Episode Transcript
[00:00:10] Speaker A: Welcome to KB on the Go and today I'm coming to you with the updates from Oracle Cloud World Sydney and I'm on the ground at the International Convention Center. Oracle Cloud World Tour will showcase innovations and explore the future of enterprise tech, from AI driven applications to autonomous databases and the next wave of of cloud infrastructure. So for now you'll be hearing from a few executives on their thoughts towards cloud computing. KBI Media is bringing all of the highlights.
Joining me now in person is Juan Luaza, Executive Vice President, Database Technologies at Oracle. And today we're discussing the role of cloud infrastructure plays in facilitating this transformation. So Juan, thanks for joining.
[00:00:57] Speaker B: Welcome, very happy to be here.
[00:00:59] Speaker A: Okay, so Juan, talk to me a little bit more about your view around how enterprises are leveraging AI to enhance operational efficiency. And then I sort of want to move to what's the cloud's position.
[00:01:11] Speaker B: AI is being used all over the place in enterprises. So yeah, operational efficiency is one thing. You can enhance the performance, you can simplify applications.
But there's a number of other areas. It enables a whole new set of capabilities that were never possible before. Like AI can understand human documents and pictures and movies, which was never really possible until really a couple of years ago. And then another really big area is for developing apps so you can ask AI to basically write the app for you. Now it's not quite that simple for an enterprise app because we have to make sure it's highly secure and the data is consistent and it's evolvable. But yeah, so that's another big area is, you know, accelerating application development.
So yeah, so AI is being used all over the place for many different purposes.
[00:02:00] Speaker A: Okay, so just going back on accelerating application development now. I was recently at Microsoft conference, they unveiled some really great technologies, one of which though was around like now we're getting to the stage where you don't really need development background if you want to leverage any sort of like from a development perspective. Right, so what does that then mean for like engineers and how do we start to accelerate that? I know people get a little bit nervous, but I'm really focused on the good, the good side of, you know, what I can bring.
[00:02:30] Speaker B: So for most enterprise applications, things, you know, things that run, you know, the world thing, you know, whether it's financial, telecom, retail, manufacturing, those things have to be right.
And so you still need technical background. Now you can develop kind of simple apps maybe without a technical background, but you know, you can't really depend on that to keep the phone system going, to keep the banking system going. We really need technical experts as well. But what it enables them to do is become far more productive. So a lot of the more mundane things that they do can be automated by AI. But you know, I can hallucinate. So that's why you need the professionals, because you need to be able to validate everything AI does. You also need to provide guardrails around what AI can generate to make sure that security is still maintained. Even though AI is generating an app, that the data consistency is maintained, that it's evolvable, that this app isn't kind of a one shot deal. You have to be able to change it and modify it in the future. There's a lot of different aspects to it, but there is a big difference between kind of smallish local apps and the apps that run the world, the enterprise apps. Those require quite a lot more expertise, even with AI.
[00:03:38] Speaker A: So you said before, far more productive by alleviating, by not doing mundane things. Right. So can you give me an example of what would you view as a mundane sort of activity?
[00:03:49] Speaker B: Well, a simple thing is writing SQL statements.
So you can now describe what you want to AI and it can generate a SQL statement for you. Then what you have to do is just, you know, read it and make sure that it actually does what you intended. Because, you know, English is not a precise language. So sometimes, you know, I can get confused about what you're actually asking for. So you have to verify it, you have to make sure there were no hallucinations or mistakes. But it greatly accelerates the process of generating because you don't have to think about, you know, in a SQL way, you can think about it more in a descriptive way. So yeah, it can greatly accelerate the productivity. That, that's one simple example.
[00:04:25] Speaker A: Okay, so focusing on the hallucinations and you're right, when you know we're pulling all of this stuff from large language models, sometimes it can come up with things that don't make sense. So I guess to your point, you do need the technical background to be able to validate, hey, is the answer correct? So if you, if you don't have a development background, for example, engineering background, and you just start leveraging AI using, you know, Python and you start coming up with stuff, but you don't actually understand the mechanics of how it works. Are you starting to see people that perhaps don't have the pedigree or the lineage to be able to decipher, hey, if this makes sense or not? Because now everyone's online or mainstream media sort of saying like hey, you don't really need a development background. You can just leverage AI and you know, run all these Python scripts, for example. So how do you sort of see that in terms of like coupling that up with like, technical capability, but also these charlatans or people perhaps that don't have the same experience as a traditional engineer?
[00:05:19] Speaker B: We don't see it as much in enterprises because data is crucial to their business.
If you're, for example, in a hospital, you can't give people incorrect diagnoses. You can't tell them that, hey, this is the drug they're supposed to take when it isn't. I mean, you just can't make that kind of mistakes. If you're in the banking industry, you can't tell someone, hey, your check was deposited when it wasn't.
So in general, enterprises are very good about making sure that everything is correct. I think more smaller businesses, things where it doesn't matter as much, you might see some of that. But again, there's the difference in these two worlds. You know, it's sort of like flying an airplane. You know, you have to make sure you're really good at it before you get behind the pilot seat and start flying an airplane because the consequences of, of messing it up are, are very severe. So I don't think we've seen that kind of problem in enterprises, but I do think you're going to start seeing things like that. And we've seen it already with, you know, people doing it for smaller purposes, for other purposes. They, you know, generate documents that might not be even accurate and they might not even realize it.
[00:06:24] Speaker A: True. I have a background in banking and I'm very familiar with Oracle, and things do go wrong, right? So to try to mitigate hallucinations and, you know, we don't want to take it over, how do we sort of mitigate things do go wrong.
So what's your view then? From an enterprise perspective?
