Artificial intelligence startup Lightning AI Inc. today announcement that it secured $40 million in funding from a group of investors led by Coatue.
Index Ventures, Bain Capital Ventures, First Minute Capital and Mantis VC also participated in the round. Lightning AI, which was until recently known as Grid.ai, previously raised $18.6 million from investors in 2020.
Lightning AI sells a commercial version of PyTorch Lightning, an open-source neural network development tool created by co-founder and CEO William Falcon. The tool is one of the most popular technologies in its category. According to the startup, PyTorch Lightning is downloaded 4 million times per month.
PyTorch Lightning has become popular among developers because it simplifies many manual tasks involved in building AI software. Creating an AI application requires not only developing a neural network, but also setting up and managing the infrastructure on which the neural network runs. Infrastructure maintenance can require a significant amount of manual labor, which slows down machine learning projects.
One of the tasks made easy by PyTorch Lightning is AI training. Large-scale neural networks are often too complex to train with a single processor, forcing software teams to use multiple chips. Using multiple chips can be technically difficult.
Lightning AI states that PyTorch Lightning simplifies the task of training CPU-intensive AI models. The tool also includes features that make it easier for developers to gauge the speed of a neural network. Additionally, if a technical issue occurs in a neural network while it is being trained, developers can use PyTorch Lightning to interrupt the training process and perform troubleshooting.
Software teams typically build AI software using graphics processing units. However, there are instances where a machine learning project may also require the use of other types of chips, such as the AI-optimized tensor processing units that Google LLC provides through its public cloud. PyTorch Lightning reduces the amount of custom code developers have to write to use multiple chip types in a machine learning project.
Lightning AI generates revenue by providing a commercial, managed version of PyTorch Lightning that makes it easier for businesses to use the tool. The startup also offers pre-packaged AI applications for common use cases. A company can use one of these apps from Lightning as the foundation for their machine learning project to avoid having to build everything from scratch.
With its new $60 million funding round, Lightning AI will enhance its platform capabilities and invest in AI research. The startup also plans to continue developing new features for the open-source version of PyTorch Lightning. Lightning AI will expand its team of 60 people, which has already doubled over the past year, to support its product roadmap.
“Until now, there was no way to build production-grade AI applications that considered the entire pipeline from development to production,” Falcon said. “Lightning AI takes the principles that have made PyTorch Lightning one of the fastest growing open source projects in history – simplicity, modularity and sustainability – and applies them to the task of unifying all development of AI and the infrastructure lifecycle.”