The Next Billion Seconds

It looks as though NVIDIA founder and CEO Jensen Huang is about to be sacked by his board. Not, as it might seem, for having an affair with a purported Chinese spy, Alice Huang (no relation), who diverted NVIDIA chip orders into China, but because of a newly-revealed affair with a subordinate.

As nearly every part of the planet’s most valuable firm reports directly to him, no one knows what this will mean for business, investment, and the forward progress of artificial intelligence.

Is this the most successful intelligence operation in history? Is the US administration implicated? And what happens when the figure at the centre of the AI business – making bank and serving as the industries lender of first resort – steps offstage?

Plus: what the US midterms mean for the data centre buildout, and observations from a recent visit to an AI conference – in China.

Big thanks to Kathryn Ryan and the RNZ Nine to Noon producers for permission to rebroadcast our conversation.

Here are my own notes for the broadcast, with links to all the relevant stories

TOPIC ONE: THE AMERICAN PUBLIC DOESN’T LIKE AI VERY MUCH

There were at least 41 state measures on the ballot last week across the United States, designed to curb the massive US $11T (NZD $16T) buildout of data centres.

I believe only one of them was voted down.

The only town in the U.S. that said yes to data centers – The Washington Post

Of the 41 local data center and AI infrastructure measures across 10 states on the November 3, 2026 midterm ballot, 40 resulted in outcomes that ban, pause, or restrict development, while only 1 favored development.

The Sole Pro-Development Exception (Harwood, North Dakota): The only outcome favoring developers occurred in Harwood, ND, where voters defeated (76% No to 24% Yes) an initiative that would have required additional municipal review before approving industrial developments, including data centers. 

Governors, US representatives and US Senators won election by taking a hard-line stance against big tech, big data centres, and AI.

So in a month’s time the US has gone from “Is AI going to kill everyone” to “Let’s vote in the folks who will pull the plug on all this.”

The reality is that the buildout is already baked in for 2027, but this will affect what happens in 2028 and beyond.

It has also placed it front-and-centre as a big issue in the 2028 US presidential election.

So AI now has a political dimension it never really had before last week.

That’s going to change the tune of the AI firms. Consider this statement from OpenAI, released a day after the vote:

https://openai.com/index/setting-new-standard-responsible-ai-infrastructure

– Pay our own way on energy and infrastructure. We’ll pay the full cost of the electricity we use and the infrastructure needed to serve us. Before major spending begins, we’ll provide financial assurances so other customers are protected if a project is delayed, reduced, or canceled.

– Protect local resources and report on our impact. We’ll develop a water plan for each campus, prioritize closed-loop or low-water cooling, and publish site-level water and electricity use. We’ll publish construction plans and independent noise studies before operations begin and annually afterward. And we will not use non-disclosure agreements (NDAs) with public officials, allowing only narrow protections covering legitimate security information, trade secrets, or proprietary or competitively sensitive information.

– Create opportunities for local workers and businesses. We’ll publish the jobs we expect to create, work with unions and training providers to help residents prepare for them, and report annually on hiring and apprenticeships. We’ll also make sustained investments in priorities residents help select. 

OpenAI and Anthropic are both deeply aware that their social license to operate is under close inspection.

That’s had a huge impact on Anthropic, which expected to go public with a US $2T (NZD $3T) valuation, but that’s been scaled back to around $1.4T. Because of this vote. Because people are worried about AI. Etc.

Anthropic sets IPO price range of $70 to $78, looks to start trading Nov. 19, sources say

So that’s one week away.

Is that the popping of the ‘AI bubble’? Hardly. Demand remains strong. And if it gets harder to build data centres, that may actually increase the profits of the folks providing AI services, since demand will very likely greatly exceed supply.

But where those profits land, and who profits from them – that’s all now up for grabs.

