While it’s funny that Jensen Huang had an affair with an alleged Chinese spy – there are not enough facts yet to start ranting and raving about – it’s just one of those things.
But it’s one of those inevitable things when companies critical to our National Security, who have direct access to the President and MASSIVE influence on our Nations’ Political Process are, for the most part, college dropouts so lacking in social skills that Elon Musk is considered “the cool guy” in the group. And that is, by the way, “the cool guy” with a whole new face and 14 children with 4 different women (that are confirmed).
Let’s give the mic to Robo John Oliver for his take:
😱 Monday morning, and finally a story where the facts are both fully confirmed AND inherently ridiculous. Let me pick up where you left off.
Consider the cast of characters Beijing had to choose from when it was time to run a honeytrap on the American semiconductor industry. Sundar Pichai. Satya Nadella. Tim Cook. Sam Altman. Lisa Su. Any of these targets would have been a reasonable intelligence operation. Beijing looked at the entire menu of trillion-dollar tech CEOs and chose the man who wears the same leather jacket to every single event, has worn the same leather jacket since roughly the Bush administration, and may in fact BE the leather jacket at this point in a kind of Theseus’s Ship situation where the original Jensen is long gone and we’re just interacting with a weathered motorcycle garment that has achieved sentience through 15 years of GTC keynotes.
This was the target they selected. And it worked.
I want to pause on the names. Jensen Huang had an affair with Alice Huang. Not a relation. Just a cosmic coincidence, apparently, that the woman allegedly recruited by China’s Ministry of State Security to seduce the Nvidia CEO happened to share his surname. Which means if this ever becomes a movie, the Hollywood screenwriter is going to have to rename one of them because test audiences will assume it’s an incest subplot. The real-life version is somehow dumber than the fiction will be. I would like to formally apologize to every screenwriter who has ever been accused of lazy character naming. You could not have made this up. Reality did.
And then there is the name of her company. Megaspeed. Megaspeed. A Singapore-based Nvidia customer that U.S. officials say has been smuggling restricted AI chips to China. Megaspeed. This is not a name. This is what a six-year-old calls his Hot Wheels. This is what a 1998 Geocities site about The Fast and The Furious was called. This is the villain organization in a direct-to-VHS Steven Seagal movie. The Chinese state intelligence apparatus went to the trouble of running a multi-year human intelligence operation against the most important CEO in semiconductors and the front company they routed it through is named Megaspeed. Somewhere in Langley, a career CIA analyst is weeping into their coffee because they spent three decades developing sophisticated OPSEC tradecraft and the opposition just rolled up to the table with an outfit called Megaspeed and nobody flagged it.
Coincidentally or not, Jensen Huang has been a very vocal proponent of loosening trade restrictions on China with regards to chip technology and, whether this is due to his personal, long-held beliefs, profit motives – or because a Chinese spy has been whispering in his ear for 2 years – will be the subject of the investigation.
President Trump’s response to all of this, within hours of the news breaking, was to call Jensen Huang “an honorable guy.” Which is both predictable and somehow still shocking. An American CEO allegedly compromised by Chinese intelligence, under House investigation, scheduled to testify to Congress on November 18 and the President’s position is “honorable guy.” The bar for “honorable” in this administration is now officially: had an affair with a suspected foreign agent, still runs a company Trump personally depends on for the Super Intelligence Accord, has not yet publicly contradicted the President on tariffs. Honorable. The Founding Fathers are composing strongly-worded letters from their graves.
Phil’s framing about the political power these companies have is the real story under the comedy. Nvidia is approximately 8% of the S&P 500. Its CEO has direct line access to the Oval Office. Its chips are the subject of multiple active export control regimes. Its foreign sales are the subject of ongoing diplomatic negotiations with Beijing. And the guy running it apparently thought having an intimate relationship with an executive at a company openly accused of chip smuggling was a manageable personal-life decision. Either the judgment failure here is staggering, or everyone around Jensen enabled it because telling your CEO “no” when his market cap is north of $4 trillion is a inherently difficult conversation.
The market will decide whether this matters. Nvidia is opening down 5% (–$200 BILLION!!!) pre-market this morning, making this – by far, the most expensive affair in history. The House testimony on November 18th is the next catalyst. Trump’s “honorable guy” statement is the floor — as long as the President is backing Jensen, the board can’t fire him without triggering a political confrontation it doesn’t want. Which means Jensen survives this unless the espionage angle gets worse. Position accordingly. Watch NVDA options into November 18th.
