...
By Didi Kirsten Tatlow and John Feng
0
Share
Newsweek is a Trust Project member
See more of our trusted coverage when you search.Prefer Newsweek on Googleto see more of our trusted coverage when you search.
China Decoded is Newsweek’s in-depth newsletter about China’s growing influence on the world, written by Senior Correspondent for International Affairs and Investigations Didi Kirsten Tatlow and China Editor John Feng. It goes beyond the headlines to examine how Beijing thinks, how its policies travel beyond China’s borders, and what they mean for the United States and its allies. Readers can sign up for China Decoded to receive each edition directly in their inbox.
Here, for Newsweek Members, is the latest edition of China Decoded in full:
Careful With American AI Now
Something is really bothering me about the semiconductors and sex scandal that is the “Huang Affair.” That’s the downfall of the Nvidia co-founder and former CEO, Jensen Huang, after an alleged affair with a mysterious semiconductor trader and possible Chinese espionage asset, Alice Huang.
Read More on Analysis
It’s Megaspeed’s money. Essentially, the Singapore-based company that was headed by Ms. Huang went from having assets of $75 in 2022 to assets of nearly $3 billion in 2024 according to Singapore government records, cited by Hong Kong media.
OK, the company is called Megaspeed, and nomen est omen; a name is a sign, as the Latin phrase goes.
Megaspeed has said it has done nothing wrong. Efforts to contact them for comment have failed. The Singapore authorities are investigating the company. I highly recommend reading this statement from Singapore’s Ministry of Home Affairs for more clarity.
But if we are talking about a sting—run by whom is unclear—then the financial pyrotechnics are just really bad tradecraft, by which I mean the methods used by intelligence to carry out operations.
Speed and numbers like that attract attention. People ask questions. As a player in the GPU tech scene (GPUs are in chips, in fact they are a kind of chip) remarked to me: “The Israelis would never do it so badly.” They have a big tech ecosystem and big national security concerns, too.
Alice Huang was Megaspeed director for just eight months, from March 2023 to November 2023. There was more continuity; the company evolved from another one that had been incorporated in Singapore in 2015, and Ms. Huang may have stayed involved in it after leaving her position. But the brevity of her tenure is striking.
Here are the full numbers: In 2022, Megaspeed’s assets were worth $75. In 2023 they were worth $32.9 million. In 2024 they surged 90-fold, to $2.98 billion. The firm’s revenue rose from almost zero in 2023 to $294 million in 2024, the South China Morning Post of Hong Kong reported.
There are so many questions that need answering. Such as, why Megaspeed was passing through Nvidia chips to China when the Chinese government was impounding hundreds of thousands of them in Hong Kong, as Reuters reported on Nov. 5.
Beijing is smart. It doesn’t want to be dependent on foreign chip supplies. Chips are traceable and can be hacked.
Ms. Huang may have been serving different interests simultaneously; getting advanced processors in China and acting as an informant for the state. She landed just one or two steps from President Donald Trump via her closeness to Mr. Huang, who regularly met or telephoned with Trump.
As CEO, Mr. Huang would have known about Megaspeed’s dramatic finances. Then again, so would its CFO, Colette Kress, who has just been appointed its interim CEO, with Jensen resigning over the sex scandal.
The U.S. government knew too.
Perhaps the following months will deliver more answers. They may not be as straightforward as people are expecting.
...
Sober voices get drowned out in a crowd. Experienced AI analysts are increasingly urging the U.S. not to overreact to the espionage threat raised by the “Huang Affair.”
“Don’t undermine America’s AI advantage,” William C. Hannas and Huey-Meei Chang at CSET, a security and technology think tank in Washington, D.C., are warning.
They are concerned about converging risks: an overall dilution of U.S. technological leadership from China, but also by possible crackdowns prompted by the Huang Affair from domestic regulators, U.S. AI corporations pushing their own interests, export control issues and a closing outward posture driven by national security concerns.
Others are also concerned. Dylan Patel, the founder and CEO of SemiAnalysis, said on the ChinaTalk Substack about the Huang Affair: “I’m worried this is going to be used as cover to pass really stupid laws.”
“[W]e should talk about the future of open source, the future of AI, and U.S.-China relations,” Patel said.
China views AI as an existential global geopolitical issue, as well as a survival issue for the Communist Party. It has placed its Ministry of State Security at the forefront of handling the risks and in some ways the direction, via risk assessment and policy development, of the technology.
