Bloomberg Audio Studios, podcasts, radio news. Bloomberg Tech is live from coast to coast with Caroline Hide in New York and Eva Low in Sent Francisco.
This is Bloomberg Tech coming up. Intel returns to profitability. It gives an upbeat revenue forecasts. Are the chip makers comeback efforts working.
Plus banks for preparing to launch a thirty eight billion dollar debt offering to help fund data centers tied to Oracle. It's a record for AI.
Infrastructure, and Mistraus's CEO, Arthur Mench joins us on the company's release of a new platform to help enterprise clients make customizable AI tools.
An important private market conversation. We get back to the public markets to end this week. We are at a new record high when you're looking at the SMP, when you're looking at the NASDAC, we're up two percent of the course of the last five days. But a real tick hire on the day ed the macro picture, the inflation cooler than many had worried about, and indeed that gives a bedroom to run when it comes to maybe some more cuts that the market has already been pricing in.
We've got China us. We've got plenty to be digesting, and you're looking at earnings.
Yeah, in the earnings context, Intel's post earnings rally has basically completely faded. A gain of more than seven percent is now less than one percent, but profitable for the first time in a long time on a net income basis. The PC market was better than they expected. They're making some progress in server, even a little bit in foundry.
The issue is that the quarter was all about getting money from anyone they could, clearing some of the debt, and there's still a lot of unknowns about how far there's left to go on this turnaround plan.
Correct, Yeah, let's get about where some of the holes are still left on Intel Counjin surmanis with US beming intelligence senior analysts covering semiconductors, and we are still left with a lot to chew on. Because clearly PC doing a bit better, maybe even CPUs getting into those data centers. But what about the next iteration of fabrication of chips.
This result shows the first sign in the right direction, sort of a turnaround, a beginning of a turnaround. You know, the gross margin headwinds. Like you said, the challenges are not gone away. But what they showed yesterday where they are running their demand is running ahead of the supply, which it will be for the first chapter twenty six.
There were structural gross margin improvements, though we will not see a gross margin expansion next year, primarily due to the new eighteen A coming on their exit from the ALTA business, which was their highest gross margin business.
But structurally under.
When you peel the onion, we like what they're doing. Most of the eighteen A will be moving from Oregon to Arizona, which is a better cost structure fab to begin with, a lot more volume will be coming back to Intel's foundries, which is better for them as they go through the initial transition cost and the revenue scale, which it is showing from the higher demand and PCA Data center improves slowly structurally the overall gross margins to start to keep well.
I don't know what happened, Conjum, but in the time you've been speaking, Intel's now flat and eliminated all of its gain of the day. And I'm not saying that's your fault. I'm just making the observation. Look, when I spoke to the CFO Dave's Instagram the phone, he was a bit surprised by some of the end markets that they're relying on. On the product side, servers in particular.
The basic argument they're making is the hyperscalers has suddenly realized that you have to at some point update your CPUs, you have to invest in them, and so now they're saying that actually on a year on year basis, they'll see unit growth.
What did you make of that?
You know, that is a very credible story to believe in. We have seen signs of this in other markets. For example, the storage server market has been seeing a big demand when it comes to HDD. So when you think about every time when we're pinging something like a co pilot, it's going and clearing or analyzing data which was sitting dormant for the past six months, so the storage servers were not really.
Using them cuddly.
Now we're pinging them four to five times, So that does you know, it is a credible story to believe that the need for CPO, even outside AI servers, is going to increase now as we involve more and more AI.
In our world.
Day Conjensavania Bloomberg Intelligence. Later in the show, we'll get more on Intel from the cell side. Meanwhile, banks are reportedly preparing a thirty eight billion dollar debt offering for data centers tied to Oracle. Sources say the deal could launch as soon as Monday, with JP Morgan a MUFG
among the lead banks. Here to discuss, Bloomberg Intelligence Senior tech analyst Anna rag Rana, there's a key point of clarification here that Caroline made to the team earlier in the chat, and it's not Oracle taking on this debt. Its advantage the developer who will do the project. Even so, when you see a news piece like that about that level of debt supporting the build out in infrastructure, what is your reaction to it.
