AI may be the hottest thing since sliced bread. But that doesn't mean it's easier to develop and run. According to a recent Boston Consulting Group poll, 74% of organizations struggle to derive value from their AI investments.
William Falcon, creator of the popular open source AI framework PyTorch Lightning, says one of the biggest mistakes companies make is underestimating the amount of work involved in AI orchestration. Masu. “Currently, building your own AI platform is like building your own Slack: it's complex, expensive, and not core to your business,” he told TechCrunch. “The value for enterprises lies in their data, domain knowledge, and unique models, not in maintaining an AI infrastructure.”
A former Navy SEAL and intern at Facebook AI Research, Falcon began developing PyTorch Lightning while an undergraduate at Columbia University. This framework provides a high-level interface to the AI library PyTorch and abstracts the code for setting up and maintaining AI systems.
After dropping out of his PhD at New York University, Falcon decided to collaborate with Luis Capelo, former head of data products at Forbes, to commercialize PyTorch Lighting. Their venture, Lightning AI, takes an open source framework and layers enterprise-grade services and tools on top.
“We have thousands of developers training and deploying models independently. [with Lightning AI] We're at a scale where we would have needed a development team without Lightning,” Falcon said.
Lightning AI handles typically tedious tasks like distributing AI workloads across servers and provisioning the infrastructure needed to evaluate and train AI. The company's flagship product, AI Studios, allows customers to fine-tune and run AI models in their preferred cloud environment.
Lightning AI development platform. Image credit: Lightning AI
Businesses can also use Lightning AI to host AI-powered applications running in private cloud infrastructure or on-premises data centers. Pricing is pay-as-you-go, with a free tier of 22 “GPU hours” per month.
Falcon said the goal of Lightning AI is to make AI development “as intuitive as using an iPhone.” He claims that the platform has allowed Cisco to reduce infrastructure setup time to two days, allowing researchers at his alma mater, Columbia, to complete hundreds of experiments in 12 hours.
“Most people don't know this, but many of the world's leading AI products are trained or built on Lightning,” Falcon said. “For example, Nvidia's model suite NeMo was built using Lightning tools, including Stable Diffusion with Stability AI.”
Lightning AI certainly has momentum. More than 230,000 AI developers and 3,200 organizations currently use the platform, and the company recently raised $50 million in a funding round.
However, there is competition. Comet, Galileo, FedML, Arize, Deepset, Diveplane, Weights & Biases, and InfuseAI offer an equal mix of paid and free AI orchestration services.
Falcon believes the market for managed AI solutions is large enough to support many players. And he's probably not wrong. According to Fortune Business Insights, the machine learning operations industry (the field of Lightning AI) could be worth approximately $13 billion by 2030.
A recent $50 million investment led by Cisco, with participation from JP Morgan and Nvidia, brings Lightning AI's total war chest to $103 million. The New York-based company, which has 50 employees, plans to use the proceeds to acquire new customers, including government customers, and expand its Lightning platform into new markets.
“With a lean, high-performance team and products with gross margins above 90%, we expect to reach $10 million to $20 million in annual recurring revenue by the end of next year and return to profitability shortly thereafter,” Falcon said. ” he said. ”