[00:06:38] Speaker B: We are talking about something called our generative development architecture. So the idea here is for an enterprise app, I mean, the wishful thinking is you can just tell it what you want and it just builds it for you. Right? Now the problem with that is how do you guarantee that the data secure? Like I can't show medical records or business records to some other person, right? It's like your bank account, you can't see anybody else's bank account, your health records. So. So you have to guarantee that kind of stuff. You have to guarantee that the data is consistent, that it doesn't mess up the data. You have to guarantee a lot of enterprise capability. So what we're doing is we're splitting that port heart out from the AI. We're going to implement that in the database. So, for example, the security that says only you can see your account data, or only you and your spouse or, or hey, you can see your child's medical data, but you can't see, you know, some other random person's medical data and the doctor can see yours if he's, if you're their patient or, you know, the nurse, the clinician can see certain kinds of data. So all that, what we're doing is we're building all that privacy and security into the database, into the data itself.
So that way when the AI comes and generates something on top, it can't mess it up because the database will know who you are and it will only show that data to the AI. So the AI can't ask for arbitrary data. You know, I think about it as providing guardrails around the AI so it can't kind of fall off and do something crazy. So that kind of security, data consistency, all that we're pushing down into the database, so it can't be bypassed by an AI and it can't be done wrong by an AI. And that's new because generally that level of user level privilege and privacy has been built into applications, not into the database itself. But now the AI is the thing that's generating the application. So again, you can't really depend on it because it might have some hallucination, you know, it's not going to follow the rules. You give it 20 rules to follow, and usually it'll follow them, but sometimes it won't.
And that doesn't work in the enterprise.
[00:08:40] Speaker A: Okay, so this is interesting. So you said before, we're going to split it out in terms of the database, in terms of sensitive information or pii. And then you said that, you know, we want to try to avoid hallucinations and all of these things. So how are you doing that? You mentioned before guardrail, that's one component. But how do you sort of walk me through the mechanics of that then?
[00:08:59] Speaker B: The traditional language of data is something called SQL. SQL. And SQL is a very powerful language. You can do practically anything in it. And the same thing with other programming languages like Java. You can, you can write anything in the world in it. And so what we're doing, in many cases, you know, depending on the use case, we will limit how AI can use the SQL. So it can't generate arbitrary things, because once it generates arbitrary Things, then it's not guaranteed to be secure, evolvable, atomic, all that kind of stuff. So that's kind of the idea is you limit, you tell it, okay, I want you to solve this problem, but you have to do it in this fashion, and the output has to be exactly this, in this format. So then we can run that and validate that that does not violate privacy, does not mess up the data before we run it. So that's kind of the idea is you really have to kind of limit the scope of what AI can do, because by default, programming language, you can do anything you want. You can mess up everything, you can give away everything, right?
So, yeah, this has become a big deal. And the other thing is, again, it's going to have to be done at the database level because the AI is actually now the app level.
And a lot of these rules used to be enforced in the app. Well, you can't depend on AI to get it all right? And the other thing is, AI can generate programs at tremendous speed. It can generate 10,000 lines of Java in a minute. And the question is, who's going to validate it? Right? Who's going to read that? 10,000 lines of Java, it's going to take you a week, two weeks to read that. And then you're never really sure because you didn't read it yourself. And so really you have to put the guardrails around it to take a lot of the really core stuff that can't be violated and make sure it can never happen. That's how we see it going in the future. It can't just be AI doing everything.
[00:10:46] Speaker A: So just a quick question on that before we move on. So you're talking about splitting it out and sensitive information. Would you say customers or organizations out there aren't splitting out and it's all just in the one database.
[00:10:58] Speaker B: I know the data is all in the database, but what's happened is the programmer says, okay, I understand the rules of who can see medical data, who can see financial data, who can see retail data. And I'm going to program them into every program that I write. And then, you know, I'm very careful about doing these things. Right now we're trying to get AI to do that in one minute. And we tell it, here's a long set of rules to follow.
And it mostly follows the rules, but mostly follows that it's not good enough. Right. And it's hard to validate. Usually when, when you're developing program, it takes weeks, months. And so people look at it there's multiple people that look at it, it can be examined, people understand it. So that's where things get tricky when you, that productivity is actually dangerous because it can generate programs faster than humans can, can validate them.
[00:11:42] Speaker A: Same with the rules. I'm guessing that that's then stipulated by a policy, right? Is that what you're saying the rules are coming from?
[00:11:48] Speaker B: It's really by the business, right? By this specific business, a lot of times by regulations.
[00:11:52] Speaker A: This is where it gets interesting. So we're working a regulated industry and it's like people sort of think policy schmollicy. Like we had policies at our eyeballs and then no one was adhering to them while we had a lot of incidents or breaches or someone's using a tool they shouldn't be using or shadow it. So then how do you sort of get from the policy then into the rule then to make sure the data is validated?
[00:12:12] Speaker B: That's the trick, which is what we believe is we're going to have to push all that stuff into the database so it can't be by. So when you ask for data, you only get the data that you're by policy allowed to see. Right. Which gets complicated in the real world. It's not just, hey, I can see my account and nobody else's. Right, because you think of a business as an employee, your manager gets to see certain data by you, your coworker gets to see less data by you. Someone in the human resources department actually sees data, a different set of data by you. Somebody in the finance department, you know, gets access to certain data that even maybe your manager doesn't see. So there's, there's a lot of different rules that have to be obeyed. And so those rules have to be pushed down into the data so that you, you can, you cannot get, you can't mess it up. So any data that you ask for, it's only going to give you data that's, that's valid for you to see or for you to update. It won't give you anything else. So it won't be possible for me to divulge somebody else's data because the database will never show it to me.