No, that's a very good point, and in fact, you know, I had a very good conversation with our credit analyst, Rob Schiffman, and we discussed this quite a bit. We have seen something like this coming from other places as well, which is when you have a contract from a particular vendor, you know you can create, you could an asset back securitization product where the contract itself says there is revenue
that's going to come in the future. So the question is whether they're going to be appetite for such large dead deals going forward as well, because I think that's critical for recognizing a lot of that RPO that's coming in into revenue. Otherwise, you know it's going to be very difficult to convert that into sales.
This is about yield, and for the facilities related to advantage, they're expected to come to you in about four years is about two one year extension options. You're getting two and a half percentage points over benchmark those. So people who are clamoring to get exposure to the AI narrative now just have to do it through equity. They can do it through debt, and they can do it in
a nice yield. More broadly, are we going to see these sorts of packages come in this sort of sheer scale because of the need that we've got out there at the moment.
So you have to because I think about it, if Open Eye has signed a deal worth three hundred billion dollars with Oracle, you know, how are you going to finance that? How are you going to convert that contract into sales? And you need data centers to come there, you need equipment, and finally that needs to be realized. So we are expecting a lot more in the market. The question is whether the private debt market has the ability to, you know, basically absorb this level of.
Debt or not.
Arak Rana, I'm pretty big intelligence. We'll see how this does get absorbed. It's coming to the market as soon as Monday. But let's just talk more broadly about what's happening in the debt markets, when it comes to tech, when it's happening in the equity markets, which are actually tech help performing in that particular area. Today, you've got earnings, you've got hopes on the China US trade talks, you've
got stock ticketing on record. After that CPI number came in with well a relief that the Fed will potentially keep on cutting at the rate in the market things. Martin Norton's with US Chief Investment Strategies at EMPOWER. I want to go back to what we're just talking about and the debt that's coming to the market and the
absorption that's going to be happening. I'm looking at a story right now saying that the Bank of England is actually probing data center lending because they're weren't about AI bubbles brewing. Is this something you're starting to hear about in the market.
Martyr.
Well, it's so interesting because a lot of the rationale for the I guess complacency around the AI trade has been that this is largely being funded by very well capitalized, very strong hyperscalers, and the companies are not taking on debt. They're using what they have to extend into this space.
But now in the most recent month, we are beginning to see more debt issuance, and I think when you think about the scale of investment and the ratio of capex to revenue that we're seeing, even the hyperscalers, it does argue that we're going to have to use a broader range of financing to meet some of the targets that we have, And of course that's the signal that folks are looking for that there is an AI bubble now,
you know. I think there's a real difference between extreme valuations, which is where I think we are in an AI bubble. I think the sentiment has to get more aggressive. I think we would have to see far more debt issue. I think we've our I guess kind of wildness around it.
We're getting a few hits to the line. Mart we need more compute I'm interested in. Look, five trillion is what the spending is likely to be up to twenty thirty if we're going to get the data center and compute necessary to fuel the AI viewpoint. To that end, when you're thinking about clients coming to you saying I want exposure, I want it to be cross asset. Are you saying don't how you talking about the diversification between equity exposure and debt exposure foot bonds and the like.
Well, at this point, I mean the primary to AI still within the equity market. I think this is something that especially for the active credit managers, it is an opportunity to vet.
You know, you can get that added yield.
As suggested, but I do think people want to be really cautious in terms of adding that debt. One thing that we're talking about a lot in the credit market is just how tight spreads are, and so that is we want to watch to make sure that isn't you know, evaluation risk within the within the credit market.
Martya earnings was a factor in this market.
This morning.
When I came to my desk, we showed Intel actually is now basically flat. Even so, in the course of a conversation we had with Intel CFO Dave's insner. He made for an interesting case study, which is that he said they had been very cautious about tariffs, but the impact wasn't there in the end, and actually in the PC segment on which they depend, shipments were very strong, stronger than they had anticipated. Just as a case study, you know, that's really interesting data to me.
What do you make of it?
