI boom be the chip makers and the hyperscalers, right? To what extent are we seeing other companies in other sectors within the s and P500, for example, adopt this technology and has it hit their margins yet?
- [29:17]
H
Yeah, you're right, you know, and this has been a kind of a buzzword more than wholesale adoption. And the market is, is looking for results at this point. So enough talking and we want to see deployment, not only deployment, but full execution and earnings. So there's been massive investments in this space. I think at this point, as I said before, I think we're in a mature phase. We want to look for second order effects. So second order investment opportunities that are indirect beneficiaries are perhaps the second wave beneficiaries of this large scale massive capex in AI. Last year I recommended utilities. Utilities was one of the top performing sectors last year and my recommendation was because it's, you know, a second derivative to now I think we need to be looking at data center operators, cloud infrastructure, medical, real estate financials. I think there are second order effects. That's where the bang is for the coming for coming years in a structural way.
- [30:18]
B
Which sector do you think will take advantage of the the most?
- [30:23]
H
Well, I, I still like health care. I think there's going to be a lot of deployment of AI in this in the sector and I and I like it one, for that reason and two also because it's a nice defensive sector. So that's going to be my top pick for this year just as Utilities was last year. So they're nice because they're both defensive and they could potentially benefit from additional upside.
- [30:43]
A
Catherine Rooney Vera of Stonex Group Always great to catch up. Thank you Tim.
- [30:49]
B
Well, it's time now for talking tech starting in space. Two NASA astronauts are returning home in a SpaceX craft after being stuck in orbit for nine months. The astronauts undocked from the International Space Station last night and are expected to splash down off the Florida coast later today where they will undergo medical checks and be reunited with their families. Plus, Apple lost its right at Germany's top civil court to overturn a regulator's decision to put it under tighter antitrust scrutiny alongside other US tech giants. The court cited Apple's vertically integrated products and services, its importance for third party access to markets and its potential to harm competition as reasons for the decision. And Cognition AI, the developer of an AI powered coding assistant, has raised hundreds of millions of dollars at a valuation close to to $4 billion. That's according to sources. The round doubles the previous valuation of the startup which released its Generative AI coding tool a year ago.
- [31:46]
A
Caroline let's take more on Generative AI, the AI sector writ large and VC Spotlight. Grace is with US partner at Luxe Capital focusing on emerging opportunities in artificial intelligence. You're based here in New York and you're on the board focusing in and helping with the investments in certain pin ups. I think think of Runway and I think of together AI for example, how are we seeing the application layer AI starting to return real value? Because at the Moment everyone have been so focused on the foundational and the infrastructure layer. Are we starting to reap rewards?
- [32:16]
D
For the last few years, we've really seen enterprises focusing on foundational AI. Now we're seeing that shift, as you say, to the applications, applications of AI and doubling down into the 20, 25 year year of the agent.
- [32:29]
A
Agents.
- [32:29]
D
Actually, we've been talking a lot about agents in enterprise production. We just talked about where we actually see that impact. We're seeing it across our portfolio in really three main areas. One, encoding, two in customer support and three, and sales. And I'll give you one quick example on that. On the real ROI and impact we're seeing. There's a company called Maven AGI, they're an agent for customer support. They are seeing real enterprise impact integrating with those legacy, legacy systems and working with public companies to actually solve these tickets in a fraction of the time at a fraction of the cost, driving that real value.
- [33:04]
A
What's interesting is we just had Daniel Pilling at Sans Capital on who's an investor in some of Nvidia. His theory is actually sort of physical AI, which I know you've also invested in, but physical AI is actually in many ways closer than agent Ki in terms of the real rewards. What do you make of, of that sort of time frame and when we start to see more bountiful effects of agent?
- [33:24]
D
Ok, well, it depends how you define agents and that's actually part of the confusion. Right. Everyone has their own definition. Right. For, for my purposes, for engineers I talk to across our portfolio, it's really an autonomous system or an LM is directing its own actions across a series of steps and these agents are getting better and better, or not really seeing that truly verifiable loop. In fields that are not kind of coding, customer support, sales, it's because in fact they're actually difficult to integrate with. Right. And that's kind of a funny kind of the irony in the space right now. If you look at the physical world, we've seen text and image highly repeatable, highly verifiable tasks, whether it's in the sciences or whether it's in math. But as you start to get to these proprietary data troves, I'm talking robotics, I'm talking warehouses, I'm talking data and manufacturing. We have to tap into this proprietary workflow of data, even for a biologist or chemist, and you have to not just get that data, but also integrate it with the workflow. That's where we're seeing really exciting advancements. But it's still early from seeing that real enterprise roi.
- [34:24]
B
Grace, it does seem like the opportunities that you just spoke about are really all enterprise focused. Really, you know, thinking about it B2B. What does it look like for consumers if this technology is adopted on a wide scale?
