Nvidia Stock Plunges on CEO Investigation While Google and AMD Expand

The market for artificial intelligence chips has run on a simple assumption. NvidiaNVDA sets the pace, and everyone else follows. That assumption is now under strain.

A boardroom probe meets a selloff

Nvidia co-founder and chief executive Jensen Huang admitted to an extramarital affair in a message to staff. The board opened an independent review into reports that he was targeted by a foreign intelligence effort.

The stock had already fallen 13% through Friday since the first reports surfaced. Monday's decline put Nvidia on course to lose more than $200B in market value if the drop lasted through the session.

The New York Times reported in October that US authorities examined Alice Huang of Singapore-based Megaspeed and concluded she was a Chinese intelligence asset. China's Foreign Ministry spokesperson Guo Jiakun called the spy claims smears against China.

The board said it has no reason to believe national security or company secrets were compromised. Huang stays in the job while the review runs, and Nvidia reports third-quarter results on Nov. 17.

Google ships two more TPU systems

AlphabetGOOGL expanded its eighth-generation tensor processing unit lineup with TPU 8tx for confidential inference and TPU 8i Max for very large context windows. Confidential inference means running AI models while keeping data private. A context window is how much text a model can handle at once.

That brings the company to four TPU systems in 2026, the first time it has introduced that many in a single year.

A tensor processing unit is Google's own AI chip, built in-house. Nearly 100 customers now use them, including leading AI labs, trading firms, retailers, and universities.

Compared with the prior generation, the earlier TPU 8t delivered up to 2.7x better training performance, and the TPU 8i delivered 80% better inference price performance, which is the work done per dollar. TPU 8i Max has twice the computing power, static memory, and high bandwidth memory of TPU 8i, in the same rack space.

TPU 8tx also pairs with AMD's fifth-generation EPYC processors, a detail that ties the two challengers together.

Oppenheimer turns bullish on AMD

Oppenheimer upgraded Advanced Micro DevicesAMD to Outperform from Perform with a $600 target, pointing to momentum in the data center business. Analyst Rick Schafer modeled data center revenue growth of 135% for 2027. AMD was up more than 3% in premarket trading.

That segment would then make up 85% of total revenue. Management is guiding to overall revenue growth above 80% next year, even with limited supply. Schafer expects AMD's AI graphics chip sales to pass its server processor sales in 2027, even as server processor revenue more than doubles.

The demand case rests on customers buying more over time. Schafer sees OpenAI driving AI chip growth through the first quarter, Anthropic adding demand late in that quarter, and the largest cloud buyers scaling up into 2028.

Spending by the five largest cloud providers is expected to reach about $1.3T in 2027, up from $800B this year. Oppenheimer dismissed worries that midterm election results will slow data center construction. It sees little political risk before the second half of 2028.

Buyers gain bargaining power

The disruption looks contained to one company for now. Panmure Liberum investment strategist Joachim Klement said the damage could spread if Huang's job comes under threat, since Nvidia might pull back support for other chip firms.

Dan Ives called talk of replacing Huang "comedy" and kept his view that Nvidia sits two to three years ahead of China in AI chips.

Three things landed in a single session: a boardroom probe, a product launch and an analyst upgrade. That mix is unusual. Buyers weighing AI chips now have a second and third option to test on price performance. They also deal with a supplier whose chief executive faces a request to attend a closed-door House briefing on Nov. 18.