Bloomberg Audio Studios, podcasts, radio news. Bloomberg Tech is alive from coast to coast with Caroline Hide in New York and Eva Low in sentences go.
This is Bloomberg Tech coming up.
President Trump advances plans for American investors to buy TikTok's US operations, with a potential value of fourteen billion dollars.
As we discussed the later day I bubble warning, hedge fund manager David Einhorn cautions that huge spending on AI infrastructure may destroy vast amounts of capital.
And our conversation with Qualcom CEO and the changing chip landscape in the wake of Nvidia's Intel deal, and.
A lot more, a lot more, including the market look over the last five days, we have seen pressure on tech stocks more broadly. You're putting in the macro perspective of maybe we lean away from just the Federal Reserve being the key function tactic here, and indeed we start
to question valuations more broadly. We're off more than a percentage point on the NASDAK not significant moves, the biggest drop in the five day basis that we've seen since the beginning of August, but still notable in terms of sentiment. Where sentiment has really changed is Crypto's strategy were formerly known as MicroStrategy, the worst performer on the week, and I'm looking at Bitcoin off by five point six percent key day for options expiring today, ED, what are you looking at?
Okay, this is a story that's moving tech and moving markets. Intel and Global Foundaries are both up significantly. The Wall Street journals reporting that the Trump administration is looking at a policy where it will require US chip makers on a one for one basis to have the equivalent number of chips manufactured in the US as they do their customers import from overseas. We'll try and match that, but
it's moving markets. Then there's Oracle, so Oracle on a five day basis on track for its biggest weekly drop since April. We know more about the structure of a deal for US TikTok, the breaking news in the last hour Bloomberg reporting citing sources that even in the event that this deal is done and completed, China and byte Dance will still take more than half the profits from US operations of TikTok. Let's get all the details with Bloombo's Alex Levine, who covers social media and has been
all over this story. Let's start with what happened out in the White House last night. The President, the Treasury Secretary all explaining to us, along with the Vice President Alex the structure of this deal, what is new and what do we know?
Well, I think what we know about the deal so far is that Oracle is set to play a huge role in helping the new US TikTok entity secure user data, and also the algorithm that is going to be leased from byte Dance for the United States to use and retrain using all US user data. What we saw from the White House yesterday at the Executive Order signing, there was actually not that much new in terms of what we know as far as China where China stands on this.
There are a lot of questions from reporters about what sorts of nod or what sorts of approval the Chinese president has given to Trump, and it seems that Trump has been sending some mixed messaging on exactly where China stands. But I do think that one thing that is becoming clear is that as we begin to learn more about the structure of the deal and what by Dance will
get from it. There are more and more questions, and I think some more scrutiny and even criticism from members of Trump's own party over what the eventual what the eventual outcome will be.
I think exactly that the Chinese embassy has made the latest statements say the US side needs to provide an open fare and non discriminatory environment for Chinese investors.
So no, yes, go for it.
You can take this part of TikTok and start reworking in the underlying algorithm. For me, the question is is Oracle the right person to be overseeing the algorithm changes? And more broadly, how on earth are they going to separate the US version? And more broadly, what about the latest voyce is reporting that we're seeing that maybe byte docetells.
End up still controlling operations with that in any way satisfied national security.
Issues hit well, I think it depends how you ask on this first question that you had, which is is Oracle the right partner for this? If you asked the Trump administration, you asked the White House. Absolutely, they said yesterday that Oracle has been partnering with TikTok for several years already on trying to cordon off US user data from bye Edance's global operations in China. But I think one thing that's very top of mind for me is that TikTok and Oracle have been working together on this
for years, and it has been an imperfect solution. This was proposed as a solution to the Biden administration back in twenty twenty two, and after years of negotiations with a framework that looked almost identical to what we're now talking about with TikTok and Oracle today, the Biden administration was not comfortable that this would adequately solve national security concerns.
So you're starting to hear from lawmakers and others in across Washington that this might just be a new version of that, and if so, that that's just not going.
To cut it, very similar to Project Texas from your own reporting, Alex Lavigne, thanks for breaking that down. Still so many questions to answer, but let's just talk about the advertising side of all of this. I'm beginning Rachel Tipograph found our CEO of analytics platform, Mickmac with all these questions remaining about how on Earth the algorithm continues and whether we all remain one hundred and seventy million US users.
