Bloomberg Audio Studios, Podcasts, radio news. This is Bloomberg Intelligence with Alex Steel and Paul Sweeney.
The real ap performance has been in US corporate high yield.
Are the companies lean enough? Have they trimmed all the fats?
The semiconductor business is a really cyclical.
Business, breaking market headlines and corporate news from across the globe.
Do investors like the M and A that we've seen?
These are two big time blue chip companies.
Window between the peak and cunt changing super fast.
Bloomberg Intelligence with Alex Steele and Paul Sweeney on Bloomberg Radio.
Of Today's Bloomberg Intelligence Show, we dig inside the big business stories impacting Wall Street and.
The global markets.
Each and every week we provide in depth research and data on some of the two thousand companies and one hundred and thirty industries our analysts cover worldwide. Today, we'll look at why the tech giant Microsoft says it will cut six thousand workers. Plus we'll disc us how President Donald Trump is looking to lower drug prices in the US.
But first we begin in the legal space.
Recent research by Bloomberg Intelligence maintains that the Supreme Court would likely let President Donald Trump fire fed chair j palell even if market reaction discourages it. It's a subject of a bipiece titled scotus would likely let Trump fire Pal even if the markets don't. For more, co hosts Alex Steel and I were joined by the pieces author Elliott Stein, Bloomberg Intelligence litigation analyst. Your first asked Elliott, if it is constitutional for President Trump to fire fedchair Pal.
We're not sure yet, but wait, yeah, the way the law is playing out is, you know, he's taken a very expansive view of executive authority and what the president can do.
You know, we see it with the tariffs. We sorry with those.
He's taken a lot of actions that presidents haven't done before, and they're being litigated. And one of the things that he's done is that he has tied to exert control over what used to be called independent agencies, so things like the FTC, the Federal Trade Commission, the NRB, the
National Labor Relations Board. He has removed commissioners at these agencies even though the statutes that govern these agencies have what's called the fore cause removal restriction, meaning the president can only remove them for cause, and he's just firing them at will.
He's also obviously.
Threatened to do the same with Jerome Powell and maybe other members.
Of the board.
And his view is that the four cause removal restriction is unconstitutional because it impinges on his power as the president under Article two of the Constitution to essentially oversee the entire executive branch and the Supreme Court. In nineteen thirty five, there's President saying those four cause removal restrictions
restrictions are constitutional. But more recently the Supreme Court has sort of try to narrow that precedent and said, at least with the couple agencies where you only have one director, those four cause removal restrictions are unconstitutional. And so the president, President Trump is sort of trying to expand that recent precedent to these other agencies.
So do we have other I mean, so the Federal Reserve is one thing. Do we have other agencies that are in the legal process now where someone's been removed and we have a suit?
Yeah, exactly. We actually have quite a few of them. So, you know, he has fired two commissioners at the FTC. Those are in litigation, sort of the most procedurally advanced cases at this point are cases where he fired commissioners at the NRB, the National Labor Relations Board, and also the Merit Systems Protection Board, which sort of oversees federal employees and whether you know they're being fired and hired
for merit or not. Those two cases, we have actually several rulings, and what's interesting is that the judges in those rulings sort of follow along.
Ideological lines and how they rule.
So the trial court judge was a judge appointed by President Obama, and she said that the firing of those commissioners was improper, that it was violated the four cause removal. Those cases are now an appeal. It's going to be an argument next week. But we have sort of preliminary ruins from the appeals court. One panel of three judges, two of whom were appointed by Republican presidents, said that President Trump would have been likely to succeed on the merits.
But then that ruling was then revisited by the entire court, which is dominated by judges is appointed by Democrats, who said that the firing was improper.
So we're sort of seeing how this might play out.
Is this are we really in a truly partisan judicial world or Is this just simply different interpretations or a stricter looser interpretations of the Constitution.
It depends on the issue.
I would say on this issue, it seems like the judges are essentially ruling along ideological lines, where you have judges appointed by Republicans saying that the fore cause removal restriction really does impinge on the president's executive authority, and you have judges appointed by Democrats saying, you know.
It's okay to put some limits.
The four cause of removal restriction would still allow the president to fire someone if they did something improper.