[00:13:10] Speaker A: So then just going back on the rules then for a moment. So would Oracle or OCR more specifically have general sort of rules? I know everything depends. But I've spoken to people before and they're like, look, we don't really know. We're sort of relying off like vendors or tech partners to be able to give us some Guide Rails. We just don't know. So is that something that you guys are offering or.
[00:13:29] Speaker B: We have a bunch of rules that are kind of generic across all businesses. Your data has to be encrypted. You have to, you know, have firewalls. There's a lot of generic rules like that. But what I'm talking about here is kind of specific rules for each kind of business. Like, you know, like I said, medical care. You know, you might be able to see your underage child's medical records. Right? That's a rule. You might have an elderly parent that you've been given custody of that you can see their medical data. So there's. There's a lot of complex rules that, that are specific to a business or an industry, and those have to be provided by the customer themselves. Right. And even customer to customer, they can differ.
[00:14:08] Speaker A: Some people are still saying to me, though, Juan, like, hey, we just still don't know. Like, there's specificity in the rules. I understand that. But then people are still trying to navigate, you know, real basic patch management. And now we're talking about AI. We're talking about all these complexities that we're introducing. So how do you sort of. How do you respond to that? People may not know the specificity of the rules to be able to maintain and splitting out in terms of the sensitive data, et cetera, what you've mentioned.
[00:14:32] Speaker B: Already have to know the rules. I mean, somebody. Somebody in the corporation has to know the rules or else how's it going to. How's it going?
[00:14:38] Speaker A: I'm saying that people don't know the rules. That's what I'm saying. There are people out there that don't know that.
[00:14:42] Speaker B: That's right. And that's why you get the people that do know it and you. You wire it in so that they. So that it can't be bypassed. Right? You don't even have to know the rules. That's kind of the idea that AI doesn't have to know the rules because we bake them into the data. And so now when they ask for data, when you say, hey, tell me about my, you know, bank account, you know, did my checks go through? What kind of payments have I made? How much money did I spend on groceries? The base itself will only show the AI data. That's you or your spouses or your family or, you know, whatever the rule is, it can't see any other data, so it can't mess it up. So AI does. The point is, instead of trying to teach AI all The rules and say you have to follow these flawlessly. We're going to bake them in at a lower level. And the same thing with, with employees. I mean, you're right, employee. A lot of times employees don't really understand a lot of the rules. So if you, but if you bake them in, then you can't really bypass them anymore.
[00:15:32] Speaker A: Does that make people nervous, though?
[00:15:33] Speaker B: I think it's, it's safer. It's safer?
[00:15:36] Speaker A: How so?
[00:15:36] Speaker B: Because you can't bypass the rule. You can't see data that you're not supposed to see. It's not like going into a file cabinet or something and seeing whatever's there. It knows who you are. And it says, okay, based on your position, your title, your group, you know, what kind of role you fill. Here's the data that you're allowed to see, and that's all you get to see.
[00:15:55] Speaker A: Okay, so I want to switch gears for a moment now, and I want to talk about database 23, AI. So why would you say, one, this is game changing as an AI platform. So talk to me a little bit more like what makes this game changing? And I, and I preface that with saying that everyone's saying what they're doing is game changing, or the big hyperscalers. Every single person I'm talking to is saying it's game changing. So I really want to, with your role, your background, you know, the caliber of person that you are, I really want to understand what does this mean for you?
[00:16:24] Speaker B: So databases have been around for decades and they're really unbelievably good at handling business data.
So what is a business? You know, like, for example, you have long, you know, account information and stuff like that. It might be, you know, millions, billions of pieces of data.
It's mostly numbers and dates and strings and things like that. And we can search that, we can find that, we can secure that, we can analyze that with unbelievable precision. 100% right every time. And a lot of, a lot of ability to analyze the data as well. So that's what we've been really good at. Now what's new with AI is it handles the kind of data that computers and databases have never been good at. And that's what I call human centric data. So a table with a billion rows is not very human centric. So human centric is things like a written document, you know, written in natural language, like English, or a picture, right? So I can take a picture of you, a picture of her, and I can say, who's in this picture? Right. We've never been able very good at that in computers. It'll tell you what pixels are in there or in a document. It'll tell you what words are in there, but it won't tell you what's the concept that's being discussed, what is the summary of what's in the document. Same thing with videos. So that human centric data has been basically impossible to really deal with. You also haven't been able to use human languages to basically interact with computers and with data. So you had to, you had to learn SQL, you had to be a programmer that knew basically the computer language in order to interact with data.
So that I think is the big breakthroughs in AI is the ability to understand these concepts that were never possible.
And so what we've done at Oracle is we've baked this technology into the database. And the important part about that, there's two really important parts which is different, which is we're bringing AI to the data.
So you have this large database that might contain financial or medical records for millions, billions of people. You can't move that to the AI. You have to bring the AI to that. So we're building the algorithms directly into the database. And then the other really important part, which I think is where we differentiate a lot, is combining business data with this AI data.
So when someone asks a question, we can look up, who is this person, what kind of account do they have, what's their history of purchases, what products do they own, what are they looking at for the future, what kind of problems have they had in the past? So we can look up all this business data about that person, and then we can look up this human centric data, like documents, images, all that kind of stuff, and we can query them both together.
So you can say, hey, I want this, but it has to be specific to me. Don't tell me about a problem on an Android phone if I have an iPhone, or don't tell me about an issue with this model of car, if I have some different model car, or hey, if I never bought this product, why are you talking to me about it? So, you know, or if my account has a limit of X, don't talk to me about buying a yacht, you know, I'm not in that price range, you know.
