Not many companies would dare to fight with Amazon, Microsoft and Google at the same time, but three-year-old Paperspace thinks it can carve out a cloud computing niche by rolling out the red carpet for data scientists. Today it launched new GPU-powered virtual machines that utilize Nvidia’s Pascal cards. With pre-installed machine learning frameworks, Paperspace caters to the emerging market of prosumer and enthusiast data scientists.
“Amazon Web Services (AWS) is amazing, but it’s hard to get going,” said Dillon Erb, co-founder of Paperspace.
To make machine learning in the cloud more accessible, Paperspace offers users a familiar Linux desktop from within their web browser. Anyone can then use either a secure shell or a web-based terminal to implement code. And in addition to interface and hardware improvements, Paperspace aims to undercut on price by offering access to Pascal chips with 2560 CUDA cores and 16GB of memory for just $0.65 per hour.
“In the last 18 months, we have seen a rise in people asking for GPUs,” added Erb.
Machine learning as a service (MLaaS) startups have taken heat in recent months. There are a number of reasons for this, but one of them is that there’s a mismatch between the market of highly technical engineers and products that simplify the development process for novices.
To be clear, Paperspace is in a category of its own, aside from startups like Bonsai and H20.ai, but the metaphor is still appropriate. Existing cloud computing providers that are already embedded in enterprises will continue to move toward democratization — that doesn’t guarantee whitespace for anyone else, especially when you take into account the sheer cost of building and upgrading data centers.
Y Combinator, NYU and Insight Data Science are early partners with Paperspace. The new GPU-powered virtual machines will be used by Insight for its professional training program. YC is also experimenting with the streamlined system for use within its new cohort of AI startups.