Why the 'software eats everything' dream has hit a wall

Back in 2011, venture capitalist Marc Andreessen made one of tech's most famous predictions: software would eventually devour every industry. Fifteen years later, it's worth asking whether the meal is almost over - or whether the world has been a tougher bite than anyone expected.

Writing for Fast Company, the argument is straightforward. Yes, the so-called Magnificent 7 tech companies now dominate the S&P 500 with valuations that would have seemed fictional a generation ago. AI startups like Anthropic and OpenAI are worth hundreds of billions before they've turned a meaningful profit. On paper, software's conquest looks pretty complete.

But the physical world doesn't care about your roadmap

Here's the problem: most of what actually makes civilisation run - housing, food, healthcare, manufacturing, energy - stubbornly refuses to be disrupted by a good algorithm. You can't ship a hospital in an app update. You can't scale a wheat harvest the way you scale a server farm. The real world has friction, regulations, labour, weather, and physics, none of which respond well to a 10x growth target.

This matters because it reframes how we should think about tech's actual role in our lives. Software has genuinely transformed communication, entertainment, finance, and retail. But transformation isn't the same as domination. The industries that involve atoms rather than bits have largely held their ground - and often pushed back harder than expected.

What this means for the next wave of tech ambition

The current AI boom is, in many ways, Andreessen's thesis on steroids. The promise is that large language models and automation will finally crack the industries that resisted the first wave of digital disruption. Maybe. But the physical constraints haven't gone anywhere. Building a robot that can reliably do what a plumber does is a genuinely hard problem - not a software problem.

None of this means technology isn't powerful or important. It clearly is. But there's a useful reality check hiding inside the argument: the companies and founders who understand the limits of software - who know when to pair it with deep physical-world expertise - are probably better positioned than those who assume code alone can solve everything.

Andreessen was right that software would become enormously valuable. He may have underestimated just how much world there is left to eat.