So combining the business data, you know, if I have this kind of insurance policy is interested in answers to that kind of question that relates to my insurance policy, right? I'm not interested in, you know, just arbitrary knowledge. So combining the data that the customers, the businesses already have about a customer that personalizes it and says, well what, who are you? What's your history, what do you own, what do you want to know about? Together with that AI data makes it much more powerful and much more relevant. And seamlessly making them work together I think is the difficult part.
[00:20:03] Speaker A: Interesting that you said there is on the database because I mean you would probably know more than I would in terms of how many enterprises are running Oracle databases, like thousands. So that I think is millions. There you go. Right. So that's a unique position you've been around for so long, like almost 50 years or so. So I do understand now where that is unique as opposed to hyperscalers on that front.
[00:20:27] Speaker B: So yeah, I mean Oracle runs, you know, most of the critical systems in the world, the financial systems, the telecom systems, the health systems, you know, all the manufacturing systems. The super critical infrastructure of the world is mostly run by Oracle. So that's kind of our heritage and that, and we're bringing AI to that.
[00:20:47] Speaker A: So I want to just touch on now talking a little bit more about the rise of self driving databases. So maybe for people who aren't familiar, like talk us through like what that.
[00:20:55] Speaker B: Is specifically one of the things that we've been working on is using AI to automate all the processes involved in a database. So when I talk about mission critical databases, those have traditionally been very difficult to create and to manage. So if you, if you ran a bank or a stock exchange or an airline, you needed that. You know, the system can never go down. It has to handle millions of concurrent users, it can never mess up the data, the security has to be perfect. That has been a very complex task. A little bit like flying an airplane. You can't just walk into an airplane and fly it. You had to be a highly trained expert, you had to have a lot of knowledge, a lot of experience.
So now with AI, what we're doing is we're, we're using AI to automate that whole process.
So now the big benefit is all this. Enterprise capabilities are now very easy to use and that means that they become democratized. So now even small businesses can get the same level of mission critical capability as the largest enterprises in the world. The same level of security, the same level of availability, same level of concurrency, performance, all that stuff. That's kind of the big benefit in the data world is takes the complexity out of these very sophisticated system and everyone gets the best. You don't have to settle for something lower. Sometimes I also compare it to like a smartphone, which is the richest person in the world has the same smartphone that I have, that my kids have, that, you know, my sister has, that the person that works on my yard has. Because the technology has become simple enough and affordable enough for everyone. And that's what's happening now. AI is making the most mission critical technologies that only the biggest businesses had access to. Now everyone can get that. And because it's super simple, so you no longer have to get in and know how to fly an airplane. You can just say, hey, take me to this city and it'll just take you there.
[00:22:45] Speaker A: So now I just want to quickly touch on perhaps some emerging trends or anything sort of shaping the future for AI and sort of cloud computing. Anything you've just discussed here today at Oracle Cloud World in Sydney.
[00:22:56] Speaker B: There's so much going on in the data world. One of the things I always say, it's, it's, you know, I've been doing this for 36 years. There's never been a more interesting time. You know, a big part of it is AI because it's, as I mentioned, it's enabled things that we dreamed about in the past. Being able to speak and get answers to very sophisticated questions. Never been possible before. There's been kind of demos people showed, but they never worked. Oh, they never really worked in the real world.
So that is super exciting.
There's also the ability to make databases much more global. We're now having a lot of regulation among countries that say, hey, my data has to stay in my country. I don't want my data to go somewhere else. And so one of the challenges for us is how do we make it look like a single database so that I can manage it, I can access data, but have data stored individually in each country. So that's another big thing that, that's in the world.
Another big thing that we're doing is we're unifying a lot of the frameworks that people, a lot of the data models that people have used to store data.
So there's been relational databases for many decades that store data basically in big tables of data. But now we have things called document databases that kind of store databases. Name value pairs, hierarchical name value pairs. We have graph databases that kind of navigate data as if it was a graph from link to link. And one of the things we did in our latest release is we unified all these models so you can use any of these models to access the same data. So you no longer have to say, hey, I'm going to build a stack, a graph database or a Relational database or a document database, you can have it all in one place against the same data. So one minute you can be treating data as a document, the next minute is a graph, the next minute as relational tables. The data market had been fracturing before. We're bringing it all back together because it's the same data, so you should be able to use it any way you want to. It's the same thing whether you're speaking in English, French, Japanese, whatever. We should be able to access exactly the same. So all these languages are being unified and AI is doing that also. The multilingual aspects of AI are amazing.
You can pretty much talk to it in any language and it does what you want. Again, never possible before, Never possible. All this has happened in the last few years.
[00:25:04] Speaker A: So, Juan, do you have any sort of closing comments or any final thoughts you'd like to leave our audience with today?
[00:25:10] Speaker B: Yeah, I would say that the world of data is changing extremely rapidly. AI, there's a lot of other new technologies, so everyone needs to keep up. I mean, this is exciting and we all have to be agile. We all have to really learn about this new technologies. Sometimes people get nervous about it, but it's a tremendous opportunity and it's going to enable things that literally were not possible two years ago.
So I think it's super exciting and everyone needs to learn. And actually, one of the great things about it is this new AI technology is actually very easy. It used to be you had to learn these kind of complex machine learning algorithms, you had to go to school, become a data scientist. Now it's like using ChatGPT, you just type stuff in your normal language, human language, and it responds to you in human language. So it's also become dramatically easier. So if you just kind of understand the concepts and work on it and understand how it fits with business data and what the guardrails are, you can get very productive very quickly. And it's not like you have to go back to school for eight years to do it. So it's a very exciting time. I think the most exciting time ever to be in the industry.
[00:26:24] Speaker A: Joining me now back in person is Chris Chalaya, Senior Vice President, Technology and Customer Strategy at JPAC at Oracle. And today we're discussing an update on OCI strategy in the JPAC region. So, Chris, welcome back and keen to have you, keen to chat with me today.