Using it as much.
What are advertising decisions being made right now?
Advertisers in a weight and C mode. Right now, it is still commonplace for TikTok, and we're actually seeing all time high traffic in September from first twenty percent of total traffic from advertisers was allocated to TikTok. If we date back to Q one, it was ten percent. So people still see TikTok as a place to build brand and drive performance all at once. That being said, major
advertisers have contingency plants. The moment that TikTok America is created, I don't think we're going to see traffic at twenty percent. These major advertisers are going to wait to see if the user base is still there, if the engagement is still there, and if the ad performance results are still there. If those things are to be true, then absolutely you'll
see dollars move into TikTok America. But in the meantime, you're going to probably see a short term boom to environments like meta and alphabet where there are other places to invest your short form video ad dollars.
How is this going to impact efficiencies for brands? Because at the moment you can go to one platform and have global reach.
That's going to be really important.
For some of these global companies if they have to silo off a US targeted ad versus global.
Absolutely, there's a domino effect here. A lot of these major advertisers, they have these global joint business plans in place where they're guaranteeing ad dollars across multiple geographies. Advertisers have major question marks. Are those agreements going to be thrown out the window? And then from the actual operating standpoint, there has been a single platform to do buying, measurement
and optimizations. It is unclear now if advertisers are going to have to log into multiple platforms and then Carolyn Prer point, what are the results going to be? Which is why major advertisers they have choices they can go
invest dollars in other places. It's the sm B businesses that have been heavily reliant on an environment like TikTok or Amazon where they might go all in, but if they don't see those results, I promise you you're going to see dollars move out of that environment back into places like Meta and Google.
Rachel.
What we do know from the reporting is that there will be a retraining or even from the ground up rebuild of the algorithm. How worried are people in your community and your network that that basically removes what makes TikTok good for them, that the algorithm is not as effective as it currently is when Oracle has been able to review it and change it.
Yeah, it's less about the algorithm, and it's more about will user engagement still hold? And will the effectiveness of the ads, meaning the cost per impression or the cost per click or the cost per sale.
Will those hold.
If those things don't hold, you will see dollars move out. So the algorithm is a mechanism to make those things true. But what advertisers are really concerned about is results.
What Bloomberg has reported is that byte Dance is still going to get fifty percent of the profits from a US TikTok, and that MGX, a state backed investment vehicle of UAE, is one of the anchor investors. If you're a content creator on TikTok, or you're an advertiser on TikTok, do you care about either of those considerations.
If you're an advertiser in TikTok and you're a global brand. I'm talking Fortune one thousand brands. You care about two things. You care about performance, but then you also care about brand safety. And so if they don't feel that their concerns are being addressed by this new ownership structure around brand safety, then yes they will care. If they believe it's being addressed, then it's irrelevant to that creators. Creators care more about the culture and the community. Will the
community still stay in TikTok? Will the vibe still be the same? They don't really care about the ownership structure as long as the community engagement is there.
Let's talk about a name that you said that maybe you will be benefiting amid all of this, and is meta and the move to Instagram in this moment by advertising dollars. But what's happening in Europe at the moment is interesting with meta as well and the UK and being forced to basically add a subscription offer rather than offering just advertising. How advertisers looking at that, how much do they anticipate people will go for an ad free version of Instagram.
I love what's happening right now because we can look at mcmac's data globally, we do operate globally, and it's really interesting. The laws that are taking place in the EU and UK when it comes to advertising is way more stricter than the US, and as a result, platforms like Meta have struggled in recent years in those environments. In twenty twenty three, Meta had to introduce the same ad free tier in the EU, and there's been no
public data whether users have adopted that or not. The only thing that we've seen happen is that the price per tier come down, which is a signal that maybe adoption was initially there. Fast forward, the UK government is adopting very similar rules as the EU, where they're putting very strict guardrails in place and how platforms can target users and essentially you have to opt in one hundred percent and if you don't opt in one hundred percent, then you have to pay for an ad free tier.
