We knew AMD Advancing AI would be an exciting week alongside our ecosystem partners in San Francisco, but the event even exceeded our own high expectations. From product launches to industry applications and countless conversations on the show floor, the event made clear that the enterprise AI ecosystem is thriving. Our itinerary certainly reflected that.
Announcement: AMD Helios™ rackscale solution and AMD Instinct™ MI455X GPUs on Vultr
The AMD Helios™ rackscale solution delivers enterprises more than GPU acceleration. The 72-GPU configuration also provides 18 AMD EPYC™ CPUs, scale-up and scale-out AMD Pensando™ networking, and ROCm™ software, with up to 1.67 PB/s of memory bandwidth and 31 TB of HBM4 memory capacity per rack.
AI is moving from experimental pilots to large-scale production services, and teams need the operational efficiency to deploy workloads fast. From agentic AI to inference, fine-tuning, and training, the solution gives teams performance, flexibility, and control for accelerated AI at scale.
TESSERA and full-stack AI applications
On Wednesday, we publicized the University of Cambridge’s TESSERA project that brings planetary-scale Earth observation to life. TESSERA takes in years of satellite imagery and turns it into AI representations of every 10-meter square of Earth’s surface, unlocking a range of applications: precision agriculture, biodiversity and conservation efforts, land-use change detection, renewable energy infrastructure monitoring, and more.
It utilizes AMD Instinct MI325X GPUs and AMD EPYC processors with Vultr Cloud GPU and Vultr Bare Metal, another example of a practical and impactful application of AMD software deployed on Vultr infrastructure. At AMD Advancing AI, we also announced a few others, covering decision intelligence in insurance, operational intelligence in healthcare, and physical AI in robotics.
These industries don’t share obvious links on the surface. But in the agentic workflow, the common denominators are many: agentic AI orchestrates agents, enterprise data provides context, CPUs coordinate agents and workflows, GPUs accelerate inference and reasoning.
The result is production-scale AI that drives business value while maintaining the critical human oversight layer.
Vultr leadership takes the stage
We got to watch our CMO, Kevin Cochrane, and Vice President of Solutions Engineering, Duncan Ng, lead a couple of insightful and practical sessions at AMD Advancing AI.
First, Ng led “Production-Ready AI Training and Inference with Vultr,” which covered how teams build training pipelines and deliver production inference on Vultr using AMD Instinct GPUs and ROCm open architecture. The session focused on data locality, secure networking, Kubernetes orchestration on Vultr Kubernetes Engine, benchmarking, scale-out, and cost-efficiency.
Then, Cochrane led a breakout titled “Scaling AI in Production with AMD and Vultr,” where he shared how our partnership enables operational efficiency and global scalability for workloads from training to inference, pointing to real-world use cases across healthcare, finance, retail, manufacturing, and hospitality.
Cochrane also hosted two new episodes of his podcast series, Agent(X), with guests from AMD and VAST Data. You can find them here.
The enterprise AI ecosystem shines
From those sessions, to customer stories and use cases, to new product launches, to conversations with visitors at Vultr booth 950, the following was clear: The AI landscape has grown to its current stature thanks to an ecosystem of partners working hand-in-hand to deliver comprehensive solutions at every layer. We’re proud of all the advancements our community is making to unlock new possibilities in the cloud. Want to learn more about them? Catch us at another event or get in touch today.