In terms of say the Supreme Court issuing and a ruling that then is ignored, is there precedent for that like over history?
Well, we've actually sort of seen it with TikTok right, where the Supreme Court upheld the TikTok ban and the president instructed his attorney general to ignore the ruling and ignore the ban. So we have sort of started to see some issues like that. But in this case I expected Supreme Court to if if he fight well on this issue, generally I expect the Supreme Court to side with President Trump.
Interesting, so what happens when a president defies the Supreme Court.
Like in TikTok, nothing right so.
Far, nothing so TikTok's still The remedy is for Congress to.
Impeach the president and then the Senate.
To convict, highly unlikely in this political environment.
I also like pushed out the rights. It sort of feels very blurry. It doesn't feel like he completely ignored it's it's like we're working on it.
We're just pushing out the deadline.
That feels a little differense Yes, that's essentially what he's doing.
But at the same time, there's a law that Congress.
Passed that said, you know, it has to be banned by a particular date. The Supreme Court upheld that law, and he's essentially ignoring the details of the law. I agree with you, he is using that as a reason, but the law is still on the books and it's.
In violation our Thanks to Elliot Stein, Bloomberg Intelligence litigation analyst, we moved next to some of the news in the aerospace industry this week. We heard that China has removed a month long ban on airlines taking delivery of Boeing planes, and this follows a breakthrough in trade talks between the US and China that temporarily slash terrace on each side. The resumption of deliveries to China will be an immediate boost of Boeing, with around fifty Boeing jets to be
delivered to China this year. For more co hosts, Alex Steele and I were joined by George Ferguson, Bloomberg Intelligence Senior Aerospace, Defense and Airlines analyst. We first asked George for his take on this week's news.
I think what it means for Boeing in the short term is that they could probably get you know, some of these inventory airplanes they've built. They're sitting on Tarmax, they're maintaining them, costs bowing a lot of money. They can get them delivered into China. Serium expects thirty four of those airplanes to go in through the remainder of
the year. They can get them delivered into China, get them off their books, you know, sort of get rid of that extra expense they it costs for them to manage those airplanes, so they could just get onto focusing on the factory floor and building brand new airplanes. So I think it's a positive. We'll see how long it lasts. But it's definitely a positive.
Was this sort of baked into the US China trade talk or was this irrespective?
Do you have an idea now, I'm not sure what's baked into the US China talks. I do think that there's an opportunity here for Boeing to get to even new orders from China. We haven't really seen material orders out of China since twenty eighteen, so last decade before the pandemic. Relations have been a bit strained since then.
I think it's just it's one more piece of evidence that shows Look, as the US and China get into negotiations over trade, you're going to see heavy fracturing, and then you know, we see the markets kind of get
very distressed about that. But behind the scenes, there's discussions about, you know, putting that relationship back together, and Boeing probably factors in high in those negotiations, and so I think you could go from a, hey there's despair here to hey, look there's a three hundred airplane order for Boeing coming out of China. So it's going to be a real up and down I think roller coaster.
Here, George, are there airplane manufacturers in China?
Or do they depend solely on Boeing and Airbus.
They depend largely on Boeing and Airbus. There is a manufacturer, Comac, you know, they've been building a seven thirty seven, a three twenty you know, you know, substitute. They call it the Comac C nine one nine. The airplane's rolling out slowly. I think we saw fifteen or twenty deliveries last year. This is a country that probably needs at least one hundred and fifty narrow bodies a year, so not going
to fulfill all their needs. There's about that many flying right now, and when we look at the flying statistics for that airplane, it's not flying as frequently as a three twenties and seven thirty seven. So it's it seems to us that the maturity that product isn't there yet for the Chinese to substitute out all of the Western jets.
Right, So there's still a little bit of a way to go. You also have to wonder it's just in general the comeback that Boeing has been able to make. George, how would you describe the situation that Boeing is and now versus a couple of years ago?
Yeah, you know, I think this year is what we expected last year to be before the door came off on the or the door plug and the Alaska chat. We're starting to see good flow through the factory and the delivery numbers we've seen. We're starting to see them deliver that inventory of built airplanes. Like I said that Dragon costs pretty heavily. We're hearing good things about supply chain. They're going to pull Spirit Arrow Systems into Boeing by midyear.