- [34:36]
D
Well, I'll give a simple example on that. You know, I tried to book a flight from New York to San Francisco using OpenAI's operator a few weeks ago. Simple task in theory, assure a flight, lots of us do. In reality it's pretty complex. You know, you need to get the right airline, you need to get the right airport, you need to get the right time and all my preferences. And so that's a common issue we're seeing. These AI agents are complex systems with compounding errors. So even if there's one little error like not getting access to that one criteria, it compounds across that 50 steps in the Agentix system. So a consumer experience perspective, it can be pretty frustrating when you don't book that flight exactly the way you want it.
- [35:15]
B
Are you seeing crowding out come from big companies? Right now you have the hyperscalers valued at multitrillion dollars. Is there an opportunity for small companies to come in and actually provide solutions and complement these hyperscalers, complement these large chip companies?
- [35:34]
D
Most definitely. In fact, we're seeing it across our own portfolio. I'll give another example. They are called together AI. They're an open source AI cloud. They actually help you run and productize these fantastic open source models and you can use them, you know, not using a cloud provider in tandem or working in tandem with your own existing suite of tools. And so you're seeing the emergence of this new infrastructure stack which is still neat adjacent but exciting. We're using these cheaper chips, you're actually using these open source and even smaller models to achieve that same outcome.
- [36:03]
A
We're looking at your portfolio companies and some of them have international flavors and certainly hugging face. I think a climb over there running that business, you're in second as well. Tokyo based at this moment where we're almost putting up this race of China versus the U.S. what other talent pools are you looking to for AI experts, expertise and excellence?
- [36:22]
D
Well, expertise is everywhere and I think if you focus on China for a second, right. China has fantastic engineers and researchers and I think deep seat was a lot of a wake up call for the US who were saying wow, you know, whatever you thought deep sea cost to train, it was a true feat of hardware efficiency and engineering efficiency. And I actually think that's a blueprint. We're going to see use internationally. So not just in Japan with companies like Sakana AI using those smaller agent based models. Also in France with companies like Mistral building really, really strong open source models as well. But how do you actually leverage and do more with less? How do you use less compute? How do you actually use more efficient chips, in fact cheaper chips and actually have a more efficient infrastructure stack using that open source technology?
- [37:04]
B
Grace, here in the US the change in administration not even two months ago, I'm curious about how you view the landscape differently now that somebody likes David Sacks for example, whom Caroline and Jackie spoke to just a couple of weeks ago, is in the position as AIs are. Do you see a difference in the landscape here?
- [37:21]
D
I think it's early to know but we're really seeing a huge excitement for the industry and really the industry being invigorated. Right. We just talked about Deep Seq as that being a wake up call to invest more heavily in engineering startups, in awesome AI startups who are building right here on US soil. And we're actually leveraging this open source and transformation transparent innovation to ultimately create really exciting applications. I'm really excited for new applications in some of those physical areas we talked about earlier. So robotics, space, defense and manufacturing all have those proprietary data sources and that unique workflow. You need to understand the workflow of a defense contractor if you're actually building a fantastic application. So that mix of kind of public and private is something that's really exciting me.
- [38:06]
B
Lux Capital's Grace is for the grace. Thanks so much for joining us. Do appreciate it. Well, coming up, a lawsuit alleges that Google and the startup character AI are to blame for the death of a 14 year old boy. We'll discuss the claims and the implications for Silicon Valley. This is Bloomberg.
- [38:28]
H
You can get the news whenever you.
- [38:30]
A
Want it with Bloomberg News Now. I'm Amy Morris.
- [38:33]
D
And I'm Karen Moscow, here to tell you about our new On Demand news report delivered right to your podcast feed. Bloomberg News now is a short 5 minute audio report on the day's top stories. Episodes are published throughout the day with the latest information and data to keep you informed. Yes, there are other products like this.
- [38:52]
A
From a variety of news organizations, but.
- [38:55]
H
They usually rerun their radio newscasts throughout the day.
- [38:59]
A
That's not what we do. We create customized episodes that can only be heard on Bloomberg News now and.
- [39:05]
D
We don't wait an hour to publish breaking news. When news breaks, we'll have an episode up in your podcast feed within minutes. So you're always getting the latest stories and developments.
- [39:14]
H
Get the reporting and the context from.
- [39:16]
A
Bloomberg's 3000 journalists and analysts.
- [39:19]
H
We're all over the world.
- [39:21]
A
Listen to the latest from Bloomberg News now on Apple, Spotify or anywhere you listen.