What's interesting is that right now in the UK, seventy five percent of ad investment is happening in Meta. Meanwhile in the EU it's closer to thirty six percent, which is another signal that if this happens in the UK, you're going to start to see ad diversification outside of Meta into other platforms. So it's actually the reverse of what's happening in the US.
In the US.
Opportunity for Meta to grab market share. In the UK, they might actually lose market share right.
Now, Nick mag Ceo, Rachel Tippograph back on Bloomberg Tech, Back and forth.
Great to have you on the show. A story we're tracking.
Jaomi unveiled a smartphone designed to take on the just released iPhone seventeen, underscoring the Chinese company's broader ambitions to take on US rivals from Apple to Tesla. The Jaomi seventeen, which will be offered in Pro and Promax models just like Apple's Marque series, starts at six hundred and thirty dollars for the most basic model and rising to about eight hundred and forty dollars at the top end, more than one hundred dollars cheaper than the base iPhone seventeen.
Carol, Let's go back to AI now, because coming up computing powerhousers that are feeding the AI boom, they're seeing explosive growth of their own.
How will this compute market evolve? One of that next This is a Boomberg Tech.
The AI revolution is fueling explosive growth for companies selling computer power. Call Weave, despite forecasting losses this year, is valued at over fifty billion dollars, while Oracle's added two hundred and fifty billion so its market cap. That surge is subject of this weekend's essay by Bloomberg Ideas and Culture senior writer Felix Salmon, who joins US now. I spoke to the core Weave CEO last night and I basically put to him that all of the people in
this ecosystem are writing checks they can't cash. The demand is there, but their ability to pay for it. The infrastructure needs to be built, maybe it's not. And you make a really smart case on just how wide some of the losses are with all the spending as as much as they have momentum in what they're doing.
And what that in financial terms is called is counterparty risk. That if you are selling, if you're building data centers, and you're expecting trillions of dollars of demand, then what you're worried about is that the people who are contracting to pay, like you know, Open the Eye has said that it's going to pay Ora called three hundred billion dollar, that they just won't have that money because they're losing money, and where are they going to be able to find it?
You know, in video can't just throw billions of dollars at them forever. And so the big question then comes, how are they going to be able to hedge that counterparty risk? And the answer to that question is what if compute is actually traded as a commodity on markets on exchanges. That solves a huge number of problems.
And people are already walking that line and already building that offering. It seems like Felix push us forward as to how we could get what you call really these combinatorial auctions.
Combinator yeah, is something called the Milgram assignment auction. And all markets are auctions, right. If you think about the stock market, people are bidding every microsecond for how much they're willing to bid and offer on various stocks. And that's a very simple auction. It's just like there's one object and you can have a single price of that object.
It's the share of stock. Other auctions, other commodities have more complicated things, like in the oil market, you have to invent these standards like ren crude or West Texas Intermedia that people can sort of agree on. When you're talking about something as complicated as compute, where there are hundreds of different variables that people care about, what chips you have, how much uptime there is, what time of
day you're using, all of this kind of stuff. Then you need a very sophisticated auction that is powered itself by AI. But we have these companies one Kronos and Autonomics that are actually building that right now, and really it's super necessary. Without it, we aren't going to be able to see the level of investment that people are sure is coming.
Food for thought, and that's exactly what job description is. Felix Salmon is a great really go dive into it this weekend. But for more on AI and the questioning of valuations around AI as well, let's talk to Stephanie Aliaga.
She's global market strategist at JP Morgan. No, we just have the latest David Einhorn coming out saying I'm.
Worried about the shaft scale of AI infrastructure investment to the moment that billions is going to be lost in terms of capital.
What are you thinking of these warnings around a bubble.
Yeah, it's clear we're seeing explosive demand demand growth that we've really never seen before. That doesn't mean it's not real, but it does mean that we should continue to really look at these valuations and perform these sanity checks.
And what gives me.
Some confidence when I'm performing my sanity check is that this infrastructure wave, as significant as it is, is powered by real demand growth that is already showing up in the numbers. It's funded by real cash flows, and even after all of this enormous spending the four major hyperscalers, they're still free cash flow positive, and so there's a lot of demand, and still the supply constraint on infrastructure is there. So there's no house of cards underneath this
AI infrastructure wave, I'd say. But still I think the AI theme is it's a lot term. One deep seek was one upset that we had earlier this year. There will surely be others, and right now we're in this kind of period of digestion. But I think we're just waiting for that next piece of like bullish commentary and data, and then we're going to appreciate the fundamentals once more.
