He adds: “In many cases, there’s data estates [that] look like a large, dusty closet with things shoved in a bunch of different drawers. You can’t take that and then have a clean, precise analysis based on data that you can trust.”
Improving the quality of the data is also something companies can control, unlike other factors, such as the demand for graphics processing units (GPUs) outpacing supply, which is unlikely to change soon.
To illustrate how better data unlocks real results, Sansbury cites the example of a pharmaceutical company that has used Cloudera to unify and make sense of decades of research data. “The system [surfaced] connections that otherwise would not have been available without bringing the data together at an enterprise scale and applying AI-based search tools.” This sped up the drug discovery process, saving the company about 18 months in drug discovery time and hundreds of millions of dollars a year in research and development costs.
Another example is a financial institution that used AI to improve its customer support processes. “When you’re on a call with this bank, AI is listening and predicting what you’re going to say next. So if your cell phone breaks up and gets garbled, the AI predicts what you’re going to say and picks back up. That capability helps them resolve issues 20% to 25% faster. When you apply it to thousands of call centre employees, this is significant cost savings, but I’d also say it’s a very novel application of technology to solve what seems like a relatively mundane business problem.”
See also: Should bond markets fear an AI bubble?
Multi-cloud providers are gaining importance
As the use of AI becomes more prevalent, using multi-cloud providers is also going to be “increasingly important,” says Sansbury. “[It’s] for the simple purpose of economics. Large customers want to have a second source of supplier to make sure that everyone at the table is behaving fairly from an economic perspective,” he adds.
The irony is, most enterprises did not set out to be single-cloud shops. In his view, the rush to the cloud was driven by “speed and convenience”, which led to an outcome that saw most companies having a preferred cloud service provider and a second provider that is “very small in nature”. When that happens, companies are likely to lose economic decision-making power.
See also: AI’s next act: From shiny pilots to measurable impact
Amid the “super crowded” AI and data platform market, Cloudera offers its customers the best of both worlds by providing convenience and ease as well as scalability and capacity, says Sansbury, adding that customers were forced to choose between either in the past.
Cloudera’s platform counts its scale, security and manageability of its tech as factors that enable it to stand out among its competitors. “We founded the big data market 15 years ago, and in that time, we were the undisputed leader in technology across on-promises and private cloud environments.”
He adds: “I think our key advantage is our ability to offer a compelling return on investment or total cost of ownership, an enterprise-class, modern data platform that addresses large customers’ needs for security and scalability. But with some of the development we’ve been doing in the last two or three years, and with our acquisitions, we’re also starting to provide them more of the convenience and ease of use that they want.”
As a self-described neutral enabler of AI and the “Switzerland of data infrastructure”, Sansbury shares that Cloudera tries to be “agnostic” about where its clients want to use the data by allowing workload portability to support hybrid and multi-cloud.
He adds that Cloudera’s role is to simplify what their customers are looking for, since they usually have some of everything, including mainframes, data centres and cloud-based infrastructures. “They have what we call accidental architectures. Where they are today is a function of 20 or 30 years of decisions, acquisitions… It’s not an architecture that’s put together on a clean white sheet of paper to be perfect. Because of that, they’re looking for vendors like us to simplify and not complicate their lives.”
When it comes to governance and compliance, Sansbury notes that there are different use cases for cloud-based infrastructure and owned infrastructure, where businesses prefer convenience for the former and data security with the latter. Cloudera’s customers can find both on its platform.
Cloud-based infrastructure isn’t always cheaper
To stay ahead of the latest tech trends, click here for DigitalEdge Section
Between cloud-based and owned/on-premises infrastructure, the former isn’t always cheaper. While cloud-based providers allow enterprises to use rented infrastructure with the promise of convenience and ease of use, Sansbury says most enterprises don’t use it like an occasional “rental car”, which means costs add up at scale.
He adds that owned infrastructure can cost half as much as cloud at a certain point, citing Cloudera’s research showing that after roughly 200 workloads, running on owned infrastructure can actually become more economical.
“The argument was that cloud-based infrastructure was going to be cheaper… but that hasn’t happened. The two big cost components of that are computing devices and electricity, and neither of those costs are going down in the next 10 years. They’re both getting more expensive,” he says. “So I think that’s brought realisation to customers that cloud capabilities are critical to their long-term strategy, but not sufficient to do everything they need to do.”
Yet, Sansbury isn’t writing off the use of cloud-based infrastructure. “If it’s for a test application that’s being evaluated or if it’s something that has significantly variable compute needs, like it may only need to compute for an hour a day, those should be on cloud-based infrastructure. But if something is running all the time, like fraud detection and prevention at a financial services organisation or a credit card company, or if something requires significant security and/or proprietary data sources, those are the types of workloads that should be running on owned infrastructure.”
Abhas Ricky, chief strategy officer at Cloudera, adds that cloud economics are also “way more expensive” than hybrid economic. That aligns with what he is seeing in the Asia Pacific region, where hybrid cloud (or a mix of cloud and owned infrastructure) is more prevalent as organisations are still in the early stage of digital transformation or bound by data sovereignty requirements.
In Ricky’s view, the value behind any technology investment comes from one or two familiar levers: cost reduction, cost avoidance, risk reduction, time to market and service improvement.
However, how organisations measure those value drivers may differ. The metrics can range from the total cost of ownership, drug discovery and research acceleration, or a reduction in false positives in anti-money laundering cases.
