AI is one of the least concise-about technologies of the moment. It seems like every new business calls itself an “AI company.” This is not too far from the truth, especially since data capture and manipulation, cloud computing, and machine learning are all getting cheaper and more common. AI is used by a lot of tech companies these days. But true AI startups, whose products are built on proprietary, computational, deep-tech AI, are just starting to pop up in larger numbers. They are finding customers in a wide range of industries that want to automate their businesses with their technology. Startups and bigger tech companies that want to sell AI technologies must choose the right ways to make money.
I’ve assisted scores of entrepreneurs as they created their AI firms from the ground up over the last 20 years as an investor in automation startups, helping them refine their go-to-market plans and business models. AI companies have to handle sales differently than most enterprise SaaS companies. Still, they have great long-term recurring revenue prospects if they set up their models correctly. That doesn’t mean there’s only one right way to sell AI software. AI companies can make money in at least five main ways. Here is a summary of each of these models; hoping these rules will help both founders and investors.
Profit Share:
Few AI companies use a revenue share model today, but it’s likely to become more popular in the years to come. For AI companies that work in marketing, advertising, or online shopping, for example, they could get a cut of each sale that started with their AI chatbot and went through to the end. The “cost-savings share” is a part of this model. The AI company gets paid when a customer saves money because of an AI company’s platform. In customer service, healthcare, or insurance, for example, if a customer doesn’t call the call center, chooses a lower-cost plan, or chooses an in-network provider because of an AI application, the cost savings could be shared with the AI company.
SaaS Innovation:
This way of making money is becoming more popular for AI companies that offer a one-time or annual service. Often, companies that have developed sophisticated artificial intelligence systems on the back end, such as robotics companies, will offer their solutions as a service. An example would be a robot that picks apples in the fall and weeds farmers’ fields twice a year.
It could be an automaton that measures a building project’s floor plan. SaaS also workings well for software-only AI companies that offer similar one-time services. For example, it could be AI software used to train a robot to drive itself. Once the robot is instructed, the service is no longer needed regularly and may only be used once a year to update the robot. This model works well for people used to paying for services, like a farm that pays x dollars for y tons of apples.
Based on usage
This model is similar to SaaS as a Service but a bit more complex. It lets AI companies charge for their software based on pre-defined usage metrics, such as the number of queries, the amount of data served, the number of models created, or other metrics based on consumption. AI companies in retail, finance, engineering, and modeling self-driving cars are in a good position to use this model successfully.
Subscription to SaaS:
Most companies that make enterprise software use this traditional way to make money, and AI companies can do very well. The SaaS subscription model has many different types, such as monthly per-seat, monthly all-you-can-eat, annual, etc. This model might make the most sense for AI companies that offer a specific vertical application, like those in HR, healthcare, education, real estate, or logistics. Most of these users pay for software through a SaaS subscription model.
Advertisement
AI companies use transactional models less and less, but they are becoming increasingly important. This way of making money means that an AI company gets paid a fee for every transaction on its platform or every time its software is used to do a specific task. The transactional model works well for HR, retail, and insurance companies that use AI.
These models can be put together in different ways, and many companies that make AI software choose to use two or more of them. And every day, I talk to the founders of AI companies who think outside the box about how to make money from their technologies. Some robotics companies give away or rent their hardware for cheap in exchange for promises to use their AI software for at least a year. Some are giving away free versions of their core software and only charging for good services for big businesses. If you are the founder of a software company based on AI, think about how you make money from your technology.