Stephanie, every phone call I get at the moment about specific infrastructure deal there's an element of debt involved. How closely are you tracking that and are you worried about it?
So the infrastructure spending boom is beyond just those four major hyperscalers right that fifth in line, as you know, came out with a bond deal this week.
So we are.
Seeing that AI is going to cost a lot of money, and you know, the debt markets are coming in to help fund some of those gaps and spending. I'm not incredibly worried about that just yet. I mean, the bargain that bond investors are making is that there will ultimately be enough revenue generated from this to pay back the bonds. That's maybe a more conservative bet than what equity markets are making by putting more money into these lead megacaps
that already have sky high valuations. But more broadly, I think it just shows that, you know, this infrastructure wave has a lot of urgency behind it, and it's beginning to touch many other corners of the market. The corporate bond market is one of them, but we're also seeing it in utilities and infrastructure and public and private and so forth.
I mean, a forty year bond with what was it ended up a yield of just about one point sixty five percentage points.
So for treasuries feels.
Like, I mean, relatively little to have to pay for the price of that longer term debt, but there is belief.
I want to go back to the comment you made, this isn't a house of cards. How do you feel though about some of.
The circularity arguments that are going on that basically it's the same people are buying, basically giving money in equity for promise of buying their GPUs later.
Absolutely, markets have been rewarding large capital spending commitments, and we should not be rewarding redundant capital spending commitments. And I think that is one of the I guess the sanity checks that we've all performed this week is okay. It seems like capital slashing from one balance sheet to another balance sheet. But a lot of the comparisons that are being made to the dot com bubble of companies selling bannerats to one another trying to juice up their valuations.
What's different this time around is the spending is grounded in real infrastructure, real chip spending, real data centers, and so that gives me a bit more confidence on that. But we know absolutely should not be rewarding companies for redundant capital spending commitments.
Stephanie, really quick, can the global energy sector.
Meet the demand?
That's the million dollar question, right, billion trillion dollar question. I guess The reality is it's going to take all hands on deck. And while a lot of data center power commitments have been made in fact this year enough to power New York, Chicago and LA for a year. That's the degree of power commitments that have been made. A lot of that's in the pipeline. The reality is
we are confronting in very aged infrastructure grid. It's going to take time until the twenty thirties at minimum, until nuclear power, even new natural gas power is going to be there. In the meantime, there's hope that we can get a lot more efficient and that the demand is it's tracking up, but it's also tracking up gradually, right, so we're not too worried about that real I think bottleneck moment happening, it seems a few years into the future.
Until then maybe we see more efficiency gains or the costs of compute have come down ninety eight percent already.
Stephaniely Aliaga, global market strategist at JP Morgan Asset Management, thank you very much. Now coming up, Bloomberg Tech Asias sat down with Holton Nu from bay Too Apollo, China's biggest robotaxi operator at conversations.
The next this is Blenberg.
Tech the race that bring driverist taxis to the world. It is on and while the likes of Weimo have focused on establishing domestic strongholds, Chinese companies are aggressively expanding internationally. Here is by do Apollos, Global manager of overseas Business. You speaking with Blumberg Tech Asia, Aldreders.
I think both Waymo and Bido Apollo, we are the only two companies who provide over one thousand cars in all the cities.
What does your expansion plan and your and your scale up look like.
We have already signed several legally binding contract with different partners globally. Any single one of them is regarding one thousands level of robotaxis deployment. I believe once we remove the safety driver from the car, you know, the numbers of deployment can can raise gigantically right, since what we are aiming for is uh is the real AUTI of striving is without safety driver.
With working with us companies Uber and Lyft. Do you see any risk of geopolitics coming into the equation as well?
From bidose perspective, we think both China and the US have enough ability and wisdom to overcome such kind of geopolitical problems or difficulties. So we believe since we are also commniests working for for new technology, also for profit for profile, generating profits for for both companies, so I think we're yeah, everything will be good in the end.