That was a big part of their problems. I think pulling them in helps them control quality at Spirit. So I think this story this year should be all about improving quality, improving throughput in the factory in the back half, some good cash generation. So I think this year is whatever they thought last year would be, which could be a pretty good year.
So where are you, George and your model in terms of seven thirty seven kind of build rate here? Because I know that just with the economic model for Boeing that just drives cash flow the more planes I can get through the line.
We're looking for like a forty five to fifty percent increase in seven thirty seven deliveries this year, so really head a increase, and again that's going to be the big driver of cash generation for the company over the year and probably the ex the year. A build rate per month of thirty eight, we think even maybe into the forties kind of build rate, which is an area where they should be able to generate some profits.
Our thanks to George ferguson Bloomberg Intelligence, senior Aerospace, Defense and Airlines analysts. Coming up, we'll continue with more on the planemaker Boeing, which landed its biggest ever aircraft order. You're listening to Bloomberg Intelligence on Bloomberg Radio, providing in different research on two thousand companies and one hundred and thirty industries. You can access Bloomberg Intelligence.
Via b I go on the terminal. I'm Paul Sweeney. This is Bloomberg.
You're listening to the Bloomberg Intelligence podcast. Catch us live weekdays at ten am Eastern on Apple, Coarclay, and Android Auto with the Bloomberg Business App. Listen on demand wherever you get your podcasts, or watch us live on YouTube.
We continue with more news from the planemaker Boeing. This week, Boeing landed its biggest ever aircraft order, with Qatar Airways agreeing to purchase up to two hundred and ten wide body aircraft, including the seven eighty seven Dreamlier and the seven seventy seven X model. It's a deal worth ninety six billion dollars and this comes a year after Boeing slogged through a deep executive shakeup and prolonged crisis.
For more.
Co hosts Alex Steele and I were joined by Sid Phillip Bloomberg, Deputy team leader for Global Aviation. You first as Sid for his take on this week's news.
This is a significant deal for Boeing because they can sort of add another sort of milestone order to their backlog, especially for their seven eighty seven, which is the sort of state of the art white body jet, as well as potentially for the Triple seven X, which is again a jet that Kataraways has ordered and is still to
be certified. And so as airlines sort of look to the next generation of aircraft that they're looking to order, this is sort of these are these You're going to see these mega deals coming up where everyone's looking to sort of replace their current generation planes with newer models which are more advanced, better fuel economy, and that sort of great for Boeing in terms of just building out that backlog and sort of selling out aircraft past the end of the decade.
What is the delivery time for something like this, But it's clearly had some operational execution issues exactly.
I mean Boeing has been struggling to deliver both the seven eight seven as well as seven three seven Max and get the seven Triple seven X, which is their latest sort of generation aircraft certified. So it is not clear on the details for what the sort of intricacies of this Qatar deal are, but potentially these will sort of play out and go out into the next decade because I mean, they've got a significant backlogo aircraft that
need to deliver in the short term. So this is not going to be an aircraft deal that sort of happens in the next two or three years. Is going to be a drawn out deal that goes into the next decade.
Well, Bombing's finally digging itself out of its hole from the last several years, so that kind of goes to the big question said, it's all about production, you know, getting planes off that and people just like George Ferguson fro Bloomberg Intelligence says, just focus on the seven thirty seven Max deliveries.
That's the number.
Is there a sense that Boeing's making improvements there?
We have seen Boeing deliveries inch up, and we've sort of they've had sort of milestone in terms of just hitting better production levels on the sevent three seven, And we've heard from Aline customers, including United Airlines, saying that the deliveries of the seven three seven Max are sort of going up and they're waiting on the seven eighty seven, which is the aircraft that we understand that airways ordered. That's also going to be sort of key to getting production and profitability going.
What other kind of pitch can Boeing make to outstrip air Bus? Because in a duopoly market, and I appreciate that China has some smaller players, but if you're dealing with a duopoly, like I mean, people got to buy your planes exactly.
I mean, we have a duopoly in at the moment.
And does that change it all? You think? It sort of has.
