Upscale AI Unveils SkyHammer Architecture to Redefine Scale-Up AI Networking

Upscale AI today unveiled SkyHammer, a clean-slate architecture purpose-built to power the next generation of scale-up AI networking. Designed to overcome the limits of traditional front-end switches, PCIe fabrics, and proprietary topologies, SkyHammer is designed to deliver deterministic latency, extreme bandwidth, and predictable performance across the world’s largest AI clusters.

Upscale AI officially launched last month with over $100 million in seed funding. "Our deep expertise enables us to architect cohesive AI infrastructure solutions, partnering with leading xPU vendors and AI innovators to meet today’s demands and power what’s next,” Barun Kar, CEO of Upscale AI, said.

Unlike conventional data center networks, SkyHammer was designed from the ground up for unified compute domains, where CPUs, GPUs, and memory operate as one coherent system. Efficiency and scale, Upscale AI argues, can’t come from retrofits — they must start with intentional design.

“With AI compute demands exploding, the industry urgently needs open, scalable, and cost-efficient solutions; SkyHammer is a step squarely in that direction,” said Robert Hormuth, Corporate Vice President, Architecture and Strategy, Data Center Solutions Group, AMD.

SkyHammer is AI-native by design, built on open standards and optimised for collective AI workloads rather than legacy server architectures. It promises simplicity, resilience, and interoperability across diverse hardware and hyperscaler environments — eliminating vendor lock-in and enabling choice.

“AI infrastructure is entering a phase where traditional data center fabrics can no longer keep pace with the scale and synchronization AI demands,” said Alan Weckel, Co-founder of the 650 Group. “Upscale AI’s SkyHammer architecture represents a decisive step toward purpose-built, open, and predictable interconnects that are designed for AI from the ground up.”

Built with deterministic flow control, real-time telemetry, and adaptive load handling, SkyHammer integrates observability and resilience into its core. It supports emerging interconnect standards such as ESUN, UEC, and UALink, ensuring adaptability to future technologies.

“The AI era demands infrastructure that is open, efficient, truly scalable, and economically sustainable,” said Sameh Boujelbene, Vice President at Dell’Oro Group. “Upscale AI’s SkyHammer™ marks a pivotal move toward that vision.”

Products based on the SkyHammer™ architecture are planned for release in 2026.