Happy New Year and My Predictions for 2026

Wishing all readers a successful new year. May 2026 be productive and leave room for interesting projects.

My predictions: Agentic coding won’t make big technological leaps, but integrations will improve dramatically. The real shift is happening elsewhere. Existing code tools were built for humans. Next year, we’ll see the first tools built exclusively for LLMs. This will allow models to navigate codebases more precisely than with conventional tools.

AI-assisted coding will become more reliable, and I expect fewer people will bother learning to code from scratch. This sounds like a loss, but paradoxically drives up the value of genuine programming skills. Those who truly understand what’s happening under the hood will be more sought after than ever.

By the end of the year, businesses will show more willingness to start software projects again. This time with AI support as an integral part from the beginning, not as an afterthought.

Sudden mass unemployment among software developers won’t happen, at least not in Germany. The skills shortage is real, as is demographic change. AI will make developers more productive, not obsolete. Demand for software is growing faster than AI can meet it.

The AI bubble won’t burst. Unlike crypto, there are real productivity gains here that companies can already measure. Valuations of individual startups may be inflated, but the technology itself delivers.