Changed a little bit in terms of just the sort of there's a lot of politics behind aircraft orders again, and that's something that we've seen where Trump's been announcing these mega Boeing orders. I mean, he announced a deal for thirty planes in Saudi Arabia, and so there is a lot of politics attached to aircraft orders. But at the same time, both Airbus and Boeing are sort of sold out of aircraft until the end of the decades. So it is sort of you you're notching up all
these orders. But at the same time, the biggest thing that airlines are talking about is give me my plane, because that's really the biggest question where airlines are saying that plane deliveries are delayed, and the air Bus is repeatedly talked about. How supply chain issues continue to haunt the industry now five years after the pandemic.
Yeah, which I can't imagine, although again I'm what I understand is one of the bottlenecks is labor getting the This is not just you know, sitting then assembly line and hitting a hammer. You've got to be really, really skilled at labor, and during the pandemic you lost a lot of that and retraining these folks are getting new people in it.
I'm hearing that's a big problem.
Labor is a major problem as well as I mean, just one part being shot holds up the entire production line. I mean, you have millions of parts that go into an aircraft, and e US and Boeing are the final assembly lines where the plane actually gets put together. But then if parts are not available in the sort of process, you can't actually build those planes. And that's where the bottlenecks come up because you have smaller suppliers who are
struggling financially. You've got parts in short supply. I mean EBUS talks about it's sort of it's whack a mole where you get one part sorted out and something else, and so it's really a question of getting that supply chain to sort of hum along nicely and then actually building those planes and delivering those planes because you've now got record backlogs at both Boeing and Airbus.
So are they exemmed from tariff issues?
That's again a really important question. No one's quite clear about in terms of what the tariff impact is going to be because for fifty years almost that the aviation industry has largely functioned tariff free, and the tariffs added
complexity to it. Because now everyone's not quite sure if aircraft are example, I mean, you've seen like all sorts of industry players and industry groups making representations to President Trump saying that we would like to have tariffs removed on aircraft and aircraft parts, and so we're trying to see where that sort of actually lands up, because potentially aircraft manufacturers and airlines may have to pay tariffs on planes depending on what they're in and what parts they have.
Our thanks to Sid Phillip, Bloomberg Deputy team leader for Global Aviation. We move next to news in the tech space. This we heard that the tech giant Microsoft will cut six thousand workers across the company to reduce management layers. This will amount to less than three percent of total headcount for more. Co hosts Alex Deal and I were joined by anorak Rana, Bloomberg Intelligence technology analyst. First ask anarog for his take on why Microsoft is cutting jobs.
The numbers three percent does not say much, but at the same time, Microsoft is not hiring at the same pace that it used to. And I think one of the areas where it is deploying AI is coding or you know R and D. That's an area where we anticipate productivity to improve because of the tools we have, and not anticipating headcount to grow.
Up.
Second thing, Microsoft is investing very heavily in AI right now. The capex is going to be extremely high, you know, as North doos eighty billion dollars for the next twelve months, and you know that they have to offset some of that pressure on gross margins with cuts on the operating expense side so that the net operating margin doesn't get impacted. So I think I think it's a little bit of all the above.
Was this a surprise, not really.
I mean three percent to me does not come as a surprise. Three percent is you know, in an environment like right now where companies are not anticipating the real attrition rate. So on a normalized basis, if a company is seeing attrition employee attrition between ten to fifteen percent, we're not seeing that right now, so the company has to do something in order to get back to the
normal range of their financial planning. So it is I said, this doesn't sound too much to me, but they could be a little bit more based on you know, who they are trying to get rid of. If it's higher on the paychecks, then I think it has to do with, you know, protecting your margins.
An Rock talk to us about just what the investment thesis is regarding Microsoft right now, given some of the tariff uncertainty, given some of the geopolitics uncertainty, what's the call on Microsoft these days?
You know, for the last few years we have been saying this is probably one of the cleanest stories in AI and software land because it doesn't get impacted with tariffs. There is some marginal secondary impact because of you know, if we do get into an economic slowdown, But the real benefit for them is they are hosting chat GPT, the most favorite app of consumers right now, and they
get the benefit of that. So if you have let's say, you know that your user base goes from two hundred million users to three hundred to four hundred now we are close to about five hundred million users, it really helps Microsoft because they are hosting that app, and as those users use chapid chat GPT for a longer period
of time, they get to make more money. So I think that's really one of the bigger benefits where they are offsetting the risk that we are seeing that other software vendors are seeing because of the lack of headcount growth across the tech landscape.
