Cobus Greyling argues that software is undergoing a historic collapse in cost and complexity similar to previous waves (storage, CPUs, bandwidth), fundamentally transforming software production and the development industry.
No AI bubble according to Marc Andreessen
Unlike the dot-com bubble, where 97% of deployed fiber optic cable remained "dark" (without demand), current AI faces no demand drought—"there are no dark GPUs". Tech waves surge when the expensive becomes cheap enough to be squandered. AI has reached that point: people "waste" AI on everything, from meme generators to code tweaks, creating "Work Slop" and "AI Slop" (low-value outputs flooding workflows). The intuitive interface (natural language) makes AI massively accessible, triggering explosive consumption.
History of technology cost collapses
The article recalls Gmail 2004 (1GB free vs. 4MB for competitors), dial-up where every byte was precious vs. today's background 4K streaming, CPUs evolving from room-sized mainframes to pocket supercomputers. Now it's software's turn.
Software couldn't "eat the world"
Marc Andreessen declared in 2011 that "Software is eating the world," but for decades, the astronomical complexity and cost of software development held back production, accumulating massive societal technical debt. Software could not truly "eat the world" as long as it required scarce, costly resources like skilled programmers to build even basic applications.
The current collapse
Today, software "devours everything" precisely because its cost and complexity are collapsing. Generative AI tools boost developer productivity by up to 55%, drastically reducing development cycles and costs for everything from prototypes to production code. As content has become "permissionless" (YouTube, blogs, Twitter enabling creation without barriers), software is also becoming permissionless. "Content is now an application; apps are now content"—the lines are completely blurring.
Enterprise vs. consumer divergence
Greyling qualifies this, however: this transformation will play out differently within enterprises. Legacy systems, compliance obstacles, and scale requirements will not disappear instantly. "AI might democratise the edges, but the core stays gated" in an enterprise context. Concrete evidence: while permissionless consumer AI thrives via decentralized platforms, 95% of enterprise generative AI pilots fail to reach production due to security and governance gaps.
Employment consequences
The ADP Research chart shows a steady decline in US software developer employment since 2018 (peak ~112% in January 2020, falling to ~82% in January 2024), suggesting a deep industry transformation already underway.
Gutenberg moment
The article frames this collapse as "Software's Gutenberg Moment"—a revolution comparable to the invention of the printing press, democratizing software creation in an unprecedented way. The societal technical debt accumulated over decades could finally be resolved thanks to this new accessibility, although the transition creates major disruption for traditional professionals.