|
Hey Reader, Last week, Nvidia announced that Jensen Huang had resigned as CEO of the company he co-founded 33 years ago. That company makes the chips that are central to the AI revolution. This came after the Wall Street Journal reported that Nvidia’s board was preparing to oust Huang over two affairs: one with a businesswoman U.S. officials suspect of spying for China, and another with an Nvidia executive who reported to him. Sadly, it reads like a tabloid story and certainly doesn’t help with the general vibe that Big Tech is run by people who can’t be trusted. Today’s newsletter will get you caught up, including on the biggest policy fight at the center of all this – Should Nvidia be allowed to sell AI chips to China at all? 🤔 That’s the question behind the scandal. In one corner… Nvidia wants to make more money by selling more chips. Some people in the United States think that’s fine. And the counterpoint… Others view these chips as military technology and believe they should be restricted in a way that reflects how powerful they are. And I think that’s the key: yes, you can do a million things with the AI chips Nvidia makes… but they’re also so powerful that “military tech” is a reasonable way to think about them. And that’s where the (suspected) spying comes in. (By the way, I’ll avoid saying alleged here because nobody in authority has really alleged that any specific crime was committed… it’s mostly just embarrassing and scandalous.) A chip smuggling network was (reportedly) centered around a woman with whom the Nvidia CEO was having an affair. 😬Jensen Huang admitted to having an affair with Alice Huang (no relation), a former executive at Megaspeed, a Singapore company that was a major Nvidia customer. U.S. officials put Alice under surveillance because they suspected her of smuggling Nvidia chips to China, which is how the affair came to light. The Information reported that a network of companies tied to her and her family helped move about 240,000 export-restricted chips, worth roughly $7.5 billion, into China. For scale, Meta trained its biggest open AI model on about 16,000 Nvidia chips. The caveats: Singapore police found insufficient evidence that Alice committed offenses there, though they barred her from the country for giving them inaccurate information. And U.S. officials don’t believe Jensen knew she was reportedly passing intelligence to Beijing. Here’s how I’d break down both sides of the debate: The case for selling chips to ChinaIn 1980, Jimmy Carter cut off grain sales to the Soviet Union. The Soviets turned to Argentina and other suppliers, American farmers took the hit, and Reagan lifted the embargo in 1981. The sellers say chips will go the same way:
If you believe those three points, the embargo is a waste, and only holds the American economy back. The case against: this is about more than moneyThe people who want restrictions aren’t arguing about Nvidia’s revenue. They’re arguing that AI chips are dangerous in the wrong hands.
If you believe AI chips are more like weapons than they are like laptops, a few lost sales are a fair price for keeping them out of the wrong hands. My takeThe Nvidia story is about balancing the well-being of a big American company (and America’s stock market and GDP) with national security. Lockheed Martin sells F-35s to our allies, but not our adversaries, and everyone agrees that is a reasonable thing to do. The weird thing about Nvidia is they don’t think their chips are military tech (or at least they don’t say so out loud). They think of them more like laptops or iPhones, and they are frustrated that they’re not allowed to export as much as they want. That puts Nvidia in the awkward position of having to argue that their chips are less powerful so they’re allowed to sell them to more people. 🤷♂️ I think it’s better to be safe than sorry, and if that reduces Nvidia’s sales a bit in the interim, that’s probably a sacrifice the American government should be willing to make. Until next time, Rob Howard 🤩 New: The AI Executive System for Teams More than 100 executives have gone through our AI Executive System program. Now we’re bringing it to whole teams. We learn how each person works, build done-for-you AI agents around their job (two for every team member, 10 for each sponsoring leader) and train every team to use them, with Demo Day in 90 days. No technical experience needed, and it comes with a 60-day make-it-right or money-back guarantee. |
Coaching, community & curriculum to help everyone thrive in our AI‑powered future.