I mean, it's totally interesting, and it really echoes what we've seen over the course of twenty twenty five, which is this sea change in trade policy and yet really no discernible effect from either an earnings perspective or from
an inflation perspective. Our emphasis over this period has been expecting to see some hit to earnings, less concern on the inflation front, and I think one thing that we're really learning over this process is that you can see a sea change in policy and yet not necessarily see the impact immediately. I think the concern I would raise to investors is just because you don't see an immediate
impact doesn't mean there isn't an impact. When we look at that long term implications of globalization on earnings, it has been a real positive for the US and globally, and so as we roll that back, it stands to reason that it is a headwind. Maybe a headwind that we can overcome, but a headwind nevertheless.
Marta, please respond to what is one of the top news stories of the day. President Trump says he's halted trade talks with Canada because of an advertisement that they ran, citing a Reagan era comment, you're aware of it. In response, Mark Carney says that trade talks were making progress before the president's post on true Social again as a case study of trade and policy in this market.
Your reaction to it.
Well, I guess I would say that that trade talks are the toddler that just won't go to bed.
I don't think this story is really over yet.
I mean, we have a lot of certainty, I guess in terms of the overall effective terrif rate, but the fine details of what those trade talks look like, yes, with Canada, but also of course there's still a lot.
Of negotiation with China.
So I think this is something that will be this kind of lingering volatility within the market.
Marta, you have a sea of parents of young children who are just feeling that pain analogy really beautifully. Thank your just I want to go back to the very star of our conversation and they say that valuations you think are extended where exactly and what will earnings have to vindicate which types of companies.
Well, when we're looking at the US market broadly, really every sector from our vantage point, with one exception, is quite expensive. And when we talk about expensiveness, we're looking at sectors and de siles of valuation relative to their own history, and what we're seeing is valuations that are really in those ninth and tenth deciles. And that's when that starts to matter to perspective three year returns. When we look under the hood and technology, we really see
it across industries. Now, that doesn't mean that these stocks can't grind higher, and if we do see upside surprise within earnings, potentially that could push the stock market even further. I think the idea here is that there's just low immunity. So if we were to get an earning season that people expect, I think that's a lot harder to move the stock market.
I will do a shout out to healthcare.
This is something we've been surfacing since the summer This is one of those areas that is more attractively priced. It has a lot of problems that it faces, but it also has, you know, the potential to benefit from something like AI. So that's one of those areas that we're looking to in this environment.
Martin Orton, chief investment Strategist to empower, thank you very much. A Clara Resources and Exploration and Mining Services provide a plans to build a two hundred and twenty seventy seven million dollar rare Earth's plant in Louisiana, the first of its type in the US. This comes as Western nations attempt to reduce reliance on China, currently the dominant supplier.
The facility will process material from m clara's clay deposits in Brazil and Chile for magnets used in evs and wind turbines, and construction is expected to be completed by the end of twenty twenty seven.
And coming up so much for pet rocks, hey. JP Morgan takes another big step into the crypto world. More on that next. This is Bloomberg Tech.
JP Morgan is set to allow clients to use bitcoin in Ether as collateral for loans. That's according to sources the move marks a deeper step into crypto for Wall Street. Bloomberg's digital finance reporter Emily Nicole broke the story, joins us, now, I think it's really important to go through the mechanics of what we think this will work as right, institutional clients, what they can do is collateral and then what the bank will do in return take it away.
Yes, so what we've been hearing is that JP Morgan will be allowing this by the end of this year. What the process would look like is where institutional clients come to JP Morgan with holdings of bitcoin and ether that they already own that would then be transferred to a third party custodian that JP Morgan well the point, who will look after the cryptos so the bank doesn't have to touch themselves, and that can be used as collateral for financing that the bank can lend against.
What's really interesting is the context of JP Morgan and shall I say Jamie Diamond's relationship with crypto, because it was a few years ago he was calling them pet rocks, hyped up fraud and now most recently in your story, which is a beautiful quote talking about I don't think we should smoke, but I defend your right to smoke.
Is that why they're not going to be custodian while they're letting others like State Street Bank of New York, Mellon Fidelity still play that game rather than JP Morgan itself.