That was by do Apollos Holton New along with our own Annabel Drulers. Check out the full episode of Bloomberg Tech Asia online or on the Bloomberg terminal. A lot of stories coming out of the region now coming up. More on TikTok as President Trump advances plans for US investors to buy the social media platform in the US. About that story. Next, This is Bloomberg Tech. Welcome back to Bloomberg Tech Cara. I'm sounding the tech ipo watch Clackson the alarm, and the name I'm looking at is Klana.
Down almost six percent in the session. But the reason that's notable is the shares trading at thirty nine dollars nine cents a share, meaning they are down below the forty dollars IPO price pretty quickly as well. Like think of all the other IBOs you and I've covered in recent weeks and months, and some of them had a lot of long term momentum. This feels like a pretty quick drop back, and there is a broader sell off in fintech right now. But it's one to watch this Friday.
Sorry to bring the mood down, but it's a it's a notable moment.
Oracle. Sorry, I've just got carried away. I'm having a good time.
Oracle over the course of five days is down almost eight percent, also bearing in mind broader momentum that it's had in recent weeks, right, But to kind of sell the news week where we got a lot more detail on its participation in US TikTok and how it's going to operationally and to anchor, investor and board seat member run things. But that's five day drop, biggest in April. Please take it away.
I will, because yes, Oracles run far higher in recent months, and no ononder there's been a profit taking. But we aren't going to drill into Oracle into potential TikTok investment and the role that it plays with Bluemost brody Ford who covers the company. And there is a great opinion piece out of our colleague Davely who really puts out really the question of what expertise does Oracle, a cloud company have when it comes to social media algorithms, brody.
Not a whole lot. I think the UL maybe they a would tell you as much as well. I think it's very interesting how much rhetorical value the administration has placed on Oracle's involvement. They are the kind of technological underpinning of a lot of the operations, but they're the cloud vendor, right, I mean, they are not an algorithm designing company. They're not a business operations social media. They're
not a consumer company period, right. And so as Oracle takes a very large role on the stage of how this TikTok deal is rolled out, my anticipation is that their role doesn't change a whole lot from what it is today, which is a cloud infrastructure vendor.
Brady, you and I have spent a lot of the last thirty six hours on the phone trying to work this out, so let's swap notes. My understanding is that the Trump administration went to MGX which is a basically state backed investment firm in the UAE, and invited them to participate. You also have some details on these kind of anchor investors and what role they'll take.
Just give us your notes please, right.
So, these three anchor investors will likely get about a fifteen percent share each and a board seat. The exact numbers may shift around with the idea is that these are the kind of three anchor non Chinese investors, kind of each coming from somewhat different backgrounds Oracle. We've discussed MGX being an Amorati investment company than Silver Like being a US tech focused investment company, and so I think the idea is showing that, hey, there's a breadth of
folks behind this initiative. Though of course the reporting is very live in terms of exactly what these three companies are going to be buying.
The most brady Ford busy week. Forew a lot more reportings come, Thank you very much. Let's bring in Teresa Payton. She's the CEO of Foota List Solutions and the former White House Chief Information Officer under George W. Bush administration. If we may, I want to go to the UAE piece of this first, because you know, the emphasis from the administration was making US TikTok American. The UAE is an ally of the United States in defense, for example.
You know there is weapons and other agreements there. But in the context of something that is digital, what do you make of MGX participating as an anger for investor and potential board seat holder of a US TikTok.
Well, I think it's very interesting.
Obviously they have deep enough pockets to be a part of this, which is kind of part of the deal negotiation. But I to have some open questions because there's some open questions about how they feel about social media for their own citizens. And so now we're talking about we're going to have some type of a twin algorithm, because by Dance is still going to have the algorithm, but we're going to have it here in the United States and we're going to be retraining.
It on US data.
And so I've got a lot of questions around that board seat and what kind of voting rights, what kind of direction, what kind of influence they may have over this sort of digital twin of an algorithm we're going to have the US and how that retraining's going to happen.
Okay, from a pure technology standpoint, how do you think this is going to work read the algorithm or.
Is it going to work at all?