So based on that, do we think that We're going to see other tech companies follow suit in this particular way. Like if Microsoft is a creme de la crem as you see it, who's next.
So Alex, when you look at it, as I said, three percent does not scare me as much. But if over the next six months we start companies come out and say, well, I'm laying off like seven to ten percent of my workforce, that is concerning. Because that's concerning for two reasons. One, these companies do buy software from other companies, such as a work Day or a sales force, because these two companies are seeing some pressure on their seat growth because their end customers are not hiding as much.
So lack of hiding is concerned not just for the companies or the industry themselves, but for the software landscape as well, because primarily this is a seat based monetization model.
An RAG before we got all enmeshed in terroris. Over the last few months, we seemingly talk to you every day about the AI story that's kind of full off the radar screen a little bit here for I think Global Wall Street, not for you guys. What is the AI story today? Where what's the feeling in the marketplace? About AI as a growth story for tech.
Yeah, one of the things code one of the things that deep Seat did for the entire space was divided the space in two different buckets. Was the AI infrastructure play and the other one is more on the application side of it. And Microsoft I think guided them a little bit towards that framework because when you look at it, Microsoft is really focused on the inferencing side of it,
which is hosting the applications. Things that are more on the infrastructure side, whether it's the chip building from Nvidia or a recent hype like core Viv, they have not done as well as some of the other vendors because people are wondering, why do I need to play the infrastructure game because it's CAPEX heavy and we do not know how things are going to shape up in the next two to three years. One thing is sort of certain that even after three years, the number of use
cases on AI is going to rise. So you want to be on the application or the inference side of things rather than the infrastructure side of things.
Right, and that's going to be a lot more profitable going forward, But it also requires something very different. At the end of the day, Right, you're going to have basically locations that are closer to the actual use point, right, so closer to where I am and where you are, versus out in the middle of nowhere, but you have a huge heavy load. How does that change either investment or power usage, et cetera.
That's probably one of the most important questions out there right now. So somebody like a Microsoft is saying, for my training needs, I will go to those data centers that are out in the boondogs and middle of nowhere, and that could be my data center, or I could lease it from somebody like a Corview. But for the application side, I'm going to create or lease smaller data centers that are well connected where I'm going to do the infrancing or from the user point of view.
Our thanks to Anna rog Rana Bloomberg Intelligence. Technology analysts coming up on the program will break down how the Chinese AI start a deep seek became world famous. You're listening to Bloomberg Intelligence on Bloomberg Radio, providing in depth research and data on two thousand companies and.
One hundred and thirty industries.
You can access Bloomberg Intelligence via bi go on the terminal.
I'm Paul Sweeney. This is bloomber.
You're listening to the Bloomberg Intelligence podcast. Catch us live weekdays at ten am Eastern on Applecarplay and Android Auto with the Bloomberg Business App. Listen on demand wherever you get your podcasts, or watch us live on YouTube.
We move next to news in a pharmaceutical space.
This week, President Donald Trump sign an executive order to lower drug prices in the US. You order as drugmakers to lower prices voluntarily or face regulatory measures. Pharmaceutical companies had expected and feared action on drug prices. However, the order was far weaker than they anticipated. For more on drug prices, co host Alex Steel and I were joined by Sam Fazzelli, Bloomberg Intelligence, Director of Research for Global Industries and senior pharmaceuticals analysts.
We first asked Sam to walk us.
Through why we pay more for drug pricing in the US and whether that should be reduced.
The thing in the US is that it's a price that's set by negotiation between farma companies and insurers, pharmacy benefit managers, and payers. It is not dictated by the government, still wasn't until recently, whereas in Europe and many other countries you have a single payer, and that's the government, and they play hardball so they can get lower prices
because that's the way they operate, and they had. There are many situations where in the UK the National Institute for Clinical Accident refuses to believe or accept that a drug has got its value added, so it's not even offered even at the negotiated lower price. So that's the issue here, which means that as soon as innovation happens is out in the market in the US, the day that the product is approved, they can get on and
sell it. Over in Europe it could take a year two years for a drug to come through, and maybe with this new executive order sometimes the companies will go where I'd rather not launch in Europe.