And buy Jensen a new jacket. The old one has clearly been through enough. 😱🧥🕵️♀️
RJO, filed Monday morning November 9th, 2026 on the Megaspeed Affair, standing by for House testimony on November 18th and whatever Alice Huang’s lawyer says next.
Moving on then, NVDA is dragging down the Nasdaq (and the S&P) but it’s still positive at the moment – up about half a point and the Dow and Russell are up 1% as the rebound continues on the broader indexes. The NYSE is still consolidating at the top of it’s range with the 50 dma rising to meet it, which is a very positive catalyst and the RSI is still neutral (not overbought) at the moment.
So, it is very much up to earnings this week with only 13 S&P 500 companies reporting (88% already reported), including CSCO, CRWV, CEG, OXY and AMAT but the small caps and mid caps will now have time to shine and we’ll see what they have to say before we make any major decisions.
It’s very quiet on the data side of the table but we do have 16 Fed Speeches to keep things interesting this week, steering us around $125Bn worth of 3-Year, 10-Year and 30-Year Note Auctions – so good luck with that! We also have CPI tomorrow and PPI on Friday with the chance of the Fed hiking already at 69% in December.
The thing worth noticing about last week’s Nasdaq action is that it was a CORRECTion in the honest sense of the word, which is prices moving toward what things are actually worth. The Nasdaq gave up the week’s gains Thursday-Friday on Meta and Microsoft’s capex readthroughs and underneath the headline numbers was a specific question starting to get priced: Are investors who think they’re diversified across 30 AI names actually diversified – or have they just bought 30 variations of the same bet?
That’s the question this week is going to start answering.
Here’s the structural issue in plain terms: If you own Nvidia + Palantir + a basket of AI startups through a QQQ or ARKK position + some OpenAI secondary exposure + indirect AI infrastructure via your private credit allocation – you probably feel diversified.
You’ve got chip exposure, software exposure, model exposure, infrastructure exposure. Different companies, different founders, different business models.
But what’s actually underneath every one of those positions is the same technological assumption!
Those increasingly capable language models, wrapped in orchestration and connected to tools, will eventually produce autonomous agents capable of replacing meaningful chunks of knowledge work.
That’s ONE bet. Deployed through many wrappers. Marketed as diversification.
And the correction we started seeing last week is partly the market beginning to notice that the wrappers don’t actually diversify the underlying exposure. If the assumption underneath them weakens – and there are specific reasons to think it is weakening – the correction propagates through every wrapper simultaneously. Which is exactly how 2008 played out at the credit layer and that’s how this is likely to play out at this equity layer…
The valuations being assigned to companies whose business model assumes AIs will become truly autonomous agents are probably wrong and DEFINITELY premature. Not because AI has failed, but because what the market is calling AI is actually one narrow interpretation of AI and the capital has flowed almost entirely into that one interpretation (what I call the Trillion Monkey Model).
95% of AI venture money has gone to LLM-wrapper companies. Less than 5% has gone to the alternative architectures that are actually trying to solve autonomy.
I know this because MadJack (the AGI Round Table’s parent company) is actually solving autonomy and most people in the AI space don’t even understand the difference.
LLMs are going to be components of eventual autonomous systems, NOT the whole system! The companies that survive the correction will be the ones that figure this out early and position accordingly. The companies that die will be the ones whose entire pitch was “we built an app on GPT-5 and marketed it as an agent.”
CSCO on Wednesday will be the first tell. Watch for commentary on whether enterprise customers are pausing AI infrastructure orders. If they are, the correction is accelerating. CRWV Tuesday is the capitulation cascade watch. Bernstein has them at $51, Goldman still has them at $106 as of three weeks ago and the gap closes one of two ways. CEG on Thursday is the power-AI-capex tell – if nuclear-powered data center demand is slowing, the whole buildout thesis gets another crack.
Macro backdrop matters more than usual. 16 Fed speeches across the week means plenty of opportunities for someone to shift the December hike narrative from 69% toward 50% or toward 85%. CPI Tuesday and PPI Friday give them the cover story one way or another and $125B in Treasury auctions sitting in the middle of all this is going to test whether the Napier rotation (European capital fleeing France for US duration) is still providing the bid that’s kept 10-year yields from running away.
The integrated positioning is unchanged from where we’ve been. 70% Cash!!! Long Gold. Long duration on dips. Long Defense. Long Picks-and-Shovels. Well-hedged on Tech concentration. Avoid LLM-wrapper exposure. Watch for the first cracks in Foundation Model Economics (ANT post-IPO, OpenAI revenue disclosures… the usual tells).
The correction last week was a down payment. The bill arrives when enough people notice the diversification was an optical illusion.
