What do educators teaching deep learning workshops, startup teams spinning up collaborative data science environments and researchers needing GPU access all have in common? They’re stuck in what Jonathan Guinegagne, senior software engineer at Amazon Web Services, calls the “in-between” — needing more than a laptop can provide but lacking the resources of enterprise cloud infrastructure.Jupyter Deploy, a new open source command line interface (CLI) to deploy Jupyter to the cloud in a few minutes, was just launched at JupyterCon 2025 this month to bridge the gap.In this On the Road episode of The New Stack Makers, Guinegagne sat down in San Diego with TNS Editor in Chief Heather Joslyn to discuss how this new tool is making cloud deployment accessible to teams of 10 or fewer who need to collaborate but don’t have dedicated cloud engineers on staff. Solving the 15-20 Resources ProblemData scientists, researchers and educators use Jupyter notebooks to experiment with code, visualize data and document their findings. But when running Jupyter on their laptops, they also run into limitations.Distributed teams cannot securely provide direct access to their local JupyterLab application over the internet, making collaboration difficult. And if their workload requires more compute than a laptop provides — say, GPU accelerators to fine-tune deep learning models — things become even more difficult.“We saw that even simple setups needed something like 15 or 20 cloud resources,” Guinegagne said. “This is not something where you set up your infrastructure with a couple CLI commands.”Now, Jupyter users no longer need to figure out networking, authorization and cloud components on their own. Jupyter Deploy orchestrates an entire end-to-end encrypted stack — from Docker and Terraform to OAuth2 and Let’s Encrypt — with no upfront cloud expertise required. At first, anyway.“Jupyter Deploy gets you started, but it’s not a magic wand that will solve all your cloud problems,” Guinegagne cautioned. “To run this over the long term, you will eventually need to understand the components behind it.”Deploying Any Use Case, on Any CloudWhile Jupyter Deploy comes from AWS’ AI Open Source team, its architecture deliberately avoids vendor lock-in.The team used a template-based approach, allowing Jupyter users to create the combination of services that suits their particular use case.“The base template essentially sets up an EC2 instance as host, running Docker services in Docker Compose and using very simple images that are built on the fly,” Guinegagne said. “But because it is open source, the community will be able to contribute templates — essentially, deployment recipes — for any cloud provider, authentication method or compute environment they need.”Led by Project Jupyter co-creator Brian Granger, the team envisions Jupyter Deploy eventually becoming part of the Jupyter ecosystem itself, governed by the Jupyter council and integrated into the Jupyter CLI.Check out the full episode to hear more about how Jupyter Deploy handles the conda-versus-UV dependency management challenge, why the team integrated Pixie for scientific packages and what’s coming next on the roadmap — including native Kubernetes integration that could reshape how enterprises deploy Jupyter at scale.The post Jupyter Deploy: The New Middle Ground Between Laptops and Enterprise appeared first on The New Stack.