The proletarianization of software engineering
May 11, 2026
- Version française
The world of software development is undergoing an extremely rapid and radical transformation driven by progress in generative artificial intelligence (GenAI). It is a highly divisive topic that generates many more or less far-fetched theories. Still, I believe we can venture to predict the future of the sector by drawing on a clearly documented historical phenomenon: the proletarianization of artisans during the industrial revolution.
I am relying on a few assumptions:
- The use of GenAI will generate considerable productivity gains in the field of software development.
- The costs associated with using (inference) GenAI will rise once tokens stop being subsidized by venture capital.
- The best GenAI models will remain the exclusive property of a handful of providers, the only ones with the required computing capacity.
Assumption #3 is certainly the most uncertain: it is entirely possible that the development of open models we can run locally will undermine my predictions. Either way, it remains very likely that the most powerful models combined with high computing capacity will deliver the greatest productivity gains (assumption #1).
Granting the assumptions above, the consequences have already been clearly described by Marx: it is the proletarianization of the developer profession. And these consequences have been verified many times in the history of capitalism. For example, among artisans during the industrial revolution. In that case, artisans were subjected to the dissociation of labor and capital (the means of production). This transformation was accompanied by the following phenomena:
- Appropriation of labor by capital.
- A considerable increase in productivity.
- An increase in overall profits but a decrease in unit profits.
- Deterioration of wages and working conditions.
- The implementation of the division of labor, which would give rise to Taylorism and then to Fordism (i.e. assembly-line work) during the second industrial revolution. This is what Marx would call "alienation".
- Increased extractivism to meet the rise in production.
Following this parallel, we can predict that today's developers will lose their privileged white-collar status to become workers of the IT industry. In practice, we can expect the following phenomena:
- The companies that survive will be those with privileged access to models and infrastructure.
- The developer profession will shift from expert to low-skilled. For example, we can expect software validation (Q&A) to still be done manually, without any technical skill required.
- A small number of expert developers will play the role of foremen/architects and design the production lines: this is what we already see with the design of "agent harnesses" and agent orchestration.
- Environmental impacts will increase considerably to meet growth requirements.
The fundamental difference with the industrial revolution is the speed at which this revolution is unfolding before our eyes. Considerable changes happen within a few months, whereas the industrial revolution played out over several decades.
Whether this transformation is desirable or not, and what can be done to encourage or prevent it, are left as an exercise for the reader.