Copyright SiliconANGLE News
Enterprise infrastructure is no longer measured in servers or racks but in the architectural decisions that determine what those systems can accomplish at scale. As organizations race to deploy artificial intelligence workloads, the question is less about acquiring technology and more about integrating it through reference architectures that support rapid iteration without forcing teams to start from scratch. That integration challenge took center stage at the recent Nvidia GTC Washington, D.C. event, where partnerships between silicon providers, networking vendors and systems integrators demonstrated how reference architectures and validated ecosystems are reshaping deployment timelines. When enterprises can extend proven environments while meeting new performance demands, adoption accelerates, according to Gilad Shainer (pictured, left), senior vice president at Nvidia Corp. “The data center has become the unit of computing,” he told theCUBE. “The way that you connect [graphics processing unit] ASICs will determine what that data center can do. We also know that we’re on an annual cadence of technology because there’s a fast pace of progress. We want that AI technology to be accessible by everyone.” Shainer and Will Eatherton (right), senior vice president of data center, internet and cloud infrastructure engineering at Cisco Systems Inc., alongside other technology executives and systems integration leaders, spoke with theCUBE’s John Furrier and Savannah Peterson at the recent Nvidia GTC Washington, D.C. event, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. Conversations centered on how reference architectures, validated ecosystems and mission-driven innovation are accelerating AI infrastructure deployment across defense, cloud and municipal environments. (* Disclosure below.) Here’s theCUBE’s complete video interview with Will Eatherton and Gilad Shainer: Here are three insights you might have missed from theCUBE’s coverage from Nvidia’s GTC Washington, D.C. event: Insight #1: Reference architectures make AI infrastructure accessible without reinventing enterprise systems. Nvidia’s strategy centers on transforming data centers into AI supercomputers that can evolve at the same pace as modern workloads, requiring a shared architectural language between compute and networking that reference architectures provide. Cisco’s collaboration provides that bridge, extending Nvidia’s AI stack into the enterprise with familiar operational tools, according to Eatherton, who was joined by Mike Trojecki, senior director of the AI practice at WWT. By combining Nvidia’s Spectrum-X Ethernet technology with Cisco’s Nexus software, customers can deploy thousands of GPUs while maintaining the interfaces and systems they trust. “For us, being able to bring in Nexus, Nexus Dashboard, the new hyper-fabric AI, these are ways that our customers with small teams and expertise that they already have can add that,” Eatherton said during the event. “They shouldn’t have to be experts on [RDMA over Converged Ethernet] and [Remote Direct Memory Access] and all those aspects of troubleshooting. That’s something that the tooling and management software should do that for them. They can trust us to put that together and validate.” Cloud platforms such as The Constant Company LLC, known as Vultr, are moving beyond the traditional hyperscaler landscape by prioritizing early investments in AI infrastructure and stringent compliance requirements. Vultr’s platform reaches 90% of the world’s population in 2 milliseconds, and the company recently completed FedRAMP compliance — a government-wide program for cloud security assessment, according to Kevin Cochrane, chief marketing officer of Vultr. As new GPU providers enter the market rapidly, security and compliance have become key differentiators. “In this day and age of AI infrastructure, you have a lot of new providers,” Cochrane told theCUBE. “Now they’re renting those GPUs out to you. you have to wonder how secure, how compliant are these things.” The Cisco N9100 switch runs both Cisco’s Nexus and SONiC operating systems while built on Nvidia’s Spectrum ASIC, allowing customers to tailor infrastructure for neocloud providers interested in GPU-as-a-service models. World Wide Technology LLC, as Cisco’s largest integration partner, bridges innovation and implementation through its AI Validation Lab and AI Proving Ground, where hardware, software stacks and customer-specific architectures are tested and operationalized, according to Trojecki. “When we look at what we’re doing across AI, and we’re here talking a lot about AI stuff, it’s really across high-performance architectures,” he told theCUBE. “It’s some of the AI solutions and physical AI, robotics and digital twins. We work very closely with Cisco and with our customers to develop some of the coolest things on the planet right now.” Here’s theCUBE’s complete video interview with Kevin Cochrane: Insight #2: