Traditional Coding Is Dying; Software Engineering Is Not
Traditional coding is the manual process of writing and debugging low-level syntax line by line, while modern AI agents in software engineering are systems that can plan, reason, and execute multi-step development tasks so that engineers focus on higher-level design, safety, and orchestration instead of routine code typing.
That is the uncomfortable truth: if your professional identity is tied to hand-writing every function, you are training for a role that AI is rapidly absorbing. Nvidia CEO Jensen Huang is blunt about it, arguing that the manual process of writing syntax like Python is being replaced by agentic AI. In an interview released by the company, he said his engineers are gladly leaving traditional coding behind, because “coding is just typing now.” The work is not disappearing; it is moving up the stack. The future of coding jobs belongs to people who can tell powerful systems what to do, not those who compete with them line by line.
From Code Typists to AI Agent Architects
Nvidia is already living the next phase of AI agents software engineering. In an interview published by the company on Wednesday, Huang said engineers are embracing AI because it lets them focus on more creative and higher-value work. Instead of spending most of their time writing code line by line, they are designing AI systems capable of carrying out complex tasks autonomously. These AI agents can plan, reason, and execute multi-step tasks by breaking larger objectives into smaller actions.
According to Huang, engineers now spend less time on routine programming and more time developing AI agents, creating benchmarks to evaluate performance, and building guardrails so systems operate safely and reliably. “You’re taking all the mundane work, and you’re trying to get this agent to do it,” he said, adding that the work requires imagination, creativity, and a lot of technology. This is the software engineer career shift in practice: from code writers to AI agent architects who define goals, constraints, and evaluation criteria, then let the agents execute.
The Job Market: More AI, More Engineers, Different Skills
The popular fear is that if AI writes the code, the future of coding jobs vanishes. The numbers tell a different story. The global developer population has grown from roughly 5 million in 2010 to an estimated 28.7 million today and is projected to reach 45 million by 2030. In the U.S., the Bureau of Labor Statistics expects employment for software developers to grow 17% through 2033. That growth is happening alongside exploding interest in agentic AI, not despite it.
This matches a basic economic pattern: when a resource becomes cheaper and more efficient, demand often explodes. AI coding assistants lower the cost of producing software, which increases demand for people who can design systems and orchestrate AI agents. Anthropic illustrates this: despite its CEO Dario Amodei predicting that raw programming is a shrinking skillset, the company is hiring for more than 400 roles, with some engineering positions offering salaries of up to USD 405,000 (approx. RM1,863,000). Exposure to agentic AI is correlating with job posting growth, not collapse.
What Engineers Must Learn Instead of Routine Coding
If coding is “being done by the AI models first,” as Amodei puts it, what should engineers master to stay competitive? The answer is not giving up on code, but changing your center of gravity. As AI coding assistants grow more capable at generating routine code, engineers are expected to spend more time defining problems, designing AI workflows, validating outputs, establishing safety guardrails, and integrating autonomous agents into business operations.
That means building AI agent design skills: specifying agent roles, tools, and boundaries; constructing evaluation frameworks; and thinking in terms of systems rather than individual functions. It also means prompt engineering in the practical sense—turning business goals into precise, testable instructions for agents—and AI system architecture: how agents call each other, how they interact with data, and how failures are contained. Anthropic’s own engineers “don’t really write code the same way anymore… They let Claude write it. They edit. They review. They architect.” Editing, auditing, and designing around AI is becoming the new core craft.
Evolution, Not Extinction, of Software Engineering
This shift is not the end of software engineering; it is its next stage. Huang argues that AI is expanding the role of engineers rather than making them obsolete, pushing back against fears of widespread white-collar job losses. In Huang’s view, the amount of work required to bring AI into the world is “really quite incredible,” and it is creating a whole bunch of jobs that his software engineers actually prefer.
Agentic computing is set to define the rest of 2026, as interconnected AI workflows demand new infrastructure and system thinking. Tech leaders expect AI-assisted development to reshape virtually every application businesses rely on today. For engineers, the choice is clear: stay attached to tasks AI is learning to automate, or move toward the design, evaluation, and governance of the agents themselves. AI is not the enemy of software engineers; it is the new medium in which their best work will be written.