TOPIC TWO: L’AFFAIRE JENSEN

The NYT reports that Jensen Huang, co-founder and CEO of Nvidia, the company at the centre of AI – they make the chips – has confessed to having an affair with an individual identified by US intelligence as a Chinese Spy…

Nvidia Opens Investigation Into Jensen Huang’s Relationship With Chinese Spy – The New York Times

To say this is an earthquake understates it. Yes, it’s the ages-old story of men being unable to keep it in their pants, but it’s also the story of the geopolitical rivalry between the US and China for global domination of artificial intelligence, which both considers to be a race that neither can afford to lose, lest they face being wiped out by the other.

This makes it, dollar for dollar, the most successful intelligence operation in history. China has managed to discredit the most important person in the industry of AI, thrown the US markets and AI sector into turmoil, and created an opening they can use to drive their own sector’s champions into businesses.

This all has a very Matta Hari or Cold War feel to it.

U.S. Cleared Nvidia Chips to Chinese Firms With Empty Offices and Troubling Ties – The New York Times

So that’s gotten way way way​ more complicated.

https://www.theglobeandmail.com/business/commentary/article-nvidias-board-signals-jensen-huang-is-too-big-to-fire

For months, Nvidia has been dogged by allegations that Mr. Huang’s affair – he has been married since 1985 – influenced the company’s chip sales to Megaspeed, along with how Nvidia did business with China.

In May, short seller Culper Research published a 40-page report on Nvidia titled “The China Problem.” The New York-based activist fund pointed out Mr. Huang had “an uncomfortably close relationship” with a Megaspeed executive. Culper also alleged Megaspeed circumvented U.S. trade restrictions by smuggling Nvidia chips to Chinese companies.

Roughly a year ago, U.S. intelligence services told the White House that Mr. Huang’s affair was a potential issue in the country’s relations with China, according to the New York Times. Chinese authorities deny the Megaspeed executive is a government agent who successfully executed what spy novel fans know as a “honey trap.”

VERY. COMPLICATED.

Trump was warned multiple times.  Did the NVIDIA board know? 

There are so many ugly questions to be answered.

U.S. Commerce Dept. Inquiry Into Megaspeed Was Halted After Nvidia Intervention – The New York Times

Timeline of NVIDIA CEO Jensen Huang Affair

Who Is Alice Huang, the Woman in the Middle of Nvidia’s Spy Scandal? – The New York Times

Despite Denial, Trump Was Told on Trip to China About Spying on Nvidia’s Jensen Huang – The New York Times

Made worse by the fact that Jensen has 60 direct reports – NVIDIA reports to him and him alone. If they lose him, do they founder? Will they have a choice?

China panel chair Moolenaar demands answers on Nvidia CEO’s alleged affair with a Chinese spy

Senior Republican on US Senate banking panel calls on Nvidia CEO Huang to resign | Reuters

Trump Officials Clashed Over Megaspeed, Firm at Heart of Nvidia Affair – The New York Times

TOPIC THREE: THE VIEW FROM CHINA

And on that point, spent last week in Shanghai, at AI Engineer Shanghai. The view from inside China isn’t terrifically different from the view outside. Everyone has the same classes of problems, everyone is trying to get agents to work at scale, reliably and safely.

Full program | AI Engineer Shanghai 2026

Everyone is sharing what they’ve learned along the way.

What’s different is that the Chinese aren’t worrying about P(doom) – that’s the US obsession with a poorly-aligned ‘superintelligent’ AI rising up and wiping humanity out.

Instead they’re working on hard problems – like how do we build models for robot brains that allow them to react quickly and flexibly?

The only time AGI got mentioned as in a talk about the future of marketing.

So there’s no doom, plenty of work to do – and, in a first in my experience – as many women as men at an AI conference, all deeply participating in this field.

Big emphasis on open source, and sharing their work.  That’s a mandate from Xi from a few months ago, but it’s been a standard practice amongst Chinese AI firms for almost four years.

So it all feels somehow more normal over there in China. Less fevered. Not at all messianic.

And yet, at the same time, we have Chinese spies seducing American AI industrialists….