[00:26:39] Speaker C: Thank you very much for having me back again.
[00:26:41] Speaker A: Okay, so I was saying before we jumped on that, you know, I sort of follow you around the world. I didn't interview in Vegas, but I did interview in Singapore. So I want to sort of go back in terms of timeline back to April last year. So maybe fill us in what sort of happened since April 2024 to now.
[00:26:57] Speaker C: We continue to innovate with OCI and around the differentiating capabilities that we've talked about in the past that we've built into the cloud. That really makes us different as an infrastructure and data provider. What's happened since we spoke? Well, we announced Dr. 25 where we've shrunk our distributed cloud, our dedicated region down to just three racks and that can be deployed inside our customers firewalls. They become the sole tenant of that cloud, but it reaps all of the investment benefits and all the innovation that we have already deployed in our public cloud. So all of the services in our public cloud are available with an entry footprint of just three racks inside a customer's environment.
The other thing that we've delivered on as well is our multi cloud strategy and we talked about this or it was on the horizon and it's now delivered. We've now got partnerships across all three of the other CSPs, Microsoft, Google, and in September last year with AWS as well, where we're actually building an entire OCI Cloud region inside our, you know, inside the other cloud providers, data centers inside of their cloud regions. What it means for customers is that they can seamlessly access the innovation that we have without having to change cloud providers. We're moving the capability, the innovation that we have into their existing choice of hyperscaler and with that we're giving them access to all of their data assets and we've then brought AI models into that as well. So third wave of this innovation is around the AI models. So Dr. -25 shrunk the footprint, running the footprint inside the other CSPs and then bringing AI models into that data. So you're now getting really I used to call the encyclopedia of the world is what you get in large language models. It's every bit of content that's out there in the world. But the encyclopedia of the world doesn't differentiate organizations. They're differentiated when they can combine that model with the model that the data that they have inside the organization and that's what we're offering customers, all three of those pieces of innovation. Since we last spoke, we've been busy.
[00:29:07] Speaker A: Okay. So what's interesting about this is as we know, and I've spoken about this publicly with a lot of your executives, that Oracle was late to the cloud game, you know this it's known fact. What's interesting that you're saying, especially with all the integration with the other cloud providers. So would you say that Oracle's plan is just to sort of just be the incumbent then not necessarily be the front player because you're integrating with all the other major hyperscalers, right?
[00:29:29] Speaker C: Not really, because just by integrating with the other cloud providers or being present inside the it doesn't preclude us from what we're doing on our public cloud regions as well. So the rate at which we are growing our cloud Presence, we're at 101 cloud regions today with 176 planned already. So the deployment and the scale is really widespread. And what we're trying to do is make sure that customers can access all of the cloud innovation without having to change what they're doing or how they're doing. So if you're already in CSPA or cspb, well, turn on the innovation that we provide you. If you don't have a CSP or we can offer you a better experience, then you bring your workloads to the Oracle cloud. Or if because of regulatory requirements you can't go into anyone's public cloud, Oracle's or the others will bring the public cloud to you. So it's really about giving customers choice, also about giving customers scale, the ability to get that innovation consistently and starting as small as possible and getting to be as large as you need to be. As you've probably heard, we're running the largest of the AI models out there are being trained and running inference on the OCI cloud because of that, not just the database, because of what we're doing with the AI infrastructure. So our GPU superclusters are running training for the largest of the LLM model providers out there as well. So we have a multifaceted solution for the smallest of the organizations to the largest world scale organizations.
[00:31:00] Speaker A: Do you think as well that I know you and I have spoken about this at length, do you think people, as in customers are moving away from Oracle, the cloud provider or OCI more than a database company?
[00:31:10] Speaker C: No, I think, I think what AI has done, it has really brought up the significance and the prominence of data into the equation. And I think in a while OCI is Oracle cloud infrastructure. So we're providing infrastructure at scale at a significant amount of performance that underpins what we've done at oci.
And then what AI has done is really brought up or raised the significance and prominence of data. So now every ounce of data. I know that's not a measure of data, but every ounce of data that you have within your organization, customers are looking to squeeze the benefit out of that. And how do you squeeze the benefit out of that? Well, you squeeze the benefit out of that by bringing your corporate data as quickly as possible with the LLMs that are, that are out there. And that'll give you company specific, government agency specific, enterprise specific AI outcomes. And that's one step away from then building out company specific, corporate specific, government agency specific AI agents. So really I think infrastructure and data goes together. It's what you may hear us refer to that as the AI data platform. It's a combination of high speed, powerful infrastructure at scale, squeezing every bit of every bit of every ounce of knowledge that a company has within their proprietary.
[00:32:34] Speaker A: So just on that, a little bit more. So would you say Oracle's advantage over perhaps other hyperscalers is because you've got millions of people, millions of customers in the database?
[00:32:42] Speaker C: Right.
[00:32:43] Speaker A: So it's a sort of a natural progression into the OCI world.
[00:32:47] Speaker C: It's beyond that, I think, you know, it's no longer just a database, it's data. Certainly Oracle runs and powers mission critical outcomes for, you know, pretty much every industry you can think about, whether that's in the hotel, travel and transport, logistics, utilities, et cetera. Oracle powers that at the database level. But what we've announced with 23AI and the AI data platform, we're actually looking well and truly beyond the data that's just inside the database. We're looking at images and videos and sound files and stuff that's in spreadsheets and unstructured documents throughout the enterprise. And it's our ability of bringing 47 years of pedigree in data management where we're able to access all of this data seamlessly. But yet respect, I use the word respect, the security, governance and privileges that have been allocated to these various data sources. So data sources, not database sources only, that is where we excel. Because now customers don't have to rewrite the whole data blueprint, right? Because it's already in place in the enterprise. We respect that, but we make it all seem as one. And that is a huge differentiator for customers.
