Paperspace

Market position

Paperspace positions itself as a lower-cost, simpler alternative to major public clouds for GPU compute, claiming savings of up to 70% versus large public cloud providers or buying servers, and citing one of the broadest GPU selections in the cloud. Secondary commentary places it among GPU cloud providers alongside RunPod, CoreWeave, DigitalOcean and Hugging Face, and notes it established an early lead in GPU cloud services while the competitive landscape was still forming.

Emphasis on radical simplicity and speed of onboarding (signup to model training in seconds via pre-configured templates), a broad catalog of GPU instance types with per-second billing and no commitments, transparent on-demand pricing with billing alerts and limits, free GPU tiers, and an integrated path from notebook development through training to deployed API endpoints. Gradient's ability to run multi-cloud, in private clusters or on-premise distinguishes it from purely hosted offerings.

Technology

The platform provides an abstraction layer over accelerated computing: instant provisioning of NVIDIA GPU instances, job scheduling and resource provisioning without server management, pre-configured ML framework templates, automatic versioning, tagging and lifecycle management for reproducibility, and browser-based streaming for remote desktop access. Access is available through a web console, desktop apps for Windows, Mac and Linux, an open-source CLI, a RESTful API and a Terraform provider. Infrastructure options include multi-GPU instances up to 8x, SSD storage, shared network drives, private networks and IPsec VPN tunnels; Graphcore IPUs are offered alongside NVIDIA GPUs, and the company describes itself as an NVIDIA Cloud Partner.

Go-to-market

Self-service product-led acquisition through a free tier and free GPUs, Google/GitHub signup, and low-cost entry subscriptions, supplemented by a sales motion ('Talk to an Expert', Contact Sales) for business and enterprise plans. Developer marketing includes a technical blog with community-contributed tutorials, a public forum, documentation and quickstart guides, a referral program, education discounts, conference presence and meetups, and comparison content such as a cloud GPU provider guide.

Individual developers, ML/AI engineers, data scientists, researchers and students on the low end; research groups, startups and professional ML development teams in the middle; and enterprises requiring private clusters, on-premise or hybrid deployments, team management, permissions, resource tracking and regulatory compliance support. Vertical use cases highlighted include machine learning, GPU infrastructure, virtual desktops, 3D workstations and rendering, visual computing, gaming, and medical imaging.

Geography

Headquartered in New York City. The company markets global scale with 24/7 engineering support, and secondary commentary describes expansion of data centers in additional locations to reduce latency for international users.

History

Paperspace went through Y Combinator in Winter 2015 and built GPU virtual machines for emerging workloads, particularly deep learning. By the time of its $13M Series A announcement it reported more than 150,000 active user accounts and a growing business customer base, and had launched what it called its most significant product to date along with institutional partnerships and a developer community. The company later added the Gradient ML platform, a Gradient Installer for running Gradient on any compute cluster, a Graphcore IPU partnership, and support for NVIDIA H100 GPUs. In 2023 it was acquired by DigitalOcean, after which the brand operates as Paperspace dba DigitalOcean and customers are pointed to DigitalOcean's GPU and AI products.

Risks & controversies

Available sources give conflicting founding details: the company blog states the business started in the fall of 2014 and completed Y Combinator in 2015, Y Combinator lists a 2015 founding year, and a secondary business-analysis site variously gives 2014 and 2013. The same secondary site's funding figures (a $3.5M Series A in 2017 and a $16M Series B in 2019) conflict with the company's own announcement of a $13M Series A bringing total funding to $19M.