Miris is transforming the way 3D content is delivered at scale. Learn how Miris uses AI and CoreWeave Cloud to stream volumetric 3D content without relying on per-user GPUs.
Challenge
Miris, a cutting-edge 3D technology company specializing in real-time 3D streaming and spatial content delivery, saw a significant opportunity quickly evolving. Advances in generative AI were dramatically lowering the barrier to creating high-quality 3D content, unlocking a surge in volumetric data across industries.
But delivering that content at scale remained a challenge. Existing approaches forced a tradeoff between quality and scale: either compressing assets for web delivery or relying on pixel streaming that couldn’t efficiently reach large audiences.
To unlock the full potential of this new wave of 3D content, Miris needed an AI cloud platform that could deliver photorealistic experiences in real time, without compromise.
- Preserve visual fidelity without compromising quality across devices
- Scale to global audiences without per-user cloud GPU dependencies
- Deliver 3D content faster with minimal latency and load times
Solution
Miris partnered with CoreWeave to build a new purpose-built architecture for delivering photorealistic 3D content—one that eliminates the traditional tradeoff between quality and scale.
By combining AI-driven volumetric streaming with CoreWeave’s AI cloud, Miris can now process massive 3D datasets upstream and stream them more efficiently to end users.
- Upstream GPU processing
Instead of allocating GPUs per user, Miris processes 3D assets once upstream using CoreWeave infrastructure, dramatically improving scalability and efficiency.
- Volumetric streaming architecture
Miris uses GPU compute upstream to optimize 3D assets into an adaptive spatial streaming format, enabling high-fidelity delivery across devices and network conditions.
- High-performance AI infrastructure
CoreWeave’s cloud platform, including tightly integrated storage and Kubernetes orchestration, supports massive throughput and parallel processing.
- Elastic scaling for asset pipelines
The platform handles thousands to millions of assets per day, ensuring consistent performance for demanding workloads.
- Adaptive streaming delivery
Content dynamically adjusts to network conditions, enabling instant interaction without buffering or degradation in experience.
Outcomes
With CoreWeave, Miris delivers photorealistic 3D experiences instantly at global scale. This unlocks new applications across retail, media, robotics, and digital twins while eliminating traditional tradeoffs between quality, cost, and performance.
Global scale without GPU bottlenecks
No more GPUs-per-user. Miris processes and optimizes 3D assets upstream on CoreWeave, then streams them efficiently to end users. This enables high-fidelity experiences for millions of users simultaneously and dramatically reduces infrastructure overhead.
Near-instant interactivity
Load times and user drop-off? Solved. Miris delivers interactive 3D content instantly, with adaptive streaming that responds to network conditions. This keeps users engaged and improves conversion and retention.
Higher GPU efficiency, lower cost per workload
By streaming only what each device needs and optimizing memory usage on both ends of the pipeline, Miris extracts significantly more output from the same GPU footprint. Upstream, asset optimization runs efficiently on CoreWeave infrastructure with high VRAM utilization. Downstream, adaptive delivery means client devices receive only the spatial data required for what the user is viewing, not the full asset. The result is more throughput per GPU and lower total cost of ownership.
