This is the second of three articles that recount a recent conversation with Tom Siebel, Chairman and CEO of C3 IoT. Siebel is a veteran at creating companies that make and sell enterprise software, including Siebel Systems, an application software leader that merged with Oracle in 2006. In the first article, we covered the history and present state of software and applications. In this article, we look forward at what Siebel believes the Internet of Things (IoT) will mean for technology as a whole, as well as for how society functions.
Dan Woods: So the game you’re playing now is fundamentally different. The game in the early days was a game of engineering, where you were breaking new ground in fundamental engineering practices. You were putting together components at Siebel in new ways and you probably had to make up the difference where there were gaps. So, what’s the game now in a modern company? It seems to me that even smaller companies don’t have single products anymore for very long. They very quickly, within six months or a year, have platforms and then integrations between those platforms. And then you have to create an ecosystem of different players who are going to connect to that platform and use it, and each of the players usually has a different motivation. How do you deal with that? What skills are important now in defining successful enterprise products and which companies are really good at it?
Tom Siebel: I think the agile programming methodology is important. Very frequently, agile is used in the absence of rigor in software engineering, and some of it can get pretty sloppy. But this idea that we do things incrementally in small bites, and if you do it with rigor and substance and discipline, it’s a whole new means of getting things done and getting things done quickly. So that’s important. That’s exciting.
Another skill that’s absolutely critical to everything that’s going on is data science. Data science is the real deal, and it is all about AI and machine learning. Machine learning is a field where the cell has divided maybe four times and I think what happens there in the next 20 years is going to be exciting in many ways. It’s going to be socially beneficial. It could be environmentally beneficial. It’s going to be economically beneficial.
Woods: But if I was the best agile development shop and I was really good at data science, that’s still different from defining a product with good market fit. When you were developing C3 IoT, it’s not just agile, it’s not just insights, it’s not just the components that you’re bringing to bear and pointing at something. It’s what you’re pointing them at and why you think it’s valuable. It’s the product vision, right?
Siebel: Yes. When we got involved in the relational database business, we believed, well before the market developed, that people would be applying set theory, which is what a relational database is all about. We were applying set theory to data storage and retrieval. And the market did develop. And as of 1993, we had applied information technology to manufacturing and office automation, to accounting – yet sales, marketing, and customer service remained largely untouched. And I believed that was unlikely to continue to be true. Now, that might not be that great of an insight, but for some reason, nobody else believed that. Now, I believe that what’s going on now with big data, predictive analytics and IoT looks like a complete replacement market for everything that’s going on in enterprise application software.
Woods: So you don’t think that the IoT opportunity is just industrial manufacturing?
Siebel: Absolutely not. It's consumer products, it’s entertainment, it’s travel, it’s the works. It covers everything.
Woods: You’re taking a radical point of view and saying everything is a supply chain?
Siebel: Yes. There is no device in ten years that is not sensored. The phone, the watch, your heart, your pulse, the refrigerator, the pool pump, the thermostat. Everything that touches everybody’s life in the civilized world will have a sensor that’s putting out a signal once an hour, maybe 60 times a second depending on the need, but in ten years everything is sensored. At the beginning of this century, we had half a billion sensors deployed. Today there are 19 billion. In five years, there will be 50 billion. It's growing exponentially. Everything about everybody’s life, the way that they work, entertain themselves, transport themselves, and communicate and will be sensored. The human being is highly sensored. Whether it’s devices that are wearable, devices that are embedded, be they embedded in the cerebrum, in the aorta or in the gut, those devices will be there.
Woods: Got it. And you’re creating an infrastructure that addresses that dramatic growth in this use of sensors in a supply chain and the need to make use of that data and do predictive applications?
Siebel: We combine data from the enterprise with data from the extranet into a unified federated image to be able to process these very, very large datasets. They grow at breathtakingly rapid rates.
Woods: It seems like C3 IoT will thrive or not depending on how well it defines the use cases of its applications.
Siebel: No. That’s not quite true. We really have built a platform.
Woods: But do people buy platforms?
Siebel: Yes, they do. Absolutely.
Woods: So how do you define platforms?