Well, hopefully they've got some really great engineers working on this, so whenever you kind of go into this mode of there's going to be a master copy and then we're going to have our own local, regional copy, but it's going to be essentially the same, except for it's not we're going to be training it on us data. I end up having a lot of questions like what's the quality assurance for this?
Will it still feel and look the same?
I mean, TikTok may lose it's hip and cool factor once we take sort of the global piece out of it and start.
To run at locally.
So it'll be very interesting to see what the knowledge transfer will be. I know, as part of Project Dallas, TikTok Usa has been doing a lot of work to be in compliance with the Foreign Control of you know, kind of regulations around Siphius, and so it could be they already have people very well trained up to take
the reins of TikTok USA's algorithm. But if they don't, I'm a little worried about how it's going to be too owned and maybe a little bumpy once we kind of do this split and have this algorithm twin of sort of the master copy, but it has to be retrained.
And let's face it, if they're Byte Dance TikTok employees who've been working on Project Texas, who says that byte Dancers going to want them to move to us TikTok?
So we have to question what.
Engineering talent follows through theresa. I'm also interested in. There's been some Reuters reporting and Bloomberg's yet to been confirming on it.
But the idea that actually byte.
Dance as a whole has a lot more control than perhaps have been previously thought. Yes, they perhaps give over the algorithm in a sunch way and it becomes updated and monitored by the likes of Oracle, but they then hold a separate division that will continue to be wholly owned, and they'll control the revenue generating business operations such as e commerce and advertising.
How does that technically work.
And how does that make us understand it from a national security perspective?
Yeah, this is very interesting to me. So I think a couple of things. One, we know the data has been physically separated, but the algorithm itself. The reason why an algorithm like TikTok's was so popular and beat out most US based social media companies, at least as it relates to gen Z, gen A and even you know, candidly other generations. A lot of people have adopted TikTok, and the reason why it's so good is that algorithm
is constantly learning. How does it learn? It has to train itself on data and where you come from, where.
You go to and all of those things.
I don't know how the algorithm stays hip and cool and edgy if it's completely separated from ByteDance, So how does it learn if it becomes a black box of learning? So lots of questions still around how that engineering is going to work and how are they going to set this up. So, for example, there is a term called digital twin where you can actually have a.
Full production copy of production and.
Essentially have it be the same and you know, potentially have the data be separated. We're going to have to have some really smart engineers on the US side of TikTok to really say whether or not there's backdoors, whether or not some of the data itself is truly anonymized, or if we still have a real issue with kind of foreign interest being able to see USB citizen.
Data and theresa I follow up with a relatively sensitive political question here, because a lot of anxiety has been put into the potential biases that creep into this algorithm. If it's Chinese overseeing controlled whatever, you might make a
byte dance. If it's now overseen and controlled by Oracle, what can they put in place to ensure that the users that may or may not migrate over feel comfortable with the control of algorithm by them and what biases they need to oversee in the same way that we question other parent companies such as Metro and Snap.
I love the question asking here because of our lack of governance and guardrails over social media platforms, over big tech, as far as understanding the transparency around these algorithms, how they train things. Do I, as a user get to reset my algorithm? Do I get to see transparency into what's being served up to me or to my children or to my customers if I have a business account.
So, because we don't.
Have that today, this is a great question and it has to be asked of all the social media and big tech platforms, and you're right, we don't know potentially who's sitting on that board, not just Oracle, but the other investors. You know, are they going to have sort of a heavy thumb, if you will, on the algorithms and how they work more to com Obviously.
Evolving story trees are paid in there with us, so appreciate it.
CEO for Talis Solutions form of.
Bush white House CIO coming up of an AI stargate ambitions could super charge them on for aipower, big implications from video and Oracle.
That's next. This is bloombag Tech.
Quil Com CEO Cristiano Arman said that the Nvidia Intel deal from earlier this week is in a way, and I quote the moment they had been waiting for. He spas why talks about the impact on Qualcom and the rest of the industry when the news broke this.
Week, when you actually dig deeper on it, and it was interesting, Actually when did announcement happen, it was interesting.
We were slightly up.