There's definitely a bit of a divide in terms of how banks want to touch crypto. Some like JP Morgan are happy for clients to touch it, but they don't necessarily want to be involved in touching it themselves. So earlier this summer, for example, this started out with JP Morgan allowing clients to use crypto ETFs as clateral, so traditional finance rappers that they're used to touching, but it
tracks the price of a cryptocurrency. This is a step further than that, but they're still not getting to the point of why they'd want to touch the crypto directly. It should be no though, it's pretty difficult for banks to touch cryptodirectly sometimes because there are rules like the buzz or rules on banking that prevent them from touching crypto without having to hold an equivalent amount of capital
and reserve at the same time. It can get pretty expensive unless you've got the right structures in place.
There's a lot of institutions and a lot of private individuals have a lot of crypto gains they want to be able to use as collateral. Emily Nicole, it's a great story. Thank you coming up. Google says it's reached another new milestone in quantum technology. It's a step that could bring the long promise power of quantum computing closer to reality. One of the next this has been their tech.
Google says it achieved a major milestone in quantum computing, bringing the technology's immense processing power a step closer to real world use. The company says it ran an algorithm on its Willow quantum chip that is verifiable and thirteen thousand times faster than today's best supercomputers. Let's discuss with
Koreana Chow, chief operating officer at Google Quantum AI. I think actually the best place to start is the basics of quantum computing versus accelerated or supercomputing, because if you explain why that thirteen thousand x performance is notable in simple terms, it gives some measure of why it's a piece of news this week.
Absolutely, today's computers classical computers, they use bits right zeros and once they use that to calculate, it's useful.
For a number of different problems across the world.
However, quantum computing is different uses a combination of zeros and ones at the same time that enables access to different types of problems, like the quantum echoes algorithm we announced this week.
You have brought Willow with you, You've put it in an impenetrable safe housing and casing. There it is on the desk in front of me. I read a lot of academic papers this week, and many point out that you didn't use a scalable or fault tolerant chip for this demo, and so the argument follows that it would be a big challenge for you to commercialize or scale out that technology. Is that a fair concern from the academic community.
That is the goal of quantum computing, to get to fault tolerant quantum computing. Nobody is there yet. It is a long journey, but it is very exciting. We've been excited last December to announce for the first time that error correction can work. We demonstrated that with our Willow chip, and we continue on this journey pushing the number of cubits and also bringing the errors way down.
And it's the timing now, Karina, what is it five years do we get something that actually will show quantum computing being really applicable and useful in the areas of science, in the area of medicine. Why that timeline in particular.
Yeah, we are optimistic that we'll see real world applications that are only possible on quantum computers in the next five five years. You know, this breakthrough that we announced this week is a great milestone towards that path. We showed that quantum echoes, this algorithm is not only thirteen thousand times faster on a quantum chip, but that it can be used to simulate and calculate the exact structure of a molecule. So we think this is an important step on the past.
Someone on your team, we're on the twenty twenty five Nobel Prize in physics was among the winners. You have hert and leading the charge. We have real stellar well talent, but also a lot of investment. What did you make of the news the flow that maybe the US government is there to support other smaller quantum players here in the United States? Is that important? Yeah?
We are really excited and pleased to see all of the investment, the support, the excitement about quantum computing across the board. US government has been a strong supporter and partner in this for the last many years, and we're very excited about those investments and continue partnership across the ecosystem.
Crini, the COO of the Quantum Group, right, And we reflected earlier in the week on the show when the headline hit the Bloomberg about this breakthrough. Actually markets reacted, alphabet shares moved markedly. Would you reflect a little bit on what the days that followed that were, like, did the phone ring off the hook from various parties that are interested now to know more about how they might be able to use the tech.
Yeah.
For us at Google, we are really excited about all the interests that we're seeing across the board. Our mission at Google Quantum AI is to build quantum computing for otherwise unsolvable problems. I think that's why there's increasing interest with every breakthrough. People in other fields are getting excited about what can be possible, and we look forward to continue.
So the bar has been set now for you, what is the next milestone that we should judge you by in progress.
Yeah, there's going to be continued work across the board. One really important marker of progress is continuing to push the hardware. So you can see on this Willow chip here it says one hundred and five Q. That's one hundred and five cubits on our willow too fast.
Give the camera man a chance to catch up, but he will get it.
Yeah, yes, one hundred and five cubits.
