Ever since I discovered computers 55 years ago, I’ve thoroughly enjoyed exploring firmware design, compiler theory, and high level architecture, both as a salaried employee and on my own time. For me, software is the delicious intersection of art, architecture, and high-tech tools.
The recent advances in AI technology supports my life-long fantasy that someday, maybe in my lifetime, Isaac Asimov’s sci-fi robots might just be a real thing. But recent developments have also cast a dark shadow over independent software development.
Before the recent AI revolution, many kinds of software development was globally democratic. A lone programmer with low-end hardware in a developing country could produce many kinds of software that was competitive with software produced by a programmer working for a top high-tech company. The cost to the lone programmer was time; the cost to the high-tech company was the price of labor for the same amount of time.
But now, the cost to high-tech companies is split between labor and the cost of semi-autonomous coding platforms:
“[Nvidia CEO Jensen] Huang suggested evaluating engineers based on their AI token usage rather than lines of code written. For a $500,000 engineer, he expects at least $250,000 in token consumption annually…” [1]
An independent programmer in a developing country faces a new kind of exclusion. A lone programmer living on a few hundred dollars a month might not be able to afford even the lowest tier of AI assistance. Free local LLM models exist, but they don’t work well for even modest sized projects, and our lone programmer might not be able to afford the required hardware anyway.
A high tech company can now buy a force multiplier for their employees that time alone cannot compete with.
This has tilted the playing field toward capital. The once-sufficient resource of time has been sharply devalued. The tools are now in the hands of the rich, out of reach of the lone programmer.
Maybe, over time, the gap will narrow with new technology, or new platforms may subsidize access. But for now, software is not the democratic endeavor it used to be. The excitement I felt from bringing competitive software to life on my own meager hardware may be out of reach for many young independent programmers.
[1] MLQ.ai, “Nvidia CEO Proposes AI Tokens as Major Bonus for Engineer Productivity Boost.” 2026. https://mlq.ai/news/nvidia-ceo-proposes-ai-tokens-as-major-bonus-for-engineer-productivity-boost/
Leave a Reply