• Paperspace says that its platform enables programmers to quickly make a neural network accessible via a programming interface.
  • DigitalOcean intends to incorporate Paperspace into its cloud platform following the acquisition.

DigitalOcean Holdings Inc. recently announced its USD 111 million agreement to purchase Paperspace Co., a Brooklyn-based startup that offers cloud infrastructure for artificial intelligence developers.

DigitalOcean was established in 2011 and offers a namesake public cloud platform. The platform is popular among small businesses and developers because it is made to be comparatively simple to use. In the most recent quarter, DigitalOcean’s cloud services brought in USD 165.1 million, a 30% year-over-year increase.

Paperspace is a venture-backed startup that had previously raised USD 35 million in funding. According to the startup, 500,000 developers who have used its infrastructure for 100 million compute hours make up its user base.

Paperspace also runs a public cloud platform similar to DigitalOcean. The startup’s platform, however, has a much more constrained purpose: it was explicitly created to support AI projects. Nvidia Corp.’s AI-optimized graphics processing units are available through Paperspace and software tools for creating and deploying neural networks.

Through a Jupyter Notebook implementation that runs in the browser, users communicate with the Paperspace platform. That is a popular open-source coding tool for developing AI. The tool’s popularity is mainly because it eliminates the need for tab switching by enabling developers to write an AI model and view the model’s processing results in the same interface.

Software development teams can use Paperspace’s cloud-based GPUs to deploy neural networks they have already created. Various compute instances powered by Nvidia silicon are available from the startup. The most potent example on the market contains eight Nvidia A100 graphics card series chips, which, before the H100’s release last year, were the company’s premier AI chip line.

In addition to its compute infrastructure, Paperspace also provides two types of data storage. Developers can make dedicated block storage and shared drives for each instance. The latter feature makes performing tasks like data exchange between AI models easier by allowing multiple instances to access the same data repository.

Companies typically productize custom neural networks by incorporating them into their applications after developing them. Deploying the neural network on cloud infrastructure and making it accessible through an API is one typical approach to the task. Workloads can send tasks to the AI model through the API and receive processing results.

Paperspace promises to streamline the process of creating APIs. The business claims its platform enables programmers to make a neural network accessible through a programming interface quickly. It achieves that by automating a large number of the typically required manual infrastructure-level tasks.

DigitalOcean intends to incorporate Paperspace into its cloud platform following the acquisition. The company claims that its customers will find developing and implementing AI models more straightforward, thanks to the startup’s technology. The agreement will also allow DigitalOcean to cross-sell its current services to Paperspace’s customers.

Chief Executive Office of DigitalOcean, Yancey Spruill, said, “The combined offerings allow customers to focus more on building applications and growing their businesses and less on the infrastructure powering them.”

According to the company, the deal won’t significantly improve DigitalOcean’s financial results this year. However, the business anticipates that starting in 2024, the Paperwork platform will accelerate its revenue growth.