Siebel: The way that we use the term platform is that we tend to operate on top of cloud infrastructure—this is basically compute and storage in the cloud. We refer to that as infrastructure as a service. On top of that, we provide a platform as a service that serves two purposes. It provides a set of application development tools that allows people, whether they be in the aerospace industry or the healthcare industry or the telecommunications industry, to rapidly design, develop, deploy, provision and operate big data, predictive analytics, and IoT applications. The second aspect of that platform is – not only is it a development environment – it’s an operating environment for the execution of those applications. So that’s platform as a service – we use the platform as a service to both develop software as a service applications that might be predictive analytics for aerospace, customer churn for telecommunications, the next generation of CRM, whatever it may be, but the platform also is the operating system for that application.
Woods: Right – it’s deployable immediately. You don’t have to repackage it up. As soon as you're done developing, that development is deployable?
Siebel: And it provides the necessary services like access control, encryption, queuing, encryption in motion, encryption at rest, ETL, MapReduce, and data science services. All of those services are there for the operation of that software as a service application. So those are the three layers, infrastructure as a service, platform as a service, and software as a service.
Woods: And software as a service is the result.
Siebel: Yes, those are the applications themselves. In the future, I don’t think we’re in the business of building applications. We’re in the business only of providing a platform for others to build applications.
Woods: Right. But usually, in a sales process, somebody puts down money because they have a use case with a return. They think, maybe we can do ten of these and the tenth will be a lot easier than the first, but what they’re buying the first time is a solution to a specific use case. And if it’s supported by a platform, maybe that makes them happier because they realize they can do more with it. I have not seen people say, “Let’s buy the platform and figure out how it’s going to be useful.”
Siebel: It’s a little bit of a hybrid in this case. This is a very rapidly growing market. We’re running into companies where the vision is there, it’s all about IoT, big data, and predictive analytics. They have a hundred or a thousand use cases. And the idea is let’s pick one that is sufficiently complex and representative of the use cases that we have, and at the same time sufficiently tractable that we can knock it out in a short period of time and demonstrate that it works. So they’re not really buying something to solve their first use case. They’re demonstrating that by solving the first use case, they can be satisfied that they can solve a large class of use cases. We tend to go from the pilot to the enterprise transaction.
Woods: What kinds of companies are coming to you now with a platform vision, not just a use case vision?
Siebel: Well, in the utilities space, there's a company called Enel based in Rome. They’re a pretty big business, a 71 billion euro business. They have 60 million meters in 40 countries. Putting that into perspective, there are approximately 120 million meters in the United States, so this is bigger than a breadbox. Now, in Italy and Spain, they have 42 million smart meters out of 44 million meters. So this is basically a smart grid, a big IoT infrastructure. And with Enel, we’ve aggregated 10 trillion rows of data into a petabyte size data image that grows at 300 gigabytes a day. We take the union of the data from 15 enterprise information systems and a 44 million sensor network as well as weather and terrain data. We aggregate those data on a unified federated image, we process those transactions as data arrives in the elastic cloud at rates that exceed a million transactions per second, and then we apply AI and machine learning to these transactions to do things like predictive maintenance for 1 million kilometers of the medium voltage distribution network, predictive maintenance for conventional generation, and predictions related to customer churn. The potential economic benefit is 261 million euros a year – recurring economic benefit. That is the largest IoT application on earth. But we’re also working with a healthcare company now where they need to aggregate patient records for a population of 100 million – about a third of the United States’ population – including radiology, hematology, pharmacology, and patient history data. The company wants to aggregate these data in a unified federated image and then apply machine learning for disease prediction. Now, if you use machine learning and data science, we can predict with very high levels of precision, say 80-90% precision, who’s going to come down with diabetes in, say, seven years. They used to do that with rules-based systems. Those were very blunt instruments. And they required maintenance. Now we do it with machine learning. It’s very precise. So what’s the benefit? The benefit is they can deal with these people clinically now in the next seven years rather than deal with them in the operating room in seven years.
Woods: Coming back to platforms, what you’re saying is that, if you’re doing an enterprise software product, to be successful, you have to create a platform that takes advantage of all these layers. You have the infrastructure as a service layer, the platform as a service layer that enables the software as a service layer, and that’s the way you do it. But for most significant enterprise software, they’re also going to be creating platforms aimed at solving dozens of use cases because of the power of the infrastructure in the modern world.
Siebel: I think so. My prediction is that this software segment in, say, 2023 is a quarter of a trillion dollar software market. That’s larger than all of information technology when I went to work for Oracle in 1983. That’s a big market. It is a game changer.
