Why a Microsoft VP Calls This the Best Moment to Become a Developer

The show opens with the question the whole industry keeps asking itself: as AI writes, debugs, and tests code, what is left for the developer. The host wants the answer from Amanda Silver, the Microsoft executive in charge of developer tools, heading into 2026. The claim is set early: demand for developers remains an open shortage measured in years, not a passing panic about AI.

Silver has spent over twenty years building the tools that millions of engineers rely on every day. Her title is corporate vice president, and her territory is the kitchen where software gets made. That seat lets her watch the AI wave from two sides at once: the model frontier and the working developer's screen.

The turning point, as told here, arrived earlier this year, when foundation models learned to reason step by step. A model can now take an instruction, combine it with tools, and carry a task through to completion. In Silver's telling, that leap multiplied the speed at which agents could take on developer work.

Asked whether learning to code is still worth it going into 2026, the answer comes without hesitation: absolutely. The reasoning points at the critical systems underneath airlines and film studios, which are anything but ordinary software. Agents will not assemble and sustain complexity at that scale alone, so people who grasp it stay necessary. That does not mean humans keep hand-writing every line.

This is where the industry's idea of boilerplate enters: the pattern code that repeats nearly unchanged across projects. Opening a network connection is the example given, a pattern where improvising only invites failure. That repetitive layer, the conversation argues, is exactly what agents now generate for you.

Silver frames it as software's eternal ladder. In the mainframe era every line was written by hand; then came application servers and frameworks, and shared code got reused. Each rung freed developers from pattern work and pushed them toward code only they could write for their own case. AI agents are presented as the ladder's newest rung.

The learning side of the story is the most striking: computer science classrooms have carried waiting lists for years, with too few instructors to meet demand. Silver argues that bottleneck is inverting, because everyone now carries a tireless tutor versed in every subject. Self-taught developers always existed, but the claim is that the path was never this open.

Silver also puts her own graduation story on the table: she finished school as the dot-com bubble burst, holding a fresh computer science degree and no plan. She had watched her brothers live through startup collapses and entered the field frightened. Her lesson from that stretch is that such episodes do not destroy talent; they redistribute it.

That redistribution claim is the spine of the interview: developer demand outside the tech sector, across retail, health, education, government, and finance, is said to run higher than inside it. Sign-ups on GitHub are described as breaking records, the worldwide developer population keeps growing, and wider internet access keeps widening demand. The argument borrows from economics, the Jevons paradox: when supply expands, demand grows with it.

Two skills close the conversation: stating intent with clarity and designing evaluations. How you describe the task to an agent decides the outcome, and then you must gather datasets and build the yardsticks that score the AI. The example runs through social media: views and likes as signals for what counts as engaging. The closing verdict is that the 'coder' title is giving way to the systems thinker who assembles complex wholes from software.