So that is great.
This has been honestly cutting edge ship. But to get to where we want to go to solve these important problems in chemistry, in physics, in material science, batteries, energy, and more, we want to get to a million cubits. So we're going to keep pushing the performance of our system. We're also going to keep pushing the algorithm in software developments so that we can solve these problems.
What does that take, Karina? What does that take in terms of supply chain? What does that take in terms of talent? What does that take in terms of focus?
Yeah, these are all really important questions. I'll start with talent. You mentioned a Nobel Prize winner on our team.
We are super proud of.
Michelle Deveray and the entire group of teammates and body of work that's happened over the last many decades.
We're super proud of our team.
We've got a super talented team of engineers, technicians, research scientists, program managers and others who are pushing the boundaries of this technology. We've got to keep bringing the best and the brightest to our team to push the performance of what is possible.
Kaarina Chow, it's been great speaking with you, chief operating officer at Google Quantum AI, on the back of what has been quite a week for your team. Congratulations. Now coming up, we get back to Intel's earnings and the chip makers efforts and come back from New York from San Francisco. This is Bloomberg Tech.
Welcome back to Bloomberg Tech if you're just joining us. Intel and its earnings were our top story, and the stock actually has eliminated most of the gain that it had. It opened up almost eight percent, is now up just half of the percentage point, but trading at its highest level since April of twenty twenty four. It has returned
to profitability. It saw surprising demand across PC and server, and it had some fighting talk about its place in the AI infrastructure industry, which largely was around wayfer and packaging. Not as exciting as GPU. But it's worth digging into numbers.
Character and is we can do that with a perfect person. Pierre Farrago is with US New Street Research, head of Global Tech Infrastructure joining us and look, we have seen the games fade. What did you make of the numbers? Was there enough there to keep the share price rallying as it has been.
Yeah, so the numbers for Intel on the day like yesterday are but relatively short tone on trend. And so the PC market has been very weak for some time. Meeting actually expectations that we are very high because everybody was very excited about the AIPC and things like that, which didn't really move the needle for the math. And so now what we see is the more traditional driver like the Windows refresh cycle helping enterprise demand and that's
great of course. And then servers, we've seen the server market been weak for quarters and quarters, if not years, because you had so much focus on deploying like AI servers that a lot of very large players were kind of you know, deprioritizing the deployment of traditional services and now we are entering a phase of catch ups. So the new term for Intel looks good and it reminds you that it's a good business, that there is operating leverage. But even if they've lost a lot of market shirts
with MD, there are still like a market leader. So that's great, but a very surprisingly positive but relatively short ter. I mean, that's the way I interpret the stock movement. You know, yeah, big grip on the news floor, and then you look back and you're like, well, but at the end of the day, word does really matter for Intel? Is that really like PC demand over the next couple of years or is that more so?
What matters. What matters is fourteen A. What matters is are they going to be the cutting edge of the next iteration of chips? Are they going to really rival ty SMC and get others than Intel's own fabrication? Going, what did you hear from Bhutan the team about that?
So I heard two things. The first one is mixed signal, which is at eighteen A, the manufacturing is ramping, is going to get into volumes immediately, you know, like for the beginning of next year. But comments but the yield, so you know, the quality of the process are still very very mixed. Let's face it, the yields are not good today, so early yields not being good is almost business as usual. But I think eighteen a's performance in
terms of manufacturing performance is disappointing. And then the second piece of news that that rematters the most for the long term of Intel is like more positive comments. But fourteen A. Remember three months ago, the CEO had very cautious comments, even mentioned in writing in the ten case as a possibility of not developing fortein a if Intel didn't find enough customers, you got to do it. And now they're talking positively about Fatina. On the technical front,
like the road map is progressing well. And also I don't even want to call that the commercial front, but more like the coalition front. Decided that if you really want Intel to be successful in manufacturing, you need more than clients. You need industrial policy, you need a coalition. You need partners that deeply engage with Intel, invest in Intel, commit to Intel, to work with Intel. That's the only
way the foundry is really going to do well. And honestly, on that front, it's very very early signals, but I think they are positive.