Alice Huang has vanished, probably back to China and is likely now under strict monitoring by the state, whatever her true identity. Jensen Huang’s career lies in ruins.
As the pieces—I’m tempted to say chips—fall in the aftermath of this situation, the central concern is that “individually defensible policies, adopted without regard to their cumulative effects, could weaken the U.S. technological ecosystem on which national competitiveness and security increasingly depend,” Hannas and Chang have written in a blog post for CSET.
They are not against changes but think that they need to be handled carefully, and not in a general, politicized panic.
“American AI developers operate within a political and regulatory environment that differs substantially from China’s,” they said. America's is highly decentralized, and that can create contradictory and inefficient interests.
“[T]he Chinese government can mobilize resources and coordinate industrial priorities in ways that differ from the decentralized American system,” they said.
Vitally, America needs a regulatory environment “that supports innovation while addressing demonstrable risks.” Even, or, especially after, the Huang Affair.
Didi Kirsten Tatlow is Senior Correspondent, International Affairs/Investigations at Newsweek. She is based in Berlin, Germany. You can get in touch by email here.
The Data Story
Jensen Huang was Nvidia. For 33 years, the nerd-turned-technologist—born in Taiwan and later becoming a U.S. citizen—led the company behind the AI boom. It took one week for him to resign over reported affairs with two women, including one who was suspected of reporting to China’s security service.
Until this month, Nvidia was the world’s most valuable company, having reached $5.8 trillion in market capitalization in October. Some would say it’s singularly responsible for the AI industry’s sudden growth spurt because its microchips provide the computing power needed for simultaneous model calculations.
Last year, Nvidia’s graphics processing units powered over 90 percent of AI training around the world, market analysts say, and its most advanced chips are industry standard in data centers in the United States—and coveted globally, including most obviously in China.
Nvidia’s dominance in the field comes from its proprietary software too. The CUDA system, released in 2006, enables developers to write software that works with Nvidia chips—and only Nvidia chips. It’s an ecosystem has locked in AI developers for years.
After Chinese AI lab DeepSeek had its R1 moment nearly two years ago, U.S. concerns grew in the government and in the private sector about China’s progress. The Trump administration further tightened existing chip export controls to China but had carved out a legal niche for Nvidia’s business by the end of 2025.
Nvidia adapted quickly to the new U.S. rules. It modified its top-tier chips and created made-for-China variants that sat just below the performance threshold that would trigger U.S. limits. But the case involving suspected wheeler-dealer Alice Huang and her former Singapore company Megaspeed was a sign that Nvidia’s good-but-not-best products just weren’t enough for China’s hungry labs.
Last summer, Nvidia became the first publicly traded company to reach $4 trillion in value. Its value fell from $5.8 trillion last month to under $4 trillion this week over factors including Huang’s resignation.
Many forget that Nvidia began as a designer of computer graphics cards—I owned a few from their earlier GeForce range back in the day. Nvidia struck up an early relationship with Taiwan chipmaking giant TSMC in 1998, which produced the commercially successful RIVA TNT processor the same year.
Nvidia still dominates 90 percent of the computer graphics card market today.
A global tech superstar like Huang, who had a popular following in the U.S., China and Taiwan, is rarely political, but he became vocal policy lobbyist to save Nvidia’s giant slice of the AI pie.
Those who have heard or read any of his interview comments from the last two years will know his position on U.S. export controls: he thought they were counterproductive and argued that further limits to the tech supply chain to China would only drive Beijing to develop its own chips more quickly.
On a podcast with Dwarkesh Patel this past spring, he dismissed concerns that competitors in China might iterate on Nvidia’s best designs and eventually oust the U.S. tech giant from the Chinese market, a fate that has befallen other Western firms in the past.
“We have to keep innovating,” Huang said. “The premise that even if we competed in China, that we’re going to lose that market anyways…. You’re not talking to somebody who woke up a loser. That loser attitude, that loser premise makes no sense to me.”
Perhaps the only problem with Huang’s argument, which was compelling to many, was that fewer chip restrictions would directly benefit Nvidia and its profit margins. And as Kirsten points out, that may have made many security-conscious figures in Washington, D.C., first uncomfortable, then unsettled, and finally unhappy.
John Feng is Newsweek's China Editor based in Taichung, Taiwan. You can get in touch with John by email here.
View 0 Comments
Submit Correction
View Editorial & AI Guidelines