Co-creation and validated ecosystems drive the fastest AI infrastructure deployments. The N9100 switch was developed in under a year through close collaboration between Cisco, Nvidia and WWT — a timeline that reflects a broader shift in how enterprise infrastructure and reference architectures get built. When ecosystem partners co-create rather than operate in traditional vendor-customer roles, deployment cycles compress dramatically, according to Trojecki. “It’s the fastest I’ve ever seen Cisco move, and I’ve been working with Cisco for 20-plus years,” he said during the event. “The speed at which they got the AI and compute side up to speed, and what they’re doing here … it’s a whole different Cisco right now.” The defense sector is experiencing a similar transformation as systems integrators shed their reputation for slow execution and embrace agility built on proven reference architectures. Concept-to-deployment timelines that once stretched across years now compress into months, with pace emerging as the decisive competitive advantage over traditional priorities such as price and quality, according to Bob Venero, president and chief executive officer of Future Tech Enterprise Inc., who was joined by Travis Garriss, chief information and digital officer of Northrop Grumman Corp., during an interview with theCUBE. “A lot of people look at the systems integrators as this slow-moving aircraft carrier, and with new leadership and stuff, they’re turning into a speedboat,” he told theCUBE. “This is a perfect example of being able to take something from concept to design to completion in three to four months is just unheard of in the market, and Northrop Grumman is leading that path big time.” Here’s theCUBE’s complete video interview with Bob Venero and Travis Garriss: Insight #3: AI’s real test comes in solving practical problems that matter to people and communities. SHI International Corp. partnered with Hewlett Packard Enterprise Co. to deploy AI solutions addressing critical municipal challenges in Vail, Colorado. The collaboration demonstrates how smart city technology can solve real-world problems in communities facing seasonal population surges and infrastructure constraints, according to Jack Hogan, vice president of advanced growth technologies at SHI. “What we have here is really a digital AI ambassador, our digital civic ambassador for the town of Vail, Colorado,” he said during the event. “We got involved with the town of Vail where we were able to apply some practical AI solutions to solve some big city challenges in a town that really only has 2,400 people but applies that in the times where they scale up to almost 50,000 in their peak season.” Using AI-powered video analysis through its partnership with HPE, ProHawk Technology Group Inc, IronYun Inc. USA, dba Vaidio, Kamiwazza.ai and Blackshark.ai Inc, SHI helps Vail detect the earliest signs of wildfires by transforming camera feeds into actionable insights. The system restores video quality in low-light scenarios and identifies wildfire indicators before they become catastrophic, according to Hogan. “We had the head of the fire department prevention … talked to them about the different camera feeds they had around the town, to be able to use that, to use a system with ProHawk to do restoration of that video in low-light scenarios and bring that to a level that you can now start doing video analysis through AI, through the Vaidio solution, to identify the early stages of a wildfire,” Hogan said during the event. “You can imagine the impact of a wildfire out of town like Vail, Colorado. It was something that really blew their mind and is an amazing solution, because what it’s now solving for is the loss of life.” To help Vail meet the ADA Section 508 compliance deadline, SHI deployed Kamiwaza’s Accessibility Remediation Intelligence Agent, which automatically scanned and remediated archives, PDFs and images. The town had budgeted three years and several million dollars to solve the problem; the agentic AI solution completed the work in a weekend, according to Hogan. “There’s a looming deadline of April next year … the federal government has a mandate called the 508 compliance requirement under the American Disabilities Act to ensure that all civic information is compliant for those that have hearing or vision challenges,” he told theCUBE. “The Kamiwaza ARIA solution was the perfect answer for that. The town of Vail had budgeted three years and a couple million dollars to ultimately solve this problem; we solved the problem in literally a weekend using an agent.” Here’s theCUBE’s complete video interview with Jack Hogan: To watch more of theCUBE’s coverage of Nvidia GTC Washington, D.C. event, here’s our complete video playlist: (* Disclosure: TheCUBE is a paid media partner for the Nvidia GTC Washington, D.C. event. The sponsors of theCUBE’s event coverage do not have editorial control over content on theCUBE or SiliconANGLE.) Photo: SiliconANGLE