- [39:36]
B
In February of last year, 14 year old Sewell Setzer took his own life. His mother blames the chat bot he spent months messaging with. Now she is suing Character AI, it's the company behind the bot and Google, which has a partnership with the startup. Let's bring in Bloomberg's Malafi Nayak. She joins us here in San Francisco. Malfi, this is just a heart wrenching story. We're going to get to the legal implications in just a minute. But, but what happened here?
- [40:01]
G
So this is a story about a young boy who began using this chatbot technology. And he got romantically involved with a chatbot that he created himself, sort of inspired by a character from Game of Thrones. And he got really sort of caught up in this chatbot. You know, his mother says that he began to sort of withdraw from his friends, was having trouble at school, you know, after a few months. And at some point they, the parents decided to confiscate his phone. And while looking for the phone, he happened to find his stepfather's gun, which was hidden in their home in compliance with Florida law where the family lived. And he sort of, you know, got his phone, had this last conversation where he was talking about sort of coming home to this chatbot and then he, you know, he just sort of took the gun to his head and shot himself. So it's a tragic story. And there is a lawsuit now which the mother has filed against Character AI, which created this chat bot technology, as well as Google, which has an interesting sort of deal with, with this startup. And it sort of brings into question these sort of new deal structures in AI, where companies, technology companies are not really buying these startups outright, but they are instead of buying the assets, they're sort of investing in these technologies in a very sort of unusual way where they're hiring the talent and they're just licensing the technology. So there's no real full blown acquisition here. And it's possible that, yeah, it's possible that these deals were sort of structured this way because last year with the Biden administration, we saw a crackdown in mergers and acquisitions where, you know, big companies were not allowed to scoop up their smaller rivals. So it's possible that these deals were structured that way. But it's sort of interesting that the lawsuit sort of brings these deals into the spotlight and sort of questions the legality of the sort of relationship between these big tech companies and these smaller startups.
- [42:02]
A
Right. Maltese Garcia's lawyers say in the complaint that Google in particular contributed financial resources, personnel, intellectual property, property and AI technology to the design and development of character AI's chat bots. We must be clear that both Google and of course Character AI dispute any responsibility in terms of they have come back and certainly stated that Google and Character AI are completely separate and unrelated companies. So says Google, what has been the response? When do we see it play out? Briefly.
- [42:32]
G
So in a few months we'll have a hearing on the motion to dismiss where Google and Cartridge will try to convince the Florida federal court to actually throw out this case and sort of so that it won't proceed further. So we'll see that play out in a few months. But we've seen what Google and Cartridge have said. They've said they're separate companies, that they're not related to each other. And Google says that its sort of involvement as an investor, as the mother calls it, or through a cloud circle services partnership that, that the boat had or the fact that they had this deal doesn't sort of, you know, implicate it in any way or sort of is sort of tenuously connected to the harm that was caused here. And in a lot of these cases, yes, as we've seen with say Autopilot and Tesla too, it's very hard to sort of, you know, connect the harm to the technology. So, you know, this whole question about who's to blame here is sort of a big one and still sort of unraveling in court starts when it comes to different technologies. And in this case, you know, AI is coming to the forefront in terms of these new up and coming technologies that are getting really popular. And you know, who is to blame when things go wrong and you know, when there are unexpected turns like this.
- [43:44]
A
One, Bloomberg's Multi Nike. It's a story that people must go and read. We thank you so much. Sticking with all things, Nvidia CEO Jensen Huang is set to deliver his keynote address. Invest in just a few hours time at gtc, what can we expect? Bloomberg's Carmen Reineke is here and well, the shares are selling off in anticipation. So what can you do to steady people's nerves?
- [44:07]
D
Yeah, so what people are really looking for here is some near term visibility and optimism in what Nvidia has coming. So some of the biggest things that they're looking for are comments on Blackwell Ultra, which is expected in the second half of the year. Rubin Other updates in next gen GPUs I think the other thing that people are really looking for from Jensen is countering the bear case here, which was really sparked by deep seek earlier this year. And that is really just saying that, you know, there won't be a cyclical downturn in spending from these big AI companies that they don't have too much compute going right now and that Nvidia will still be sort of the top of mind, you know, chip maker place that they're buying for, you know, much time to come.
- [44:48]
A
Have we seen any dip buying ahead of this?
- [44:50]
D
You know, dip buying in Nvidia has been really interesting this year. Usually we see people rushing back in so quickly to buy the dip and this year we've seen investors really let the stock draw down much more than it has in the past. So I mean it's down for the last two days that, you know, is after having a big sell off. So we really haven't seen a ton.
- [45:09]
A
Of dip buying $2.8 trillion company only. Carmen Reinecke, thank you so much for joining us. That does it for this edition of Bloomberg Technology. Don't forget to check out our podcast. You can find it on the terminal as well as online on Apple, Spotify and iHeart. This is Bloomberg Technology.