So is it realistic to expect that drug prices in the US will come down?
Could come down?
See the thing is, Paul, if I'm having a tough time to balance this out. If the intention is for to get the rest of the world to pay more, which of course helps the farmer companies, then I don't see how that helps the US drug prices come down.
Right.
If they aren't able to increase prices in Europe, then then the government in the US is saying, well, you should give us the same prices to giving elsewhere. Then innovation disappears because at the end of the day, if you cut your revenue by half, you won't have and you spend fifteen percent of that on R and D as what your RND faults simple maths. So how this plays out would be quite interesting to see. And of course we have to also remember that there isn't a
price in Europe. So we just check the price of we go V in the UK at the effective dose two hundred and thirty dollars okay. In the US, literally the Novo sells it for four hundred ninety nine dollars to the direct to consumer. Okay, it's about twice as much, okay. So let's say in Europe it goes up to three hundred dur in the UK, and how does that change the US number? And we don't even know whether the company actually gives that price because then there's a little
small print. We have been doing a commer with the NHS that are commercial deal, So what is the price? How is the US government going to find that price unless they incorporated into the trade agreements that they're doing, which is what should be happening.
Here, So going for so what about the PBMs, So that was singled out the pharmacy benefit managers as sort of the middlemen between insurers and the government and the drug makers, et cetera. What role do they really have and are they the ones most at risk?
I would say they are the ones most at risk. I would say they should have always been the ones most at risk, because I don't understand the reason they exist in the middle. I mean, that's the I can show you a chart of how the US payment system works, and it varies by insurance. You can go from one job to another and the cost of your drug goes up, right because you have a different insurance from one state to another. So it's it's that just really isn't is untenable?
And it's nice to see that the government administration is not talking about making it easier to sell directly to consumers. So let me give you an example. We have data on the terminal that gives you an ability to calculate what a drug revenue should have been in the past quarter volume times price, right, then you go and see what they actually reported. Sometimes for Insulince, for instance, is
eighty percent lower eighty percent. Now the PBM say, I know we passed that on back to the to medical party. Why do you even take it? Why do we need to pay an administrator in the middle here?
That's great.
So what are your companies within your food chain there in a big farmer.
What are they sayings?
Because this issue is new, we've been plaining about high drug prices forever.
What are they saying as to what is likely to really happen.
They're not saying very much, Paul, you know, and in this certal circumstances, the best thing to do is to keep your head down and see how how things work out. And of course today has only just come out, so I think they all to a t say the PBMs are a problem, we need to get rid of it. That so let's say that. But that's not today, that's been going on for ten years. They also say Europe
should pay more. But they also say that when you do the net price calculation, in many instances I actually sent you a code from Pascalsorio, the Astrisenica CEO, saying that actually in many cases the prices end up being about the same, maybe a little bit different. So all of that needs to be taken into account, and we need to decide if they're going to push Europe higher to pay more. Is the US also happy to move some of the R and D to Europe. Maybe that's
something that the European governments will ask for. Fine, will pay more, but bring some of your R and D here, Bring more manufacturing here. Oh.
Thanks to Sam Fazelli, Bloomberg Intelligence, Director of Research for Global Industries and senior pharmaceuticals analyst.
This week we focused on a Bloomberg.
Big Takes story entitled deep Seeks Tech mad man is threatening US dominance over AI. You can find it on the Bloomberg dot com and the Terminal. The story looks at how Chinese AI startup Deep Seek became world famous and swallowed a trillion dollars of stock value from Nvidia and other big tech companies. For more, co host Alex Steele and I spoke with Austin Carr, Bloomberg Technology Reporter. First asked Austin to give us more background on deep Seek and the individual behind it.
Deep Seek, as you'll recall, gained a lot of attention a few months ago when it released its new R one reasoning model as well as another one called V three, And the big deal about this was they sort of claimed to have trained it without access to a lot of the most cutting edge in Nvidia chips, and they also trained their models for a lot less, at least on the surface than a lot of the model's equivalent models here have been trained in the US, and that
raised a lot of questions about some of the fundamental notions about how many chips are needed to train these big AI models and how much it's going to cost to develop them. The reason it was also a big deal was just because the startup was sort of very mysterious.