[00:33:56] Speaker A: So I move on now. Now we spoke about a year ago about how OCI is catering more to the development community. So some of the observations that I had is, you know, younger generation, whether it's millennial, Gen Z, they're not thinking, oh, OCI first. And you have admitted that. So I want to Understand what sort of happened since then. What's Oracle, you know, doing to really cater that younger demographic? There are obviously other cloud providers that some of these younger demographics are sort of more appealing to and Oracle sort of falls back as to being an older sort of player. So I'm keen to understand what that looks like from your point of view.
[00:34:30] Speaker C: Chris. Our entry and our relevance in what the AI players are doing today has made us extremely visible to a lot of the digital natives and the emerging customers in the market.
So if you look at who's building and training these new models, well, they're relatively new companies, they're digital native companies who may or may not have had a large enterprise IT data centers in the background, right, Them bringing their workloads to Oracle. So you've heard the likes of what we're doing with Meta and what we've done with OpenAI, what we're doing with Cohere and Mosaic. Now, these are organizations that are the innovators in the market today. And what you're now seeing is a whole host behind that of digital native organizations, emerging companies that are looking at Oracle and saying, well Oracle, you're actually a true infrastructure cloud provider. Just last week I was in Singapore and we had on stage with me a four year old company, four year old company that was running on another hyperscaler, but looked at what we could deliver not just in infrastructure at a much better price, performance, footprint from what they were running at, but the innovation that we brought to them through AI services against their data. So we did a number of things. A, they came to us, B we took cost out of their existing footprint by moving their workloads from another hyperscaler to us, but we then also delivered some rapid innovation for them. So we were actually bringing AI models to their data and they delivered a service in a matter of weeks with them. The rated pace at which AI is moving in the market is seeing everybody look at where are the big players going? The big players are coming to Oracle Cloud. Oracle's very, very different. What are you different? What are you doing differently? And when we talk to how easy and portable it is for them to move their workloads from the other cloud providers, it's a no brainer because not only do we provide scale and security and choice, we also are the hyperscale cloud provider that doesn't lock you in.
So what it means is if you're running anywhere else you can, your skills are portable, your application code is portable to OCI cloud and that's seen the Rapid uptick. And you've seen our earnings and you've seen the sort of the phenomenal amount of bookings that we're bringing on board. It's customers making a decision, they're voting with their feet to come on and take the benefits. And we're the only hyperscale cloud provider that is so prevalent across the different geography locations. We've got a significant price advantage and price performance advantage. We're the only hyperscaler that provides you with performance based sla. So you know, we put our money where our mouth is effectively you don't get that elsewhere. And then we say by the way, you don't have to change anything, you don't have to retool, you don't have to reskill. You can just pick that up and point it to us.
That's a game changer.
[00:37:23] Speaker A: Okay, so there's a couple things in there which is really interesting. So let's go on the portability. I've actually been thinking about this saying if there's companies that are purely like cloud native, they can't move.
So this going back to your customer reference, how easy is it? It's a bit of complicated question for people to say that's it, I'm sick of this cloud provider. I'm going to go with oci. What does that process look like? Because you mentioned before, Oracle doesn't lock you in and that is a thing now because people don't have the trust and loyalty so they can get it faster, cheaper, SLAs sales guys better, they are going to move.
So I want to walk through this a little bit more because would you say that people are perhaps disgruntled because they are locked in?
[00:38:02] Speaker C: Well look, I'll put it towards. It's all about outcomes. I want customers, two sets of. There's customers that have got massive investments in large enterprise workloads of the past, maybe even Oracle database. Right. What we've done is we've unlocked that for them and said if you're running a large mission critical environment with the Oracle database, you can now run that in any of the other hyperscalers. Effectively we've unlocked that and you can get all these outcomes that we talked about. There's another set of customers that have got maybe no Oracle footprint as you said, cloud native customers and they have developed on the latest open source frameworks, containerized technologies. Now those technologies are made to be portable. Okay. And Oracle supports all of those open source frameworks natively on oci. What does that mean? It means unlike Some of the cloud providers, we don't fork that code and give you the Oracle specific version of that code. Right. We're supporting the native open source version of that code. So we're not locking you in. Right. And so the ability to then to redeploy what we call the development pipeline, to just point the last phase of that pipeline, you're saying where it says deploy on cloud A, or you can just say deploy an oci, no change of your code. And we've got very, very large customers and references, as you know, have moved. I mean, you've heard of customers like Uber when they moved a number of years ago. Well, Uber's, you know, it's a digital native application and they managed to pick that environment up and pick that up and move to the Oracle cloud. You look at the large LLM model trainers out there that have been moving workloads and training workloads on OCI cloud and doing that relatively quickly. It's because we support all of those frameworks out of the box natively as first class citizens in oci.
[00:39:51] Speaker A: So would you say, Chris, that we're going to see a lot more movement in the industry now? Because example what you're just saying, it's making it a lot easier for people when there's a lot less resistance and a lot less complexity and people feeling overwhelmed, they're more likely to move because they can get a better deal elsewhere.
[00:40:07] Speaker C: Would you say, think about what number portability did in the mobile phone arena. If you had a mobile phone in the past, you were tied to a particular carrier and if somebody else came up with a plan that you didn't like or had more units of messaging or data that you don't have with your current provider, you had to make the decision whether or not you change and get yourself a new phone number or you stick to the phone number. Right. So what number portability gave you was the ability for the consumer. So you had that. The operators had to keep raising the bar in what they could offer the consumers and the consumers then had the choice to move. And that's exactly what we've done. We've given customers the choice to pick to move really easily. It's no longer a religious battle. Right. You deploy in the cloud provider that gives you the best capability and what it encourages us as a cloud provider is to keep raising the bar in terms of services and outcomes and SLA and of course, price, price performance that we deliver to our customers.
