Artificial Intelligence (AI) is currently one of the most talked-about technologies, often compared to the revolutionary impact of sliced bread. However, just because AI is popular doesn’t mean it’s easy for companies to implement effectively. A recent survey by the Boston Consulting Group revealed that a staggering 74% of organizations are having trouble getting real benefits from their investments in AI.
William Falcon, who created PyTorch Lightning—a widely used open-source AI framework—points out that many businesses make a critical mistake by not realizing how much work is involved in managing AI systems. He likens building an AI platform to creating a custom version of Slack (a popular communication tool), emphasizing that it is complicated, expensive, and not essential to most businesses’ core functions. Instead, he believes that the true value for companies lies in their own data, expertise in their field, and unique AI models, rather than in maintaining the underlying AI technology.
Falcon has an interesting background; he was once a Navy Seal trainee and later interned at Facebook’s AI Research division. He started developing PyTorch Lightning while he was still a student at Columbia University. This framework simplifies the process of using PyTorch, which is a library for building AI applications, by reducing the amount of technical code that developers need to write to set up and manage AI systems.
After leaving a Ph.D. program at NYU, Falcon partnered with Luis Capelo, who previously led data products at Forbes, to turn PyTorch Lightning into a commercial product. Their company, Lightning AI, builds on the open-source framework by adding services and tools that are tailored for businesses.
Falcon explains that with Lightning AI, individual developers can now train and deploy AI models on their own, achieving results that would have previously required large teams of developers. Lightning AI simplifies many of the challenging tasks associated with AI, such as distributing workloads across multiple servers and setting up the necessary infrastructure for training and evaluating AI models. Their main product, called AI Studios, enables users to adjust and operate AI models in the cloud environments they prefer.
Furthermore, companies can use Lightning AI to run AI-powered applications on their own private cloud systems or even on their own servers. The pricing model is flexible, allowing users to pay only for what they use, and there’s a free option that includes 22 hours of GPU (graphics processing unit) usage each month.
Read Also: OPS Concerns: 61 Companies Leave Nigeria in 4 Years
Falcon’s vision for Lightning AI is to make developing AI as straightforward as using an iPhone. He noted that researchers at Columbia have been able to complete hundreds of experiments in just 12 hours using the platform. He also mentioned that many leading AI products have been developed with the help of Lightning, including Nvidia’s NeMo models and Stability AI’s Stable Diffusion.
Lightning AI is gaining traction, with over 230,000 AI developers and 3,200 organizations currently using their platform. Recently, the company secured $50 million in funding, bringing their total investment to $103 million.
However, Lightning AI is not the only player in the market. Competing companies like Comet, Galileo, FedML, and others also offer similar AI management services. Falcon believes that there’s enough demand in the market for managed AI solutions to support multiple companies.
According to a report from Fortune Business Insights, the sector focused on machine learning operations, which includes Lightning AI, could be worth around $13 billion by 2030. With the new funding, Lightning AI plans to focus on attracting new clients, including government agencies, and expanding its platform into new markets.
Falcon is optimistic about the company’s future, stating that with a small but efficient team and a product that has a gross profit margin of over 90%, they are on track to generate between $10 million to $20 million in annual recurring revenue by the end of next year and expect to become profitable shortly thereafter.
