Jensen Huang says wait two years for AI-native grads; Micron is already screening for them

In an episode of The Ezra Klein Show released September 23, Klein asked Nvidia CEO Jensen Huang what AI means for young engineers entering the field. "Wait two years," Huang said. "Because it takes four years to go to college. The mean time to graduation of this new technology is two years away." After that, he predicted, a wave of "AI-native new grads" would arrive.

It was a hopeful answer to an anxious question. Stanford researchers using ADP payroll data found that employment of 22- to 25-year-olds in the most AI-exposed occupations fell about 11% between November 2022 and June 2026, mostly because companies hired fewer young workers.

LinkUp's own data shows the same squeeze from the hiring side. Across the employers we track, active entry-level job listings peaked near 2.7 million in early 2022. By August 2026 they were around 600,000, down roughly 78%. The first drop, in 2023 after ChatGPT's release, reversed, and listings recovered through 2024. The decline that began in late 2024 has not. Removals never spiked along the way. Employers didn't cut entry-level jobs in waves; they stopped opening new ones.

Micron offers a look at what one large employer is doing against that backdrop. Using LinkUp's data, we analyzed 1,000 of Micron's most recent job postings, all published between September 1 and 26.

Still Hiring at the entry level 

Micron isn't waiting two years. Nearly one in five of those postings was for an intern or a new graduate, and in Boise, home of its new Idaho fab, it was about one in four. Three-quarters of the entry-level roles are fab and plant jobs: process, equipment, test, manufacturing and facilities. In Taichung, Taiwan, Micron posted 20 new-grad engineering roles for its fiscal 2028 hiring class, which starts in fall 2027. That is the class Huang's two-year clock points to.

The demand comes from Micron's build-out. Its job listings more than doubled between January and June as it staffed new plants in Idaho, Taiwan, and Singapore. On September 30, Micron reported record quarterly revenue of $54.2 billion and said it expects memory supply and demand to be "much tighter in calendar 2027 and 2028 than they were in 2026.

A different entry-level job  

What Micron asks of new hires has changed. Its intern and new-grad postings mention large language models, or tools like Copilot, Claude and ChatGPT, 37% of the time. Its experienced roles do so 20% of the time. None of its 21 director-level postings mention them. The gap holds within the same kinds of jobs: in fab and plant roles, 35% of entry-level postings mention these tools, compared with 16% of experienced ones.

Nearly a third of Micron's internship postings make LLM familiarity a requirement. A data-center SSD firmware internship lists "familiarity with AI-enabled tools and large language models (LLMs) such as Copilot, Claude, or similar technologies" among its minimum qualifications. A process-engineering internship at the new Idaho fab asks interns to "leverage Artificial Intelligence and data analytics tools" to "optimize wafer processing, and enhance manufacturing performance."

Micron isn't replacing its entry-level roles with AI. It is filling them with people expected to use AI from day one, and it asks that of new hires more often than of its veterans. Universities are moving more slowly. Ohio State and Purdue have made AI fluency a graduation requirement, starting with the classes of 2029 and 2030.