The K-shaped AI economy and why SMBs have been left behind in the AI age

Stripe built AI minions. Uber has a no-code agent builder. And Ramp has agents running in the background all the time.

Every week, we hear a new story about a large tech company creating their own AI software factory.

But I recently was chatting with a consultant whose clients are outside of the tech industry. They told me how unimpressed they were in how their clients and prospects were using AI.

It’s a stark contrast from what I see in my day to day. None of my clients has developed AGI, but I’m seeing productivity and capability gains.

What is a K-shaped economy?

The term “K-shaped economy” has been used the last few years to describe wealth inequality, the idea being the rich get richer and lower-income households struggle further.

Cristian deRitis of Moody’s Analytics told Yahoo Finance last year that “the top 10% of households by income account for about half of all the spending in the U.S. economy”.

The K-shaped AI economy

We have a K-shaped AI economy in that there is tremendous inequality in value generated from AI tools.

Stripe, Ramp, and Uber have all publicly shared wins from using coding agents at scale. It makes sense that these big companies with respected engineering orgs will have success.

But very few companies are Stripe, Ramp, or Uber. Plenty of companies just do some basic ChatGPT searches in the browser or maybe nothing at all. SMBs in industries like real estate, construction, and accounting likely feel left behind.

How can SMBs and non-tech businesses catch up?

Plenty of startups are trying to create the “AI software factory as a product”. As of this writing, Warp’s is in early access. This could potentially be a great value to solo founders, technical or not, and small-to-midsize startups. As long as the growth and speed is worth the cost, startups will pay the factory and token cost.

But a lot of SMBs don’t need a software factory. They have manual, inefficient processes. Putting a software factory on top doesn’t necessarily fix it.

These businesses will benefit from learning the patterns of how to work with AI. Then, once they understand the desired end state, they can prepare actionable plans and start building (themselves or with the help of consultants).

Need help taking the first step? Let’s chat.