I tried to get a sense from Intel CFO Daves and when I spoke to him on the phone about that long term you were talking about and in an AI world AI, his answer was quite simple. They do see opportunity and accelerators, but the foundry bit is what I was like most surprise. That is saying that the foundry opportunity for them in AI is in wafers and packaging. That doesn't sound hugely exciting.
Pierre, Well, maybe it doesn't sound hugely exciting, but it sounds very realistic and really playing to Intel strength. The way it's Intel today is they have two major product franchises. One is in the Piece of Business x eighty six CPUs for pieces and one was in seven. So I'm pretty sure the management wants to maximize the value of these two franchises and concentrate on them.
And it's good.
You can't invent yourself, you know, like an AI accelerator designer, like fifteen years down the lines or ten years down the line, and that's very reasonable. And then yes, Intel has an edge packaging. Intel is a strategic asset for this industry and this is probably what management wants to build on and I see that as maybe not as exciting as willing to compete against the VIDR, but very realistic and the right industrial approach probably.
You know, and for our audience pre appreciate that answer.
Because you also cover in video right, you're able to make the comparison on xAd six. I think the argument Intel were making to us was the hyperscalers have woken up a bit to the value of having the latest CPU in infrastructure. They're willing to invest and refresh on CPU. So again for them Intel, they would argue that it's a great opportunity for them.
Do you see that opportunity.
Yes, absolutely, no doubt, and it's very very strong. You have like clients asking for it. The specifics of the XAT six architecture and the way Intel implements it is
very differentiated. I've always been a great believer in the quality of the products of Intel on that from like the ability of an intell extaden ship to deal with very unforeseen city cuation in very complex work cloud is excellent, and you've had so much demand for it that it actually triggers NVIDA has moved first to open up envilling like the high bandwidth connectivity with their GPS so that Excaity six chip can be integrated into na systems, and
a five billion dollar investment in Intel and a partnership with Intel to co develop chip that will be optimized to get integrated into Nvidia systems. And my read, having followed that for a few years, is that it's not just like a nice gesture from Nvidia under political pressure or anything like that. It is a genuine, thorough and
deep interest for Intel products in the industry for that role. Now, keep things to the scale they are out on NVL seventy to server powered by x eighty six Champs chips. The chip is going to be single digit percentage of the co cost of the overall server and the other ninety plus percent would be in video.
So Pierre within video support, we know the money is yet to come through in that settle in the next quarter. Will that be enough to see Ohio breakground? Will we see an actual commitment not just to co develop, but actually to eventually fabricate on site in the United States, not just relying on TSMC.
It's a great question. You need more than that to have like a final commitment or to claim victory basically in order for Intel to successfully manufacture chips in the US package chips in the US for US clients. You need the money and major pogrests made on that front with money from the government, from NVDR, from South Bank, and I'm sure.
More will furrow.
You need also like the multi year commitment of a coalition, so you're not going to get like the right hip in the first go with. It's a dialogue that needs to last very multiple years before you get to success. And last but not least, you need execution. You needed to come up with a fought in a that's going to show up like policively, be unveiled and better convinced and make potential adopters more confident. It's going to be the right not to do something with it.
Pierre Faragu of New Street Research. Great to have you back on Bloomberg Tech. Thank you very much.
Now coming up MISTRAUS CEO Arthur Mentch joins us to discuss the startups new platform designed to help companies get the most out of AI. It's a big interview. We've been looking forward to this one for some time. Stay with us. This is Bloomberg Tech.
Today. European AI startup Mistra.
Announced the release of a new platform to help its enterprise clients make customizable AI tools that are easy to operate throughout the businesses. Delighted to welcome mist our CEO after Mensch to the program. This is a path to production right and the enterprise category between you anthropic open AI, it's a fierce battle. I think let's just start by you explaining why you feel this is such a significant milestone for Mistra.
So hello and happy to be here for us to We're a global company really focused on creating value for the enterprises, and we have had three years of experience in doing that, and so AI Studio is basically the one stop shop platform where you can build your AA application as a business, and so it brings everything that we need to create value in the enterprise, to create applications that belongs to the enterprises, that contains their IP,
that is connected to their data, and that improves over time, averaging the knowledge that is contained within enterphrises, that is contained within employee's mind.