[00:41:11] Speaker A: And you can still use the same.
[00:41:13] Speaker C: That's right. That's the number of portability. Exactly right. Consumers benefit and the industry as a whole then had to go and be more creative and deliver better, you know, better packages, better plans, better network and network reliability to retain customers on their platform.
[00:41:29] Speaker A: Okay, so a couple things in there which I really want to get into. So would you say the competition's going to be intense now, so whoever's gotten more plans, more stuff, more bells and whistles and steak knives and all these things that people want to get because that portability is there.
[00:41:40] Speaker C: I don't think it's just about bells and whistles because I think enterprises are looking for some things that are not sort of rocket science. Right. They're looking for price performance, they're looking for time to market, they're looking for uptime and reliability.
Right. And then I think with AI especially they're looking for trust because it's about data, it's about making sure your responses that you're getting with your AI agents are tied to the brand, the culture, you know, the data, the rules, the pervasive that your organization as a whole communicates. Right. And so they're looking for that sort of trust. Our ability to bring everything that we do in data management and expose that in what we do. I think that's going to be a differentiator. Right. So those are the, I think key capabilities that enterprises large or small are going to look for from their cloud provider. Time to market performance and uptime trust and of course price performance.
[00:42:36] Speaker A: So a couple of things as well. So people sort of porting over to OCI said it's easy. What about people leaving OCI and going elsewhere? Just as easy.
[00:42:44] Speaker C: Just as easy. Yeah, it's about, you know, and if you look at Oracle's pedigree and history, we've had a 47 year history. We were the first database provider to let customers write once to the Oracle database and deploy that across any hardware platform. So Oracle used to be ported to over 100 hardware platforms.
So it's not something that's new to us. It's actually in our DNA to give customers choice. Because if we not worried about locking customers in, what we're actually worried about is giving them a higher level of innovation so that they want to stay, they choose to stay. We've got a great track record of making sure that customers can write once and deploy anywhere. If you look at just prior to our multi cloud, the Oracle database has been available to run on other hyperscale cloud providers even before we announce our multi cloud capability.
Right. You don't hear that and see that from other database technologies from other providers that are available to run across all clouds. So it's something that's been in our DNA to give customers choice. Don't lock them in, give them choice. Innovate around them. They'll stay.
[00:43:52] Speaker A: You mentioned before price, performance SLAs. I'm curious to know, what does that mean?
[00:43:57] Speaker C: So price performance, you know, everything on the cloud you're paying for by the, you know, by time, by the hour, by the minute, by the second. Right. So if we can run a job quicker than our competitors, then you're paying, you know, we'd take a fraction of the time to run the job. You're paying a fraction of the cost relative to the competitors. So there's the price. That is what you see on the rate card, you know, for per unit of compute, per hour, minute, second. Okay, if we can then run your job at that lower price and run it even faster, then you're paying even less from that. Every cloud provider charges, you buy per unit of compute per time. Okay, so you buy 10 units of compute for an hour and you pay X dollars for that. Well, if we can run the job in 30 minutes, then you've only paying half of the X dollars. Yeah, because it's X dollars an hour, you're paying half of that X dollars. Because the way we've built the cloud and the way our networking works, we can actually run the job faster. So our unit cost is lower, our runtime is faster. So that's two levels of savings for the customers. The third level is we're the only ones who put performance based SLAs out there in front of the customer. So that's in our standard cloud contract for from way when we started, way back when we started OCI. No other hyperscale cloud provider gives you performance based SLAs.
[00:45:11] Speaker A: Okay, so let's get into this bit more performance based SLAs. Give me an example. What does that look like?
[00:45:15] Speaker C: So if we say that when you bought this set of compute, it will run this fast, at least this fast.
[00:45:22] Speaker A: What if it doesn't?
[00:45:23] Speaker C: We'll pay you back.
[00:45:24] Speaker A: Does that happen often?
[00:45:26] Speaker B: No.
[00:45:26] Speaker C: That's why we have performance based SLs, because we're confident that in our architecture is differentiated, we're confident to be able to deliver on what we're committed to. Right. So whereas if you look at the other, that's not in any other hyperscaler cloud provider today.
[00:45:41] Speaker A: And so just stay with me in that moment. Would you say that as part of that confidence on the SLAs, that's then engendering trust.
[00:45:49] Speaker C: It is. Because as an enterprise, what are you looking for? You're looking for. The word I'd use is predictability. If you are delivering a service as a business, you know that your customers are looking for predictability.
Want your cloud provider to give you.
[00:46:03] Speaker A: Predictability back to back predictability, which is predicated on the price, performance and the SLAs, et cetera.
[00:46:10] Speaker C: Predictability. That the service is going to be up when you need it and that service is going to perform at the pace at which you need it to. Businesses want that. You expect that. As a consumer. Right. I turn up, it's going to take me X minutes or X hours to do something and I want to have.
[00:46:26] Speaker A: That in that period of time in a transparency.
[00:46:28] Speaker C: Yeah, because we, we published that for our customers.
[00:46:31] Speaker A: Yeah. Okay, that's good. Because what was coming in my mind is where I think I'm just using a real basic example in terms of transparency and the, you know, you can sort of predict how much going to cost when you get an Uber, you sort of know how much going to cost. First you get in taxi or cab and then you're like, I don't know how much it's going to cost and the dude puts on a little bit more or whatever. So therefore, in terms of the consumer lens, you know what you're paying for. Be more willing.
