Buck Jordan here to talk about one of my all-time favorite subjects: artificial intelligence.

Whether it is a simple explainer on AI, Google’s LaMDA scandal, or the battle to be a global AI superpower, I’ve written a lot about AI this year.

But I’ve only touched the tip of the iceberg of an industry PWC forecasts will add an eye-popping $15.7 trillion to the global economy by 2030. That’s a lot of zeros (12 to be precise). It is no wonder VCs invested $75 billion in AI startups in 2021 – the best fundraising year yet for the industry!

Well, one of the most compelling opportunities in AI throughout the next 5-10 years will be industrializing AI.

You see, at this point, AI has been confined to research institutes, universities, and the very biggest tech companies (Netflix, Google, Microsoft, etc.). These companies have monopolized AI talent with enormous salaries and been the main companies deploying AI in our everyday lives (think Alexa, Siri, or Netflix’s recommendation algorithm).

However, this industry won’t add $15.7 trillion to the global economy solely through universities and big tech companies.

For AI to hit these lofty targets, it needs to be deployed widely across industries, companies, and daily lives. But to effectively do that, we need to build the tools to scale, industrialize, and enable AI to be widely adopted.

There are several key components that feed into industrializing AI, which include:

  1. Data Management: Capturing, cleaning, storing, and managing data used to train and tune AI models
  2. Model Development, Training, and Deployment: Tools that help engineers develop, train, and test different AI models
  3. Hardware: Specialized computing hardware to power the whole AI workflow
  4. Enterprise AI Platforms: Products that easily and seamlessly deploy AI models into enterprise workflows

When it comes to data management, there are several great companies pioneering the field, including Snowflake, Databricks, and MongoDB. These companies help engineers organize vast datasets, clean them, update them, and continuously manage them. This removes lots of the manual work that’s historically gone into data management, and these companies are making it easier and easier to capture, store, and feed data into the training and development of machine learning models.

Data is the tip of the spear in AI development – the first step – and there remain huge opportunities for new startups building products in this space.

Similarly, model development, training, and deployment is one of the most critical things to get right to industrialize AI.

It probably isn’t surprising to hear that a lot of the tech giants spearheaded many of the tools that serve as the foundation of model development and deployment in the market today, including Google’s TensorFlow and Microsoft’s Azure Machine Learning Studio.

Well, in the wake of the pioneering work of these big tech companies, a whole new generation of startups building on top of this work have been founded, such as Algorithmia, DataRobot, and Weights and Biases. Despite the innovation so far, there are lots of problems left to solve in model development, training, and deployment, and there will be many more startups founded in this space throughout the next decade.

Now, a lot of what we covered so far are products that help engineers build better AI products faster and deploy them into the real-world more easily. However, not every company is going to have an army of software engineers on staff who can build AI products from scratch.

In fact, the vast majority of companies won’t have very many and will need someone to build a software solution for them with AI integrated into it.

There are several companies at the frontier of this trend today, including Palantir, C3.AI, and Alteryx. Each of these companies serves a very specific type of customer, and the reality is, the majority of enterprises in the market don’t have providers of AI software solutions today.

That’s why the opportunity in enterprise AI products is probably bigger than all the other categories combined!

Despite the incredible innovation we’ve seen in AI throughout the last 10 years, the reality is there is so much left to do. AI is still a relatively niche industry composed of deep specialists who work for just a few companies and educational institutions.

Aside from a few examples, we really don’t interact with products improved by AI in our day-to-day lives very often, and most companies don’t experience the benefits of AI in their workflows.

Stage one of the AI revolution was advancing the capabilities of what AI can do. That stage is now done.

Stage two will be bringing these new, powerful capabilities to the world and making it easier for engineers to build AI products. And, in turn, for companies to include AI in their workflows.

Keep your eyes open for companies industrializing AI. They are a big part of the next wave of AI products (and, potentially, your angel investment portfolio)!

Until next week.

Buck Jordan