There had not been much written about them before sort of January twenty twenty five, and we spent a lot of time the last couple of months talking to about a dozen former employees that worked for this company, it's founder Leong win Feng, as well as forty sources closed to the Chinese AI scene, to really just give you a sense of how their sort of market is developing very differently than what's happening in the US, which is they're doing a lot more with less and sort of
just trying to innovate despite the constraints that the US has placed on it with all its export controls.
How did he do it?
How is he able to undercut the price in such a tremendous way.
Well, I think that's one of the big topics that we explore in the story. We trace it actually back to the history. Leong Win Fang had started a quant fund back in about twenty fifteen twenty sixteen, and you actually see a lot of those sort of quant sensibilities looking for efficiencies in the market, which they later applied to their AI models. It was sort of a sort of spinoff off of what they were doing with that
quant fund that became Deep Seek. One of the big things they did was called mixture of experts, basically a lot of the sort of when you used CHATCHYPT back in the day, you asked it a question, it activated the entire model. So think of that like your entire brain is activated just for the question two plus two. What Deep Seat did that was a little bit different
was chop it up into what are called experts. So every time you ask a question, it actually routes that to the particular part of the brain that is expert in math, or expert in physics, or expert in cooking what have you. So it's a lot more efficient approach. But the downside is it risks hallucinating more or perhaps sending the question to the wrong part of the brain.
But that's one of the techniques that they use to really bring this model to life for a lot cheaper and according to the Deep Seek, with a lot less or a lot fewer of in vidio chips than to a lot of the Western players like open Ai have access to.
I guess one of the things that surprised people, maybe scared some people, is how deep Seak was able to make so much progress when supposedly all these Western sanctions on technology going to China didn't really seem to slow it down at all.
What's the statistic?
I think that's that's one of the big ironies actually, if incidentally, I had interviewed Jensen Huang from the CEO of Vidia in May twenty twenty three, and that was the very month, coincidentally that you know, on the other side of the world, China was, you know, getting started with deep Seek, and Jensen told me at the time, you know, one of his big concerns about export controls is it's actually just going to incentivive Chinese chip makers to work a lot harder to develop chips that can
compete within videas. But the other nearer term threat is actually software makers sort of their constraints from the sources that we've talked too close to Deep Seek are actually what really drove this company to develop workarounds do a lot more with less, and that's become a sort of founding DNA of a lot of the Chinese startups that we've talked to, a lot of these what are called Ai dragons across Beijing and Hangzhou that are doing a
lot more with a less access to compute. And I think that that's, you know, I think the other sort of component of that is they almost have this national pride that we've heard resonate throughout our interviews that really doesn't exist in the US in quite the same way. No one's you know, quite as excited or you know, we don't think of Sam Altman necessarily as a celebrity as the same way as in China. People descend on that campus, the Deep Seek office just to see pictures
of him. It seems like a national hero they are, just because he's over able to overcome these constraints, and I think that's a really interesting component of the AI scene in China, which we don't have puite in the US.
So about forty five seconds here, what's to prevent the AI companies here in the US doing the same thing that Deep Seek does.
There's really not I mean, that's one of the things that you know, they did make their models open source for the so the most part they did, you know, unveil some of their secret sauce, but they didn't quite explain all the techniques and how they did them under the hood. But that said, yes, I mean that was part of the really compelling things. By making this model open source, they really have this ethos. One of the first quotes they put with their first public announcement was
takis cheap, show me the code. And so they have this really aggressive, let me prove it to you attitude that I think in the US space, with these proprietary models, they're not willing to do. And I think the difference here is that Deep Seek having an open source approach is really putting the pressure on Google and Open Eye to do something similar or at least undercut the costs
of their models. And you can get the same completing power for thirty six dollars through deep Seek, as it would cost one thousand dollars in Open AI.
All right, Thanks to Austin Carr, Bloomberg Technology Reporter.
This is the Bloomberg Intelligence Podcast, available on Apple, Spotify, and anywhere else you get your podcasts. Listen live each weekday, ten am to noon Eastern on Bloomberg dot com, the iHeartRadio app tune In, and the Bloomberg Business app. You can also watch us live every weekday on YouTube and always on the Bloomberg terminal