Amichai Mantinband argues that developers should not invest in becoming deeply specialized in .NET or C# purely for its own sake in 2026. The video frames that answer around what AI agents can already do well, what they still struggle with, and where human developers can retain leverage as software work moves up a level of abstraction.
flowchart TD
A[Should developers learn .NET in 2026?] --> B[Agents are strong at code execution]
A --> C[Agents still struggle with full-codebase context]
A --> D[Agents struggle with novel product and architecture ideas]
B --> E[.NET-specific syntax and implementation skill loses value]
C --> F[Developers gain leverage through system understanding]
D --> F
F --> G[Invest in architecture, cloud, and domain judgment]
- Context and framing
- What agents already do well
- Where agents still struggle
- Why .NET-specific expertise loses value
- Where developers should invest instead
Key takeaways
- The video's final answer is that developers should not aim to become highly proficient in .NET or C# in advance unless their current job specifically requires that depth.
14:0514:39 - Mantinband's main reason is that AI agents are becoming very strong at writing code in any language, so language/framework-specific implementation work is becoming less important than deciding what should be built.
1:497:57 - He identifies large production codebase context as the clearest remaining human advantage, especially for codebases around 300,000+ lines where model context limits still matter.
3:234:28 - He expects agent vision and line-by-line code review limitations to improve quickly, so he does not see them as durable areas of developer leverage.
4:576:19 - The recommended career investment is software architecture, system interaction, cloud architecture, and job-specific low-level knowledge rather than general-purpose .NET specialization.
14:0514:39