Optimizing an enterprise for Big Data can be a frustrating scrabble from open-source platform to proprietary vendor to SaaS subscription. And don’t look now — here comes cognitive computing, which is going to demand even more from all levels of the stack to get up and running. One company is already building with this in mind, purporting to offer systems with baked-in cognitive capabilities.

Tom Rosamilia, SVP of IBM Systems, said of cognitive computing, “It’s Big Data on steroids.” He told Dave Vellante (@dvellante) and Stu Miniman (@stu), cohosts of theCUBE, from the SiliconANGLE Media team, during the IBM Edge conference that while many are talking about adding cognitive, IBM is going deep into the architecture and building for it.

Cognitive class architecture

“To me it’s about building for cognitive, saying, ‘What do I need to do within storage, especially within power architectures to deliver for cognitive?'” Rosamilia said, noting that IBM’s OpenPOWER is an ideal canvas for this. Partners like Nvidia Corp. and Mellanox Technologies are pitching in with the technologies they bring to the table.

Case in point: “It turns out that machine learning, for example, an ingestion, really works very well with GPUs [of which Nvidia is a maker],” he said. “The things that we called the play on for OpenPOWER actually were designed to build for the cognitive era.”

The shrinking data window

Rosamilia said that flash storage also seems to have been made for cognitive computing, which demands excessive speed to deliver on its promise. Most of the data an organization gets doesn’t keep well, he said, so they need cognitive computing to extract value from it before the clock runs out.

“The statistic I’ve seen is 60 percent of the data that’s out there goes very quickly to zero value,” he said.

Watch the complete video interview below, and be sure to check out more of SiliconANGLE and theCUBE’s coverage of IBM Edge 2016.

Photo by SiliconANGLE