It was kind of neutral to positive to Qualcom, and we saw I think the market stock reaction on arm On MD and and an Intel was up and the way we actually look at this, we're incredibly excited because what he basically means is, I think it's Intel was sending a message that they're integrated GPU in their AI products are They're not going to likely not to continue with that because that's what the market is seeing. It's
going to be in Vidia. So then as in Nvidia comes in to the integrated GPU market, combining their GPU with Intel x eighty six, he started to bring AI into the PC market. And that is the moment we've been waiting for. And I think that's what we do. And I think the Qualcom DNA is anything that is battery power, is high performance, low battery life and a lot of AI. That's that's where we want the market to be welcome.
The CEO Christiano, I'm on there.
Meanwhile, Open ai is pushed to scale up as Stargate project is set to accelerate demand for AI computes, potentially lifting other AI sectors from in Video's hardware sales to Oracle's cloud services for more. We've got mandate seeing senior tech industry analysts of Bloomberg Intelligence, the thing about everything that's happening is a misunderstanding of where the demand is
being sent, You've done a really good job. Like Microsoft is outsourcing to Oracle basically because they don't have their own capacity to deal with this. Just present the thesis that you're trying to get the market to understand.
Yeah, look, I mean open Ai has told us that they expect to get to two hundred billion dollar in revenue by twenty thirty. Currently they will likely close the year at twenty billion in ARR. So you know, when you look at that seventy percent keeger for the next four years, I think deals like what they have signed with Oracle, the three hundred billion dollar deal and then in video signing up that one hundred billion dollar investment
for ten gigawatt capacity kind of makes sense. How they get to that capacity which will enable that two hundred billion dollar in revenue. And we have sliced the different kind of segments they will have. It won't be just CHATQPT subscriptions. They will have API sales and AI agent sales. But the enabling kind of infrastructure for that will come from additional power and data center capacity that they will
have from all the deals that they are doing. And that's the vision, And it comes down to how they execute on that vision.
There was more news with open Ai looking towards call weave a little bit more for some of its compute needs. Now the care we've CEO joined at Ludlow remain a little bit earlier yesterday.
Just to listen to what you have to say.
The challenge that we have is competition among our clients for.
The infrastructure that we have.
Right it is not a question of giving up one client for another client because their demand doesn't completely absorb all of our infrastructure. It's just a question of we have one hundred units of compute to deliver in the next twelve months. We could deliver all one hundred to client number one, all one hundred to client number two, or all one hundred to client number three. And the question is is how are we going to divvy that up among those coins.
Issue is supply side Mandle.
We say time and time again, and how big therefore is the need? I've seen numbers. We start with Baine saying two trillion. Then we saw, of course Ali Baba say it's could be four trillion.
I've just read.
McKinsey's bullish case is up to send point nine trillion by twenty thirty.
What do you have it as?
Yeah, look, I think Jensen sort of laid out quite well with the one hundred billion dollar investment. He said, for ten gigabart power, you need a four hundred billion dollar investment. They are putting hundred billion dollars open. Aye will have to find three hundred billion dollars from somewhere else. But you can actually see if we end up adding fifty gigaward of capacity, that would make up you know,
two trillion number. And so from that perspective, then you start looking at how how much will be the tokens generated from that additional fifty gigaward power capacity. And the reason why that's the right framework is because every querry right now is the function of tokens consumed. I mean, Google says they have nine hundred eighty trillion tokens per month, and that seems to be growing quite rapidly. Every company
right now is quantifying that. And so if you make it a function of power, it's easy to see where we will land up given you know, the tokens consumed per ai agent or per query or any other use case that Jenny I has.
And then we just need to source that power and deep seeing Bloomberg Intelligence, great analysis, go read it.
Thank you.
Meanwhile, coming up, we'll take a look at how Apple lagged so far behind on AI, how.
It can catch up. This is Blomberg Tech.
Apple.
There's long leed Silicon Valley in tech, but in the world of AI, it's fallen way behind. Continuing missteps could arguably threaten the ipho's dominance. Bloomberg Originals took a deep dive into what went wrong and if Apple can catch up.
Sirih Well the name of the guy I had a meeting with a couple of months ago at Cafe Cranell.
I met Zach Wingate at Cafe Grenelle.
