Software Is Still Eating the World — Just Differently

In 2011, Marc Andreessen declared that software was eating the world. He pointed to Netflix consuming Blockbuster, Amazon devouring retail, Google disrupting journalism. It was a provocative thesis about a tectonic shift already underway.

He was right. But the software eating the world of 2024 looks nothing like what anyone expected. And understanding the difference matters enormously for builders, investors, and anyone trying to make sense of the present moment.

The First Wave: Distribution and Aggregation

The first wave of software eating was mostly about distribution. The marginal cost of distributing bits approached zero, while the marginal cost of physical distribution stayed high. This let software companies aggregate demand at scales previously impossible.

Uber aggregated taxi demand. Airbnb aggregated hospitality supply. Spotify aggregated music consumption. The common thread: software intermediating a market that previously relied on physical coordination, extracting value by owning the chokepoint between supply and demand.

The economics were extraordinary. Once you paid for the platform, adding the next user was nearly free. Network effects compounded. Winners took most or all of the market. The result was trillion-dollar companies built by surprisingly small teams.

The Second Wave: The API Economy

The second wave was about composability. AWS, Stripe, Twilio, and dozens of infrastructure companies abstracted away the hardest parts of building software. You no longer needed to build your own servers, payment processing, or SMS delivery. You called an API.

This compressed development timelines dramatically. A startup that once needed 18 months and $5 million to get to a working product could now move in 3 months on $300,000. The question shifted from "can we build this?" to "should we build this?" — a profound change that turbocharged entrepreneurship.

It also created a new kind of dependency. The "API economy" made it trivially easy to build on top of others infrastructure, but with that came new risks: pricing changes, deprecations, and concentration. Today a handful of cloud providers and API companies underpin enormous portions of the global digital economy.

The Third Wave: Intelligence as a Service

We are in the early innings of a third wave, and it is more disruptive than either of the first two. The difference is that this wave is eating knowledge work itself, not just distribution or development.

When OpenAI released GPT-3, developers quickly discovered they could replace tasks that had previously required human expertise — writing, summarization, code generation, data extraction. When GPT-4 and Claude arrived, the range expanded dramatically. Today these models can pass bar exams, ace medical licensing tests, and generate publication-quality research.

This matters because knowledge work is where most of the value in modern economies resides. Manufacturing and agriculture were already largely automated; what remained was the work of minds. Software now eats that too.

What Changes, What Does Not

The most important insight here is about where human value concentrates as software eats more. In the first wave, it concentrated in platform ownership. In the second wave, in the ability to deploy platforms at speed. In the third wave, it will concentrate in judgment — the capacity to ask the right questions, evaluate outputs, and apply knowledge in contexts that machines do not handle well.

This is not a comfortable answer for anyone hoping for a clear rule about what jobs are safe. Judgment is diffuse. It lives in domain expertise, in long-accumulated intuitions, in the ability to read a room or sense when something is wrong before you can articulate why. These capabilities exist unevenly across the workforce.

The historical pattern, though, is instructive. Agricultural automation concentrated value in logistics, manufacturing, and services. Industrial automation concentrated value in knowledge work. Cognitive automation will likely concentrate value in some new category we cannot fully see yet — but it will come.

Building in the Third Wave

For builders, the third wave changes the calculus in specific ways. The moat of technical complexity is eroding rapidly — what required a team of senior engineers last year might require a sharp generalist and an API call today. New moats emerge around data, relationships, brand trust, and the speed of insight-to-product iteration.

The builders who thrive will not be the ones who master a single tool or framework but those who remain curious, who learn quickly, who can synthesize across domains. They will use AI as leverage but will not mistake leverage for insight.

Andreessen thesis holds. Software is still eating the world. It is just now eating the parts of the world that software, for the first thirty years of the internet, could not touch.