🚨 Nvidia May Fire Jensen

Admit it. You’ve wondered, how much money do they make?

Money is our last true taboo. We talk about therapy and bad dates, but guard our bank statements like classified files.

The podcast What We Spend blows the safe wide open.

In every episode, real people share their actual, unedited transaction histories. You'll hear about the secret credit card debt, the reckless impulse buys, and the hidden sacrifices nobody talks about at brunch.

It’s revealing, intimate, and strangely validating to finally hear the truth behind other people’s bank accounts.

Stop wondering. Start listening.

Nvidia's board is reportedly getting ready to fire Jensen Huang.

Yes, that Jensen. Leather jacket. Founder. The man who turned a graphics card company into the AI economy's toll booth.

On Sunday, Huang admitted to an extramarital affair. The other person is reportedly Alice Huang (no relation), an executive at Megaspeed, a data center company based in Singapore.

The problem: U.S. agencies have identified Ms. Huang as a Chinese operative. The Chinese government denies it.

Nvidia's board opened an investigation into whether the 63-year-old CEO was the target of a foreign intelligence effort. Your average Sunday.

Then on Wednesday it got worse.

The New York Times reported Huang may have had a second relationship, with Kimberly Powell, Nvidia's VP of healthcare, who reports directly to him.

Two people who worked at the company say the pair were seen on video being affectionate at a San Jose hotel during an Nvidia conference in March. The Times says it hasn't seen the video itself.

Both are married to other people…

Powell is a senior figure at Nvidia. She joined in 2008, ran the higher education and research business, and helped get Nvidia's GPUs into medical imaging machines.

One bombshell gets you an investigation. Two gets you a board meeting.

Here's the day shareholders had:

A full emotional arc before most people had finished dinner.

According to The Wall Street Journal, the board could meet as soon as today to discuss "a proper path forward."

Someone is measuring the exit door.

The board is weighing whether to name an interim CEO first, then run a wider search. It's already working with an executive search firm.

You don't call a headhunter just to see how things go.

If you own an S&P 500 index fund, you own a decent slice of Nvidia whether you meant to or not.

This is a key-person risk story. Founders who built the company are famously hard to replace, and Huang has been the face of Nvidia's AI run.

None of these reports point to a problem with the chips. The risk sits with leadership: who runs the company, and whether the foreign intelligence probe turns up anything worse.

What to watch:

Until those questions get answers, expect the stock to stay jumpy. Markets cope with bad news. A big unanswered question makes them nervous.

In March 2000, Cisco was worth more than any company on Earth. Within two years it had lost over 80%. Now the same crowd is piled into a few AI giants. See the 10 stocks we’d own before it turns. See the 10 Stocks Free

Cancer, Alzheimer's, heart disease. Gone in under a decade.

That's the forecast from Anthropic $ANT ( 0.0% ) CEO Dario Amodei.

He made it shortly before his company asks Wall Street for a cheque the size of a small country. Funny how that works.

Amodei says cancer, Alzheimer's, heart disease and mental illness will be "relics of the past" in less than a decade.

For scale, that's a shorter timeline than most mortgages.

His full vision: if AI is done right, humanity gets "an age of miracles" that cracks problems we've been stuck on for millennia.

Bold claim. Even by tech CEO standards.

Amodei isn't the only one making big predictions. The whole C-suite of AI is in a prediction arms race.

Robots building robots that build robots.

What could possibly go wrong?

This is the part your wallet should care about: Anthropic could start trading as soon as next week.

It's chasing a fully diluted valuation of about $1.4 trillion.

That's roughly Indonesia's entire annual economy. For a company that hasn't traded a single share yet.

Amodei's pitch to investors is that the company that "gets the risks right" earns society's trust and its "license to operate."

He also says it has to handle the risks while running "a viable business at the same time."

Cure diseases, keep the robots polite, and post decent revenue. All at once. Ambitious menu.

Grand visions and IPOs tend to show up together. Every CEO quoted here runs a company that benefits when investors believe the future arrives fast.

A $1.4T price tag already prices in a lot of miracles. If the timelines slip, the valuation has to be carried by the actual business.

So for anyone eyeing the listing, the real question is how much of that price is product, and how much is promise?

Revolut just cut its AI coding bill by about 90%.

Its 2,000 engineers switched their default coding model from Claude Code and Codex to China's GLM. The bill dropped to roughly a tenth.

Revolut still uses frontier models for the hardest jobs. But for everyday work, cheap won.

And this switch is part of a bigger trend.

In the week to Nov. 5, most Chinese AI baskets went up while US AI stocks fell sharply.

It was the same tech sell-off, and the two countries' AI stocks went in opposite directions.

TS Lombard's Robert Taylor thinks it signals a shift in who profits from AI. His take: China is setting itself up as the world's cheapest supplier of computing power. It's the same playbook it ran with solar and batteries.

Open-weight models now catch up with frontier systems within months. That makes the big labs' supposed three-month lead look pretty thin.

Swapping one model for another is now basically a settings change. You don't have to rebuild the whole system.

September's numbers from OpenRouter:

So usage is moving to the cheap stuff, while most of the money still goes to closed models. For now.

If anyone can build a decent model, the winners are whoever owns distribution and whoever runs compute cheapest. China will struggle to own Western distribution. Cheap compute, though, is exactly what it does best.

China added 543 GW of power capacity in 2025. The US added 53 GW.

That's 10x more electricity capacity. In a single year.

Data-center hubs in Inner Mongolia pay about 4.7 cents per kWh, roughly half the US industrial average.

The chips are catching up too. Chinese chips now match Nvidia on price per dense BF16 teraflop.

Huawei's CloudMatrix 384 delivers 1.7x the compute and 3.6x the memory of Nvidia's GB200 NVL72. The catch is that it uses about 4x the power. That hurts a lot less when your electricity is half price.

Meanwhile, the US is slowing its own build-out. Democrats took both chambers of Congress in an election shaped partly by opposition to data centers.

The governors-elect in Texas (Gina Hinojosa) and Georgia (Keisha Lance Bottoms) both campaigned partly on data-center moratoria. Those are two of the biggest build-out states.

Slower permits could leave hyperscalers paying for grid upgrades. Demand for compute won't disappear, though. Taylor says it just moves, and China is the cheapest option with spare power.

The chip fight is pushing Chinese buyers toward homegrown suppliers:

The result: Nvidia's share of China's AI accelerator shipments fell to 34% in H1 2026, with Huawei close behind at 28%.

Since 2010, solar module prices have fallen 95% and battery pack prices 90%, as China took over 80%+ of manufacturing in both.

Taylor calls the next round the "China Tech Shock." He warns it could mean shrinking profits for Western tech companies and a structurally stronger renminbi.

If your portfolio leans heavily on US AI names, the big question is who keeps the profits once compute gets cheap. Cheap compute is great news for companies that use AI. It's bad news for anyone charging premium prices to supply it.

That's analysis, not advice. But it's worth knowing before the next sell-off.