[00:46:53] Speaker C: Correct. Or another thing would be, you know, when you make a transaction of some sort, you expect it to be resolved in 2 minutes, 3 minutes, 5 minutes. Okay. And you can back that if you know that the cloud is delivering you a certain level of performance, you could say that when you do this, we will assure that this transaction will complete in this period of time. Right. That's significant because you can now build predictability to your customers. Right. There used to be, I can't remember which food retailer used to sort of say if you don't get your pizza in so many minutes, you know, you don't pay for it. So think about that. So it's predictability. How could they do that? Because they knew they worked backwards. Right. If I'm going to get you the pizza in this period of time, I've got to have enough staff to make the pizza, I've got to have enough stores close enough you to deliver it to you in time. So you think about that, that level of predictability. And that's what we offer. And we don't do this. We can only do this Let me say, because of the way we've designed oci, it's very, and I've covered this with you before, it's very, very different, right? The way we've built atomically inside out, we're able to deliver these capabilities, we're able to deliver SLAs that the others are very, you know, that's very different from the others.
[00:48:09] Speaker A: And would you say because of that it's giving like an added level of assurance?
[00:48:13] Speaker B: Absolutely.
[00:48:13] Speaker C: And it's why customers are choosing us. Coming late with something very, very different, especially as different as that for a business outcome, is not a bad thing. It's not a bad thing.
[00:48:23] Speaker A: So I want to sort of just maybe fast forward a little bit and talk to me more around some of your predictions on the next phase of AI.
[00:48:32] Speaker C: AI is moving so quickly. I think when we started, what 2022 was two, three years ago, it was all about large language models. And large language models was really ingesting data, public data, from as many places it could get it from, right? So that's kind of where LLMs are. And there's a number of large LLMs out there.
They've got effectively pretty much all the data digested. The next phase was around how do you make it relevant or specific to a particular industry or a particular vertical?
And that's what we talked about briefly. You bring the LLM in and you fuse that within your control security environment, you fuse that with your company's proprietary data. And you saw Mike Sicily had an example today about healthcare, right? So that patient's healthcare record that we're speaking to, that's protected within that doctor's practice's environment, but we fuse the speech context into it and the medical transcribing all was done within that environment. So that's the next phase is when you take public models, bring that with private data and you're getting to certain outcomes. The next phase beyond that is I think where specific processes are going to become, you know, are going to be made and automated to become more agentic with human intervention, that's going to be more agentic. It's going to learn very specific processes, deeper and deeper. And Oracle, with our breadth of industry coverage, with the breadth of solutions and applications that we have for each of these verticals gives us a head start there because we understand processes, for example, in supply chain, because we've got a supply chain SaaS application, so we understand those business processes. You take large language models, you take company specific data and now we look at that process and say how can we optimize that process? So that I think is the next phase. You're going to get very, very specific productivity gains out of agents embedding and optimizing and augmenting effectively your workforce. Right. To make everybody more efficient with the.
[00:50:32] Speaker A: Whole agentic AI, there still needs to be some like human governance at the top. Would you envision? I mean, some people are still saying no, it is operate in the background in terms of being fully autonomous.
[00:50:43] Speaker C: And you saw the example of the Mike's video today, right? So the agent was listening in on the patient doctor conversation and at the end of it transcribing and the medical professional then actually looks at that and had full control over that as to which actions were accepted and not accepted. So I think that human in the loop is certainly going to be being there. There'll be some very maybe mundane tasks that you could probably completely automate. So things like schedule matching, right?
You've got your schedule, I've got my schedule. Find the empty spot. You don't need a human in the middle there. It's going to say you've got this block here where both of you are empty.
But I think in everything else you're going to get anything that's more serious, you're going to get human intervention like that. But it's just going to make it so much more seamless and it's going to make it so much more efficient.
And through automation you always also reduce errors.
[00:51:33] Speaker A: So I'm curious to know, what do you think there's one thing that you'd like to share with our audience that people don't know about oci.
[00:51:41] Speaker C: I can't say it in one word, so I'm going to make a couple of sentences, but the main theme here is I used the analogy of the shipping industry and how that changed with the invention of containers. And it changed everything. By moving to a standard set of containers, it broadened the way the shipping industry could move goods across not just water, but across rail and road. It changed the dynamics and the cost dynamics around how much it cost for shipping. And by lowering the cost, delivering, predictability and compartmentalization of these containers, it really meant you changed an entire industry. You change global dynamics effectively. And global supply chains were formed because of that. That's what I think we've done with OCI in the cloud. And I think if you look at, you can look at this two ways. Oracle is yet another hyperscaler. Oracle is just yet another shipper, to use the analogy. Or Oracle is the only shipper that's brought containers and giving me portability of those containers to run on rail, on ship, on port, et cetera. Right. That consistency get the same experience across that very, very different dynamic relative to the other hyperscalers. And look at us, look at us, because when you ask those questions, we're able to show you some very specific, concrete examples of how that differentiation brings outcomes to our customers.
[00:53:11] Speaker A: And Chris, just lastly, do you want to leave any sort of closing comments or final thoughts?
[00:53:16] Speaker C: My final thought would be around this. I'd always ask customers to place themselves where would your enterprise be? Or where do you want it to be? With AI adoption, start with that. Picture yourself where you want to be. I say throw the ball as far out as you can and then step back and turn around and look and say what do I need to get there? And I think OCI has that ingredients of infrastructure and data. AI equals infrastructure. I say AI equals D plus I, data plus infrastructure or infrastructure plus data. And if you throw and look at where you want your company to be, turn around and go like now what I need to get there. And I think Oracle brings the most infrastructure and data consistently, seamlessly, anywhere and everywhere you go.
[00:54:03] Speaker A: And there you have it. This is KB on the go.
Stay tuned for more.