You are a global company, you're producing foundational models. Author I wonder though, how you compete against those that are already serving the enterprise. How is your solution different from that of anthropics, that of open AI, is that of X's and many others.
Well, I think our solution is much more integrated. Is that we have thought about all of the bricks that needed to be there to go from a model, which is how we started to an application with the right font end, with the right back ends, and with the right learning mechanisms. So that integration is really the reason why we've been succeeding in delivering value with enterprises. So
that's one very strong area of differentiation. The other thing is that, in contrast with some of our competitors, we really believe that enterprises should build their own AI, that they should own the system, that the IP should remain their own, that the data should remain where it is
and should not necessarily flow back to us. And so that approach of building a portable platform that can be the deployed on prend that can be deployed on private cloud, combined with the vertically integrated approach of having just one platform allowing you to go from prototype production, I think is very very different from the approach for competitors, and that has an oders to make a lot of progress in particular under.
Us Arthur, can you talk about how quickly you're able to onboard scaled enterprise customers to this and in turn therefore what the capacity and compute constraints are for you in launching a new product and then trying to run workloads with clients.
Well, a good example that I can give you is what we've been doing with one of our big customers, which is a logistic company call same Messagem and one of the biggest shipping company in the world, and so as many other companies, they were a bit struggling with the adoption of air. If you look at the MIT reports, they state that ninety five percent of use cases just don't go intoduction. And the way we have done the work with them is we use AI studio and we deployed our AI applied engineers or.
AED applied scientists.
We made them work with the business unit owners and they realized what kind of processes could be automated and to end and from ideating what we could automate to going into production. So from March to July in formats we were able to understand a full function to be automated, so cargo release and dispatching of release of containers to multiple customers. We're able to go from that to production
that is now going at the global scale. So the way we think about it that it takes a quarter to go from a prototype to production, and for him to happen, you need to have the right tools. You need to have the right tools. You need to have the right experts. You need to deeply understand what AI agents can do and cannot do. You need to connect
the agents to the data sources. You need to connect them to software which is sometimes legacy, so you need to have the right interfaces, the right plumbing.
That's what AIP studio can bring.
You also need to have the experts, so that's why we deploy people in the companies that we work with.
After there is intense interest in Mistrau as the European champion among AI labs frontier labs. The thing we're most interesting to learn about is the progress of the follow on fundraising round that's reported. You're undertaking how much money you seek in the raise, But beyond the money, why do you need to do a quick follow on? What is the most pressing need for capital at the moment.
So I'm not sure we'll talk to you about the follow on but we are not raising money at the moment.
We are.
We've just actually announced the fundraise with SML, which is the biggest European company and with whom we work with whom we have a commercial agreement where we help them make better products, automate the processes, we help connect their vertical expertise to horizontal expertise, We make them create their own AI and deploy this AI into the their systems.
So that was a one point seven billion euro race, which which makes us We've raised around three billion euro in total, and that in turn has enabled us to grow very fast and to create unique technology and to release it in particular on the.
Open source WED.
That was really interesting, this strategic investment coming from SML. You just referenced CMA, CGM, another French based giant. It's a global company, but it's French based. You've got work with stillants, You've got work with BNP, Parabar. What about US companies? How are you managing to penetrate here?
So we are working with CHIPELU US companies.
We're working with Cisco for instance, which has chosen us because we allowed them to deploy on private cloud and we deploy people to work with them in particular on the customer experience side, so that's one important customer of hours. We also work with Snowflake deployer technology in particular on document processing and all of our technology to structure and structure knowledge into to take instruct your knowledge and turn it into things that can then be crawled by a systems.
We also work, as you know, with Microsoft, which is which is a very important bartner of owers and exposes on the on voundry on models, being very competitive in particular on their cost of performance ratio. We walk with A Doubles and GCP and we work with multiple digital natives Companion as well.
Arthur your peers, Dario ama Day from Anthropic, Sam Altman OpenAI are looking increasingly to the Middle East, both for capital but compute capacity. Right this week we broke the news of Anthropic securing one million TPUs with Google and GCP. That's the scale that people are talking about. Are you making those future plans for scale? Are you looking at debt markets to ensure the future of mistrau and what you want to do?
