I was recently watching this talk from Sequoia about the startup opportunities in the AI landscape, and this figure representing the major organizations for each technological wave really stood out to me:

Figure 1: The Major Technological Organizations per Wave from Sequoia

So I was trying to ask myself the following question: what will be the most definitive characteristic of the major winning AI companies of this decade?

Certainly building unique user experiences will be one pillar, but that is a characteristic of any successful business and not just for AI companies.

So let’s take a look at some of the current leading AI companies with high growth curves to see if we can notice a pattern. OpenAI created ChatGPT which really grabbed the entire world’s attention to the power of AI, which can be summarized as a research breakthrough that aligned with a unique user experience. Another excellent product they released was DALLE, but what really drove insane growth usage was their Ghibli style feature into image generation – another research milestone that pushed the frontier while aligning with something people love.

For code, Cursor really nailed the user experience while training customized models for each part of the workflow from simple tasks like code edits to long horizon async agents. Anthropic also did excellent work on Claude for code generation, which really helped Cursor take off and has made products like Replit or Devin possible. And the list goes on ..

In the previous technological waves, especially post internet era – the frontier organizations didn’t doubt whether building a solution was feasible for a customer problem from a technology pov. It was more about moving backward from user experience to architecture backend systems that provide something people like. For example, although Netflix didn’t use entirely off the shelf tools to build their micro-services architecture, they were able to do engineering innovations by hiring top tier talent.

However, what’s unique about this technological wave is pushing the frontiers of whats possible and believing in things like the scaling laws. Even some Turing award recipients were doubtful about how far the field will be able to push the current state of the art, and have since changed their positions.

Not only is top tier AI research able to convert the impossible to reality, it also has enormous impact on defining the user experience and the costs of running a business. For example, research innovations like Flash Attention or Mixture of Experts help enormously to reduce the cost of inference which directly impacts the profitability of a business and its ability to serve more customers.

Therefore, I will make the following claim about the winners of the AI technological wave:

The greatest AI products will necessarily be built on top of frontier 
AI research breakthroughs.