Open Source, AMD GPUs, and the Future of Edge Inference: Vultr’s Big AI Bet
On this episode of the Data Center Frontier Show, we speak with Kevin Cochrane, Chief Marketing Officer of Vultr, about the company’s bold strategy to power the next decade of AI-native applications.
With a global footprint across 32 cloud regions and a $3.5 billion valuation, Vultr is staking its future on one big idea: the world is becoming one massive inference engine - and it will need scalable, cost-effective, open infrastructure to power it.
A New 10-Year Infrastructure Cycle
Cochrane lays out Vultr’s vision with clarity and conviction: the industry is entering a new 10-year infrastructure cycle, akin to the shifts seen in the late 1990s during the move to colocation and the 2010s cloud migration. This time, the catalyst is generative AI. Just as cloud-native applications redefined how software is built and deployed, Cochrane believes AI-native applications, powered by containerized code, models, and data, will require a complete rethink of how and where infrastructure is deployed.
In this next phase, enterprises won’t simply fine-tune frontier models or centralize GPU clusters for R&D. Instead, they’ll operationalize inference at massive scale,deploying real-time AI agents across geographies to support daily business functions. That shift, Cochrane argues, demands a new breed of infrastructure: highly efficient, globally distributed, and built for cost-effective inference, not just training.
AMD GPUs Purpose-Built for Inference
That’s where Vultr’s partnership with AMD enters the picture.
Vultr is now deploying both the AMD Instinct MI300X and MI325X GPUs across its global cloud platform. Cochrane explains that these chips are optimized specifically for inference workloads. With more onboard memory than competing GPUs, they allow large models to run on fewer units, reducing both power consumption and cost per deployment. For example, a model like Llama may require four GPUs from a different vendor, but only two MI325X units from AMD. That translates into significant gains in scalability and operational efficiency.
For customers looking for top-tier performance, the MI325X represents the cutting edge. But for others, the MI300X may offer a better price-to-performance ratio. Vultr’s goal is to give enterprise developers flexible options to fit their needs, wherever those needs may emerge.
Global Reach Demands Scalable, Distributed Inference
“AI-native applications are going to live everywhere, close to your customers, close to your employees,” Cochrane says. “You don’t know where demand will spike. That means your infrastructure has to be ready to scale up and down, with the best power efficiency, in every region where you do business.”
This kind of global responsiveness is only possible thanks to Vultr’s deep integration with Supermicro, whose rack-scale servers enable fast deployment of AMD GPUs at scale. The collaboration ensures that Vultr can move quickly to bring hardened, high-performance hardware online in new markets, meeting the latency and density needs of modern AI inference.
ROCm and the Power of Open Source AI
But hardware is only one part of the equation. The other is software and here, Cochrane is equally bullish on the open-source movement surrounding ROCm (Radeon Open Compute), AMD’s open-source software stack for AI and HPC.
Unlike closed ecosystems like CUDA, Cochrane says ROCm embraces open standards, giving developers more control and flexibility. “Open source and open standards are the key to innovation,” he says. “That’s always been true, and it’s more true than ever in AI.”
Vultr is actively fostering the ROCm developer ecosystem, hosting global hackathons in cities like London, Paris, and Berlin to encourage adoption and experimentation. These events have already drawn hundreds of developers eager to test ROCm’s latest capabilities, including the recent ROCm 6.4 release.
Cochrane highlights that ROCm already enjoys zero-day integration with platforms like Hugging Face. This means any containerized model published to Hugging Face is instantly compatible with ROCm and AMD GPUs, no porting required. That kind of seamless compatibility accelerates developer productivity and lowers barriers to entry, making open infrastructure a viable alternative for enterprises looking to avoid vendor lock-in.
Open Standards: The Key to Cloud-Scale AI
The broader goal, Cochrane says, is to replicate the success of open source in the cloud world within the GPU ecosystem. Drawing from his background evangelizing open-source software in the early 2000s, including time spent working closely with the Apache Software Foundation, he argues that nearly every SaaS application today is already built on open-source components. AI, he believes, will follow the same trajectory.
“We won’t see the level of AI adoption we need without open communities and open innovation,” he says. “Enterprises won’t rebuild their tech stack on black-box systems. They’ll do it with open standards that offer transparency, control, and flexibility.”
AI Infrastructure as a Core Growth Engine
As the conversation wraps, Cochrane reflects on Vultr’s growth and what’s ahead. With new capital and strategic partnerships in place, the company is doubling down on its core bet: that AI workloads,particularly inference,will become as ubiquitous as CPUs were in the last infrastructure cycle.
“We’re going to see every city on the planet with GPU clusters, supporting real-time inference in every aspect of daily life,” Cochrane says. “How people drive, how they shop, how they interact at work, it’s all going to be backed by AI running on GPUs. The scale will be 100x what we’re doing today for training.”
That future is coming fast. And with AMD’s hardware, Supermicro’s servers, and ROCm’s open-source momentum, Vultr is building the foundation to meet it.
Recent DCF Show Podcast Episodes
-
Flexential CEO Chris Downie on the Data Center Industry's AI, Cloud Paradigm Shifts
-
CyrusOne CEO Eric Schwartz Talks AI Data Center Financing, Sustainability
-
Prometheus Hyperscale Pushes Data Center Horizons to 1 GW
-
Quantum Corridor CEO Tom Dakich On U.S. Midwest Data Center Horizons
Did you like this episode? Be sure to subscribe to the Data Center Frontier show at Podbean to receive future episodes on your app.
About the Author
Matt Vincent
Matt Vincent is Editor in Chief of Data Center Frontier, where he leads editorial strategy and coverage focused on the infrastructure powering cloud computing, artificial intelligence, and the digital economy. A veteran B2B technology journalist with more than two decades of experience, Vincent specializes in the intersection of data centers, power, cooling, and emerging AI-era infrastructure. Since assuming the EIC role in 2023, he has helped guide Data Center Frontier’s coverage of the industry’s transition into the gigawatt-scale AI era, with a focus on hyperscale development, behind-the-meter power strategies, liquid cooling architectures, and the evolving energy demands of high-density compute, while working closely with the Digital Infrastructure Group at Endeavor Business Media to expand the brand’s analytical and multimedia footprint. Vincent also hosts The Data Center Frontier Show podcast, where he interviews industry leaders across hyperscale, colocation, utilities, and the data center supply chain to examine the technologies and business models reshaping digital infrastructure. Since its inception he serves as Head of Content for the Data Center Frontier Trends Summit. Before becoming Editor in Chief, he served in multiple senior editorial roles across Endeavor Business Media’s digital infrastructure portfolio, with coverage spanning data centers and hyperscale infrastructure, structured cabling and networking, telecom and datacom, IP physical security, and wireless and Pro AV markets. He began his career in 2005 within PennWell’s Advanced Technology Division and later held senior editorial positions supporting brands such as Cabling Installation & Maintenance, Lightwave Online, Broadband Technology Report, and Smart Buildings Technology. Vincent is a frequent moderator, interviewer, and keynote speaker at industry events including the HPC Forum, where he delivers forward-looking analysis on how AI and high-performance computing are reshaping digital infrastructure. He graduated with honors from Indiana University Bloomington with a B.A. in English Literature and Creative Writing and lives in southern New Hampshire with his family, remaining an active musician in his spare time.
Voices of the Industry
