Case Study: Paperspace achieves seamless global GPU cloud performance and scalability with Digital Realty

Seamless Global Colocation Assures GPU Cloud Performance and Scalability

Paperspace is a US startup that offers a GPU‑accelerated public cloud for machine learning, rendering, VDI and cloud gaming. As a pioneer, it faced extreme power and cooling demands, strict low‑latency backbone requirements and rapid growth — “a rack a week” — but wanted to avoid building its own data centres, so it relied on colocation expertise from COLOpeople to find suitable partners.

Paperspace grew from a half rack at CoreSite to cages in New Jersey and Santa Clara and, on COLOpeople’s recommendation, colocated at Interxion’s Amsterdam campus to serve European customers. Using Interxion’s connectivity and remote‑hands services, Paperspace was live in a month, achieved lower latency and better application performance for European users (including a cloud‑gaming client), and gained scalable, reliable infrastructure without any on‑site travel.

Paperspace

Dan Kobran

Co-founder