Google unleashed the general availability of a new Distributed Cloud air-gapped device that pushes the hyperscaler’s increasingly artificial intelligence (AI)-powered services toward more difficult to manage edge environments and use cases.

Sachin Gupta, VP and GM of Google Cloud’s Infrastructure and Solutions Group, told SDxCentral in an interview that the offering ties together a secure cloud experience that can take advantage of Google’s AI systems in challenging locations. This includes a connected iteration and a true air-gapped model, with both integrating compute, storage, and AI capabilities.

The former is a more traditional model where a customer can, for instance, link together edge locations into a centrally accessible cloud environment. The latter fits much more uniquely designed use case scenarios as it can operate with either limited connectivity, which is managed by the user, or “operate in perpetuity without connectivity.”

“Think of that as a small form factor, highly ruggedized – meaning temperature, vibration, dust, all kinds of things. It can handle those. It's mil-spec capable, so there's a military specification that it follows, that it supports,” Gupta explained. “It significantly expands the number of use cases and the number of places where we can bring those advanced Google services for customers.”

Gupta noted that the service was tailor-made for situations government or military use cases where security and data sovereignty were paramount. This includes certification for high levels of shock and impact resistance, with more government testing being conducted.

“Now you're talking about the ability to go to the highest level of unclassified, but also to, in the future, to secret workloads as we complete accreditation,” Gupta said.

Gupta also explained that this system could serve less chaotic use cases like medical research imaging, manufacturing, or the energy sector.

“There's going to be energy use cases where I want to be able to do predictive maintenance, optimization, analytics, but my infrastructure is very susceptible to cyberattack. I cannot have infrastructure that's openly connected to the internet or to a public cloud. I need to get those cloud capabilities locally on site, completely disconnected and air gapped so that I can feed it my data without worry,” Gupta said. “It is going to be industries that have more regulatory requirements that they must adhere to with regards to data, more sovereign requirements they have to adhere to. Data sensitivity is much higher in those industries.”

Those different use cases will also be able to tap into Google’s AI work, with a specific call out to services like translation, speech, and optical character recognition. Gupta noted this platform allows for customers to bring “AI anywhere you need it, and cloud and AI in a box, and now that sort of dream that you see at a distance can become a reality even in those more sensitive and regulatory compliant industries.”

The physical and software support aspects of the service can be tied into already approved Google Cloud partners, with Gupta naming a few in Hewlett Packard Enterprise (HPE), Dell Technologies, Nvidia, Cisco, Palo Alto Networks, and NetApp. Those can be amenable to supporting different deployment models.

“If somebody says they need to put this on a ship and they only have so much space and it needs to be different, we can work with our customers to take our software and load it up in different form factors,” Gupta said. “We do expect that more form factors will be required as we go along.”

Rounding out the GDC portfolio Those new form factors also round out the GDC portfolio.

Google last year unveiled GDC Hosted, which was a step toward serving more stringent security and sovereignty requirements. That product followed the GDC Edge and GDC Virtual products.

Gupta explained that GDC on the connected appliance side had both rack-based products and small form factors, the latter of which was missing for an air-gapped product. This left the air-gapped portfolio “narrow and incomplete” in trying to serve those “forward edge” locations.

“Bringing all of the cloud capability, including operations in a box, required more work, but now that we've completed that work we can make that complete cloud with the AI and data services in that secure environment available in many more use cases,” Gupta said. “We had to collapse a bunch of capabilities into a smaller form factor, so certainly, engineering work was required, but it was more listening to our customers and understanding the problems that they're looking to solve and realizing that we've got some unique technology capabilities to help address those.”