Nvidia Chief Executive Jensen Huang has put the spotlight on a new phrase gaining ground in artificial intelligence circles: loop engineering. The idea has sparked discussion after a post on X attributed to Huang the quote, "Nobody writes prompts anymore. The new job is to write and handle loops," describing it as a shift that could define the rest of 2026.
In an interview with the Associated Press earlier this week, Huang repeatedly described a future in which AI moves beyond simple prompts and becomes a system that searches, evaluates, reasons, uses tools and improves through repeated cycles.
NVIDIA CEO, Jensen Huang:
— darkzodchi (@zodchiii) June 18, 2026
"Nobody writes prompts anymore. The new job is to write and handle loops."
He calls it the shift that defines the rest of 2026.
Interview was out just yesterday.
Watch the 23 minute talk, then save the full framework below👇 https://t.co/9eSXpqLQgX pic.twitter.com/Kw1GZEKPjD
At the centre of the idea is a simple distinction. Prompt engineering is about writing a better instruction to an AI model. Loop engineering is about designing the process that happens after the instruction is given.
What Is Loop Engineering?
Loop engineering is a closed cycle. An AI system generates a hypothesis, tests it, scores the result against a clear objective, reads why it failed and feeds that feedback into the next generation. The pattern is to perceive, reason, act, observe and repeat.
Each pass is relatively cheap, but each result narrows the search. That is why loop engineering can turn dozens of weak or average attempts into one output that is stronger, tested and more reliable. The value is not just in the model's first answer, but in the process that forces the model to improve.
ALSO READ: Marvell Technology Shares Rocket 26% After Nvidia CEO's 'Trillion-Dollar' Projection
Why Learn Loop Engineering?
Huang pointed to this direction when he said future AI would be "less guessing and more research". He explained that even current AI systems can use the internet, search, return "two or three different versions" of information and then evaluate which version is more likely to be truthful.
That is a loop in action. The AI does not simply respond. It searches, compares, evaluates and refines. In another example, Huang described an AI system responding to a prompt by reading documents, chasing references, becoming grounded in a topic and reasoning through how to solve the problem.
This is why loop engineering is increasingly being seen as more important than prompt engineering. A clever prompt may improve a single answer. A well-designed loop can improve the entire workflow, believes Heung.
ALSO READ: Nvidia CEO Says Selloff In Tech Stocks Is A Buying Opportunity
Essential Business Intelligence, Sharp Market Insights, Practical Personal Finance Advice, Daily Fuel, Gold and Silver Prices and Latest Stories — On NDTV Profit.
