NVIDIA’s CEO delivered the keynote at CMU’s 128th commencement on Sunday and received an honorary doctorate. The address framed AI as a reindustrialisation moment for the US, and pressed both engineers and policymakers to advance capability and safety in step.

NVIDIA founder and CEO Jensen Huang delivered the keynote address at Carnegie Mellon University’s 128th commencement on Sunday morning, telling graduates that they were entering the workforce at the beginning of the largest computing-platform shift in history.

“I cannot imagine a more exciting time to begin your life’s work,” he said.

The setting suited the argument. Carnegie Mellon’s School of Computer Science created the Logic Theorist in the 1950s, widely regarded as the first artificial-intelligence programme, and founded the world’s first academic Robotics Institute in 1979.

Huang made the lineage explicit.

“AI started right here at Carnegie Mellon,” he told graduates assembled in the rain at Gesling Stadium. Huang also received an honorary Doctor of Science and Technology from CMU President Farnam Jahanian, one of the university’s highest distinctions.

The core of the speech was framed around four imperatives that Nvidia’s CEO has repeated across several recent venues, applied here to the graduates’ own choices.

“Advance safely. Create thoughtful policies. Make AI broadly accessible. And encourage everyone to engage.”

The framing was meant to land on a campus that has produced significant work on both AI capability and AI safety, including in fields where the two cannot be separated cleanly.

“Scientists and engineers,” he said, “have a profound responsibility to advance AI capabilities and AI safety together.”

Huang then placed AI within a longer arc of US industrial history. The technology, he argued, is “a once-in-a-generation opportunity to reindustrialize America and restore the nation’s capacity to build.”

The framing aligned with comments he has made in recent earnings calls and at industry events, but Huang sharpened it for the graduates by widening the addressable workforce.

AI’s benefits, he said, would reach “electricians, plumbers, ironworkers, technicians and all kinds of builders” alongside the technical roles graduates of the host institution are most likely to fill.

That distinction has become more central to Nvidia’s public communication this year, as US political concern about the AI labour-market effects has intensified.

On the question of work itself, Huang drew a careful line between the task and the purpose of a job. “Radiologists,” he said, “don’t just read scans.

They care for patients. AI automates scan reading (the task) but elevates the radiologist: the purpose.” The construction is one Huang has used repeatedly in recent quarters and represents Nvidia’s reply to the AI-replaces-workers narrative. It is also, in its specifics, easier to assert than to demonstrate.

The radiology example holds in some clinical settings and fails in others, where the bottleneck is in fact the diagnostic interpretation step.

Huang did not engage with that complexity, which is appropriate for a commencement address but worth noting for readers calibrating how to interpret the framing in other contexts.

Huang addressed the political climate around AI directly, although in the carefully diplomatic register that a commencement allows. “Every major technological revolution in history created fear alongside opportunity,” he said.

“When society engages technology openly, responsibly, and optimistically, we expand human potential far more than we diminish it.”

He pressed policymakers in the same terms used by Anthropic, OpenAI and Microsoft over the past year: “Policymakers have a responsibility to create thoughtful guardrails that protect society while still allowing innovation, discovery, and progress to move forward.”

Whether the current US guardrail debate matches that standard is a separate question Huang did not address.

The address was personal in places. Huang spoke about his arrival in the United States as a first-generation immigrant, his parents’ bet on American opportunity, and what he described as a country “not easy, but full of opportunities. Not a guarantee, but a chance.”

Then a line that read as half-rhetoric and half-credo: “How can we not be romantic about America?”

The phrasing has been a recurring motif in Huang’s speeches and tends to land particularly well with audiences carrying immigration stories of their own.

It also fits the moment in which Nvidia, having committed Nvidia’s $40bn of AI equity bets so far this year, has every commercial interest in framing the company’s success as American national renewal as much as private sector achievement.

The closing was tighter. Huang invoked Carnegie Mellon’s institutional motto and finished with an instruction.

“My heart is in the work. So put your heart into the work. Build something worthy of your education, your potential, and the people who believed in you long before the world did.”

He then waved to graduates on the way off-stage, with several thousand smartphones lifted in the foreground.

Huang’s commencement-address mode is markedly different from his keynote mode. He has appeared at GTC, Computex, and Davos in the past twelve months, delivering versions of the same five-layer-AI-infrastructure framing, with significantly more theatre and a fondness for product demos that has, in places, included a CGI version of himself.

Earlier Huang appearances that leaned harder on theatre have been part of the recent Nvidia presentation language. The CMU speech traded all of that for a more direct register, which is the convention that the venue invited.

The arguments remained largely the same. Only the production values changed.

For graduates, the practical implications are useful to set out. The world of work most CMU computer-science graduates are entering is structurally different from the one their predecessors graduated into, even three years ago.

The first US undergraduate AI degree, which CMU was the first US institution to offer in 2018, has produced its first full cohorts. Job descriptions in machine-learning engineering, alignment research, AI infrastructure operations, and applied AI products are now established categories rather than emerging ones.

Hiring across them has been uneven over the past eighteen months, with senior roles in heavy demand and entry-level roles considerably tighter. Huang’s framing of an open beginning is correct as a long-run statement; the near-term picture is more competitive than his address acknowledged.

The broader question Huang’s address raised, but did not need to answer, is what the AI build-out demands of the institutions that train the people who staff it.

Carnegie Mellon, MIT, Stanford, and the leading European institutions are all reconfiguring curricula around the technologies that have moved fastest since 2022. 

Whether that reconfiguration produces the kind of engineers who can advance capability and safety in the integrated way Huang described is partly a question of programme design and partly a question of the labour market the graduates enter.

NVIDIA is, on its present trajectory, the dominant private-sector force shaping that labour market.

That places a commencement address by Huang somewhere between a benediction and a business communication.

The students he addressed will, in many cases, work directly on Nvidia infrastructure or for companies whose own existence depends on Nvidia silicon and capital.

The keynote was generous about their prospects and clear about the responsibilities those prospects carry. It is, in its way, a fair description of where this generation of engineers actually finds itself: at the start of a technological era whose dominant company can credibly tell them that their career begins as the era does.

The full video of Huang’s address is available on Nvidia’s official blog post about the event. Carnegie Mellon graduated its 128th class on Sunday afternoon.