by Jordan Jacobs, Co-founder and Managing Partner, Radical Ventures
The Globe and Mail’s reporting on recent AI company acquisitions reflects concerns over “modest” eight figure sale prices of some Canadian artificial intelligence (AI) companies. We should celebrate every sale that enables a Canadian startup to keep building and hiring, particularly when doing so in Canada, and during a global pandemic. It’s also important to understand that exit valuations are often not reflective of a lack of ambition or anything unique to the Canadian innovation ecosystem. Price is determined by many factors, but often it is about business models.
When it comes to building an AI enterprise software business, founders are often faced with a dilemma: build a software product that generates less revenue per sale but has the potential to scale, versus taking larger short-term revenues that come with requests for customization and services. On an income statement, the former looks like a traditional Software as a Service (SaaS) company, offering similar margins or even higher. The latter has lower margins and is not scalable.
There are two important factors that motivate early stage AI companies to pursue services versus scale: 1) The revenue from services can be significantly greater per deal; and, 2) it is often easier to build an AI solution that requires customization than one that doesn’t. The service-based approach is especially appealing for early-stage companies when revenue generation can feel existential: it can be the key to securing capital and it just might keep the lights on.
I faced a similar dilemma as the Co-founder and CEO of the AI startup Layer 6. We built a deep learning personalization and prediction engine that we tested via proofs of concept with global technology companies and in worldwide machine learning competitions. Our platform outperformed the world’s best companies and research labs in many different domains including personalized recommendations in ecommerce, jobs and media (e.g. what to watch, and music personalization on Spotify); predicting who will get diabetes; and in computer vision accuracy.
Those successes produced both acquisition inquiries and offers to build custom solutions for customers. Repeatedly, we declined lucrative requests to build bespoke AI services. Admittedly, it was difficult to say no to significant early revenue from services, particularly after years of making significant personal financial sacrifices to keep our predecessor technology startup alive. But for deep technology companies looking to build a product that scales, saying yes to services comes at a cost. At best, these contracts end up being distractions from building a product that is truly scalable. At worst, these opportunities reorient the business. Once the spigot of service-based revenue is opened, it can be hard for a start-up, or its funders, to turn it off.
The pressure to build-out service capabilities often happens because enterprise customers may not have the in-house talent or specific expertise to fully leverage machine learning. The good news is that over time, we expect these capabilities to improve and some of that market-pressure to decrease. COVID-19 has already accelerated the adoption of technology across virtually every industry, and we anticipate the adoption of scaled AI products to increase dramatically for the foreseeable future.
I believe that founders need to stay true to their vision. If they have the capacity and know-how to build a scalable AI product then they must surround themselves with a leadership team and funding partners who understand the true long-term value proposition.
We sold Layer 6 in 2018 to TD Bank when we still had very few people but at a valuation reflective of the fact we had built a product company — not a service company. And we used proceeds from that sale to help launch Radical Ventures to reinvest in AI company founders, particularly in Canada. Radical is the Western world’s largest AI-focused venture capital fund. At Radical Ventures, our focus is on supporting ambitious entrepreneurs building global businesses through deeply disruptive technology. This is why we focus on software product companies rather than service-dependent companies. And, wherever possible, we try to protect founders from the slippery-slope of service-focused customizations.
We are still in the early innings of the AI revolution. Globally, and in Canada in particular, there is an explosion of real AI companies delivering products into every industry imaginable. These companies stand to shape the future of how we do business, how we take care of one another and how we live. While “Big Tech” was an early adopter of AI software, there is still plenty of room for AI startups to build truly global enterprises that change the world for the better.
— R —