Flexgrid provides a vertically integrated edge cloud that combines custom hardware, power, networking, and real‑time software into pre‑fabricated rack modules (PODs and SLOTs) which are logically linked into a unified compute fabric. This architecture enables deployments up to five times faster, achieves power usage effectiveness below 1.1, and reduces total cost of ownership by up to fourfold while delivering AI workloads with 54 µs latency across distributed locations.
Funding
Funding not disclosed
Founders
Product
Problem
Enterprises and cloud providers often face high latency, costly infrastructure, and complex deployment processes when extending AI workloads to the edge, limiting real‑time decision making and scaling across distributed locations.
Solution
Flexgrid delivers a vertically integrated edge cloud that combines custom hardware, power, networking, and real‑time software into pre‑fabricated rack units (PODs and SLOTs). These units are logically linked to form a unified compute fabric that can be deployed up to five times faster than traditional 2–5 MW sites and operate at a power usage effectiveness (PUE) below 1.1, reducing infrastructure costs by up to fourfold. The platform coordinates AI workloads across the fabric at 54 µs timescales, providing instant, low‑latency intelligence wherever it is needed while seamlessly extending existing cloud environments.
Target Audience
Primary customers are enterprises, cloud service providers, and telecom operators that require low‑latency, high‑performance AI compute at the edge to support real‑time applications and large‑scale distributed deployments.
Features
- Pre‑fabricated, optimized edge rack modules (POD/SLOT) that integrate compute, power, and networking in a single system
- Logical linking of racks to create a real‑time compute fabric spanning north‑south and east‑west traffic flows
- Deployment speed up to 5× faster than conventional large‑scale edge sites
- High density design achieving PUE < 1.1 and up to 4× lower total cost of ownership
- Coordinated AI processing with 54 µs latency across distributed nodes
- Secure, robust infrastructure that can be scaled globally without compromising performance