The world is awash in a sea of data, and it’s time for new technology to step up to the challenge and deal with it.

That’s the fundamental proposition behind Fungible Inc., a startup company led by Pradeep Sindhu (pictured), who founded Juniper Networks Inc. in 1996. Sindhu has made a career out of dealing with hard, complex problems. This may well be his greatest challenge yet.

If data is the new “oil,” then data centers are the oil rigs, pumping out and processing information at a rate previously unimaginable 10 years ago. Yet, unlike an oil well, which inevitably runs dry, data pipelines just keep expanding as connected technologies, such as autonomous cars and internet of things devices, continue to multiply.

The result is that compute has been outpaced by communications and new engines need to be built for data centers to handle the increased demand for processing and storage. This will require a whole different look at silicon, the processing engine that drives the modern computing industry.

“So many people have worked on the problem of building general purpose processors, faster and faster, better and better,” Sindhu said. “There’s not a lot left in that tank; the architecture has now played out. It’s the pursuit of what’s next is what led me to start my second company.”

Sindhu spoke with John Furrier, host of theCUBE, SiliconANGLE Media’s mobile livestreaming studio, during the Mayfield People First event in Menlo Park, California. They discussed the impact of Moore’s Law on processor development, Fungible’s custom silicon solution, a critical focus on hardware/software integration, the competitive landscape for investment and Silicon Valley’s spirit of entrepreneurship. (* Disclosure below.)

Here’s the first of the two-part interview with Pradeep Sindhu:

Moore’s Law slows down

One motivation for taking a fresh, new look at silicon is that Moore’s Law is running out of gas. Conceived by Intel Corp. co-founder Gordon Moore, the law suggested that processor performance would double every two years.

Moore was right, and processor performance doubled with regularity. Yet, a number of technologists, from the chief executive officer of Nvidia Corp. to scientists at MIT, have proclaimed the end of the Moore’s Law trend. This is the opening for Sindhu and his new company.

“As long as silicon was improving in performance every couple of years, there’s no incentive to come up with new architectures,” Sindhu said. “Now, it’s maybe 40 to 50 percent improvement every two to three years. The world has to come up with some other way to provide improved performance.”

Data-centric computing

Fungible’s solution is grounded in custom silicon optimized for data movement and data storage. At the heart of what the startup company’s engineers are building is a data processing unit, or DPU, which can transfer data quickly and connect application engines in data centers.

Fungible is seeking to address computational problems that surface when big data meets scale-out architectures and artificial intelligence. The solution is customized and programmable, made necessary by the slowdown of Moore’s Law and the potential limits of x86 processor architecture.

“Our silicon strategy is really to focus on one aspect of computations in the data center, and this aspect we call data-centric computing,” Sindhu explained. “Data-centric computing is really computing where there is a lot more movement of data and a lot less arithmetic on data. We’re building a programmable DPU to do those computations much better than any engine can today.”

A key factor that drives Fungible’s approach is a realization that storage and compute are increasingly being done at the edge, fueled by the rise of massively scalable data centers that rely on tools like Kubernetes to automate container clusters. Self-driving infrastructures will require new platforms that are agile and can deploy data services rapidly at scale.

Kubernetes and cloud automation

Sindhu got a glimpse of what this could mean before he stepped aside as Juniper’s CTO in early 2017 to start Fungible. During the previous year, Juniper made a series of acquisitions that were focused on data-driven cloud operations and the potential for more agile silicon.

These acquisitions included AppFormix Inc., a cloud automation startup, whose technology worked with any OpenStack or Kubernetes distribution. Juniper also bought Aurrion Inc., a silicon photonics platform that offered lower bit-per-second costs for networked systems.

“What we are entering is a phase that I would call heterogeneous specialized scale-out engines,” Sindhu said. “In addition to these massively scalable data centers, we’re going to have tens of thousands of smaller data centers closer to where the data is born.”

In a prior interview, Sindhu made it clear that Fungible will not be simply a chip company. The processor will be programmable, so hardware/software integration is a key element of the company’s business plan.

Fungible’s other co-founder is Bertrand Serlet, who started at Apple Inc. in 1997 and was instrumental in the development of Mac OS X. Apple provides a useful model for Fungible’s approach.

Nearly half of Fungible’s team is versed in software, according to Sindhu, who points to Apple’s ability to blend this expertise with silicon as a key element in its long track record of success. “Apple is a great example for me of a company that understands the value of software/hardware integration,” Sindhu said. “Everybody thinks of Apple as a software-only company. They have thousands of silicon engineers.”

Changing Silicon Valley landscape

Despite the presence of Serlet and Sindhu as Fungible’s leaders, obtaining funding from the venture capital community was no easy task. Juniper’s founder and Apple’s key technologist were politely shown the door by several investment firms, according to Sindhu. The company finally launched in February of last year with $32 million in financing, co-led by Mayfield Fund LLC, Walden Riverwood Ventures and Battery Ventures with participation by Juniper Networks.

For Sindhu, it was a reminder that the technology industry’s preoccupation with social media had changed the landscape compared to when he sought funding for Juniper and his concept of networking at scale back in the 1990s. “I saw a reluctance on the part of venture to take big bets,” Sindhu said. “The number of VCs that were willing to look at those kinds of problems were far fewer this time around than last time.”

Sindhu and his team are not the only ones looking at custom silicon to provide breakthrough technology for dealing with the data tsunami. This week, during its major re:Invent conference in Las Vegas, Amazon Web Services Inc. introduced its own custom-designed Arm processor called Graviton. AWS executives said the new chip was orchestrated to handle scale-out workloads.

Earlier in November, contract chip maker GlobalFoundrie U.A. Inc. formed Avera Semi LLC to produce custom ASICs. Even the carmaker Tesla Inc. has recently divulged that it was working on its own artificial intelligence chips for autonomous technologies.

“They key is the operational model and the idea of using scale-out design principles,” Sindhu said. “People who are competing against scale-out by building large-scale monolithic machines, I think they are going to have difficulty.”

Entrepreneurship drives big ideas

Despite the challenge of building new technology to solve tough, complex problems in a competitive environment, Sindhu remains enthusiastic about the life of an entrepreneur. He is no stranger to pursuing big-picture ideas.

After bringing Juniper to its major position as a legitimate competitor with Cisco Systems Inc. in the router space, Sindhu launched the Infranet Initiative in 2003. The aim was to corral telecommunications service providers and standards organizations into a common framework for network accessibility and connectivity between carriers. It later became part of the TeleManagement Forum.

In 2012, Sindhu presided over a Juniper-led project called QFabric. It was an attempt to redesign and simplify data center fabric, and portions of the technology remain in Juniper’s switch portfolio today.

The culture of Silicon Valley can be very tolerant of failure, but there is also an expectation of success. Sindhu is well aware of the game he is in. While his latest venture offers an intriguing glimpse into the future of data-driven architecture, Juniper’s founder knows that he will be judged on whether Fungible’s technology actually delivers on its promise or not.

“The fact that I did Juniper should not give me any special treatment,” Sindhu said. “It should be the quality of the idea that I come up with.”

Here’s the second of the two video interview, part of a series of interviews by SiliconANGLE and theCUBE at the Mayfield People First event. (Disclosure: TheCUBE’s coverage of the Mayfield People First event is presented by Mayfield Fund LLC.)

Photo: SiliconANGLE