So the future of if you should diligence is very linked to infrastructure, as you know, and as a consequence of that, we have created a new business unit called Mistile Compute, which is creating digital infrastructure GPUs, in particular in Europe, which is lacking infrastructure today, and so we've been growing that very fast and we are acquiring multiple hundreds of Mega what capacity in the coming year to actually be able to serve our customers, to serve AI startups,
to serve all of our existing customers and customers that come to us because they are looking for sovereign capacity. We also use that capacity to train our own models and to maintain our leadership on the open source font.
With that comes the need to continue to invest. You raised money, and is that what's being used to finance the exploration and data center build out. Because I'm looking at a headline at the moment that the Bank of England itself is really worried about a circled AI bubble and some of the debt and the financing that's going on data centers at the moment. Is that something that gives you pause?
Author, Well, we've raised equity and we are deploying that equity to actually train models.
So RANT is financed by our equity.
We also have long term contacts for our compute facilities, and with those long term contacts, we are able to actually finance them through that that's not this is bank debt and gives us confidence that we are not overly exposed to that there is a lot of investment happening on the infrastructure side today. We're really focused on creating the long term value that will justify with investments. Because we operate with enterprises, because we go all the way to delivering value for.
Them, to identifying their use cases, to.
Making things that transforming what looks like magic to something that looks like money, we are not exposed to whatever may happen on overly investing on the infrastructure site.
We invest on the infastructure site in a wise way to help.
Build mister Lai studio. Mench. Great to have you on that news today. Thank you so much, the CEO of Mistrell and co founder. Now coming up, we'll come back on the AWS shortage and outage. But it is one of the biggest in history. Remember and what would that mean for cloud Giant's future in the age of AI. This has really big ten.
Amazon basically invented the cloud business, but now it's struggling as Monday's outage shows one of the worst outages in the cloud unit's history. Bloomberg's Amazon reporter Matt Day joins us to discuss his latest deep dive and why AWS is now perceived as trailing its rivals and AI. This week's been rough outage, losing some Groundman to GPUs TPUs for Google.
Just give me the headline of your story.
So the headline of the story is that Amazon is essentially no longer the only game in town in cloud computing. They've got real credible rivals, you know, down to Oracle and Google. Now, five years or so ago was on Microsoft that was knocking on the door. And now you add on all this AI workloads and that's not a business at AIS or that Amazon rather seem to be leading in.
Just for context, though, how successful does it remain, how much market shared does it still own? And what are we starting to see in terms of inroads and so.
On on traditional cloud It's something like thirty eight percent, reckons Gartner. It's down from about half of the market, you know, five or six years ago.
Folks who have made inroads.
You know, there's there's Microsoft, there's GCP, there's Oracle and a lot of a lot of the threat there is just some of the services at Amazon pioneered like those are leaning toward commodities. Other companies can offer you a similar package of goods.
Right, Matt, When we break the story about Anthropic using TPUs, Google stock up, Amazon stock down, how have you reflected on that in the story.
Also, it's you really can't overstate the importance of Anthropic to AWS. You know, they're their marquee artificial intelligence customer. They're helping kind of code develop Amazon's artificial intelligence ship called Tranium with them. So the fact that they now have, you know, what really looks like edge or an option to go more toward Google. Should they choose, should they like the tech results better? That is a real risk or a potential risk rather for Amazon's business.
I have a feeling it's a story you're going to continue to dig into. It's so well read today. Thank you so much. Bloomberg's Mack Day reflecting on the week that was for Amazon, and we reflect on the show that was because it does it for this edition of Bloomberg Tech Ed. It was another huge busy one and boy next week to lon end.
Yeah, like we didn't even get to mention really in the course of today, but like next week is the tech super Bowl, right, Like in an earnings context, it's the biggest one.
It's going to matter.
Well, is a video there? I guess not. We've got everyone else, Meta, Microsoft, Amazon, you name it, the Vindication, the Fundamentals. They've got a lot to check out on the podcast.
I am heading to DC for GtC within video so that there might be some news there. Check out the pod. You know where to find it. This is Bloomberg Tech