Sach actor Bella Ramsey was bailed out by a new version of Apple's Siri that was meant to be available for.
All That feature still has not been released. Originally supposed to come out in April May of twenty twenty five, but now it's on track for Spring twenty twenty six.
Apple took down the ad and is facing a lawsuit over the delay, but the disconnect exposed a deeper issue. The company was losing ground in the AI race.
They totally missed the boat. There was indecision. There was lack of concern that this could become a real thing.
This is a story about how hard it can be to stay on top.
AI is something that wasn't in their DNA.
Continued failure doesn't just hurt Siri, it could threaten the iPhone's dominance. The product line accounts for more than half of Apple's revenue, so missteps with what's become its most non noticeable missing features could be devastating.
We saw this for many decades. Nokia was at the very top of the phone industry. Apple came out with a much better mouse truck. BlackBerry was at the top of the industry.
Apple came out.
With the better mouse truck. They are going to be in danger of the next Apple being created if they don't get their act together soon with some serious acquisitions changes in partnerships.
You can catch the full episode at Bloomberg dot com, on YouTube or on the terminal. Two Wall Street bulls, Tom Lee and Dan Ives are going against the grain of Wall Street, leveraging social media to bring in fresh inflows and outperformance in their ETFs. For more, Bloomberg Cross asset reporter Isabelle Lee joins us this is one of the most read stories on the Bloomberg terminal and the website.
It's not surprising two characters megatech bulls, but they've launched ETFs explain the basics what we need to know about what they're up to.
So these two hf stand out in a CE or four four two hundred ETFs in the US because they engineer a rare double performance, which is outperformance and flows. Usually you're outperformed, but you don't see flows. Usually have flows, but the performance isn't that. So both of them have those two and the both of them also are really this big figures. Tomlely has five hundred thousand followers on Twitter,
dan Ives has around two hundred thousand. So they go straight to their retail investors, and they go straight to them with their ideas and guess what these are. Retail investors buy them up by the billions as well.
Their ideas go straight to a potential retail and institutional investor. And at the same time, say dan Ives comes on this show as a buyside and sell side analyst. He's coming on and trying to say I'm independent and my view is that these companies are going to go higher and he does think they're all going to go high.
He's a permeable. But what if we start to question evaluations. What if at some point.
He goes negative on the idea of certain stocks. Does that damage the ETF inflows?
How does he remain independent in this way.
That's a good point, Caroline, And this is a new trend. We're seeing that these prominent figures are launching their own ETF. So both Tom Lee and Danives have their own really strong base because of their research. They both just send out researchers every day. You get their notes multiple times a day if you pay as a client. And then they thought, you know what, why not let me wrap my ideas into this one big ETF so that my followers or my now investors can just buy those ETFs.
In one click.
And that's a beautiful ets and one click you can really just purchase them. But to your point, that is true because followers are loyal to them, they listen to their every call.
But what if they now it's the question of.
Whether there's going to be some self interest that remains to be c and I'm sure they practice fiduciary and they're doing what's best for their clients, so.
The rest of their interests.
But then it's really interesting because these big name launches from Rubini also he launches own ETFs, even Gmo. Even Rick Reader who's now in contention to Fetchir, he has his own ETF. So it's really a new wave. And I think not all famous people when they launch their own ETFs, not all of them gain traction. For them at least it's doing well.
They are their own personal brands. Hopefully that's thrups some picks real quick. But dan Ives is charismatic. Just explain to those that don't know what he's like.
Where do we begin with Danas every quart he has this quotable. He wears these colorful clothes. He's always in a cap. Everyone is in black and white, and he's just really this big, red, colorful figure. Tomley is the same. They make this big, audacious bets and it really is a SoundBite that you've'll forever listened to.
Interesting mean, they both recently become chairs of digital asset treasury companies as well, so they seem to be in sync on certain areas of the market. Bloomberg's is Abelli great reporting and great story. Go read it that does it from this edition at Blomberg tech Head.
Yeah, don't forget to check out the podcast. You know where to find it.
It's on the Bloomberg terminal, it's online, it's on Apple, it's on Spotify, and it's on iHeart. What a week it's been from San Francisco and New York City.
It's Bloomberg