# Nvidia's hidden China sales vanish, local chips win

By @Lihong062 on X · https://f.xyz/t/98c71de1-c6b3-4e23-aa7e-eda4d5d386c8

> If over 20% of Nvidia's China-linked sales ran through Southeast Asian middlemen, they are at risk as Beijing backs its own chips, so Nvidia is overvalued.

**Result:** If held over the last 3 months (Aug 13, 2026 – Nov 12, 2026): +19.2%. Over the same days: Chip stocks −14.8%, S&P 500 −2.2%.

## What it holds

- **Short the chip king** (55%, +22.8% over 3 months): against NVDA (Nvidia) +22.8%
- **Short the server middlemen** (15%, +27.9% over 3 months): against SMCI (Super Micro Computer) +27.9%
- **Long China's own chips** (15%, +2.2% over 3 months): CAMBRICON (Cambricon) +2.2%
- **Cash** (15%)

## The plan behind it

The main bet is a short on Nvidia, because the take says its sales are less safe than the price assumes. Supermicro is a smaller short. The thread links its people to a server-smuggling case, so a crackdown would hit it too. The only long is Cambricon, a Chinese AI chip maker that gains if China buys local chips instead. A slice of cash stays aside, since this kind of story can take a long time to play out and shorts can squeeze up.

> I personally haven’t seen $NVDA bulls seriously engage with the strongest version of the short thesis. 
> 
> It is perhaps worth revisiting Culper’s May report and establishing the strongest bull case in view of the evidence presented.
> 
> THE SHORT THESIS
> 
> Nvidia says its direct China compute business went to zero after April 2025 export curbs. 
> 
> Culper argues Chinese demand instead moved through Southeast Asian intermediaries, with some GPUs allegedly diverted illegally. 
> 
> It estimates >20% of FY26 compute revenue remained China-driven and predicts a $30B+ revenue hole as Beijing favors domestic alternatives.
> 
> THE EVIDENCE
> 
> 1) Following the money 
> 
> Megaspeed’s balance sheet ballooned from $33M to ~$3B in 2024, largely on $2.9B in “refundable deposits” from an unnamed source, mirrored by ~$2.9B due from subsidiaries. 
> 
> Its Malaysian arm, Speedmatrix, pledged its assets to Apex, a Singapore Alibaba affiliate. Apex reported ~$4.2B of related-party loans and $4.1B in prepayments. 
> 
> The FT has since reported Alibaba loans to Megaspeed. Nvidia’s assurance that Megaspeed has no Chinese SHAREHOLDERS doesn’t address who FINANCED it.
> 
> 2) Following the chips 
> 
> Culper’s trade records show Speedmatrix imported $4.6B of product from Dec 2024–Jan 2026, including ~$4B from Aivres, an ex-Inspur Nvidia Elite OEM with Chinese ties. 
> 
> Giga Computing allegedly shipped >$500M in H200/Blackwell servers to Speedmatrix and Novagate, a suspected affiliate.
> 
> In March, DOJ charged Supermicro-linked individuals in an alleged $2.5B server-purchase scheme routing AI servers to China via Southeast Asia and using dummy servers to deceive inspectors. 
> 
> Separately, Bain’s Bridge Data Centres dropped Megaspeed; Culper cites one anonymous source alleging dummy servers were used there too.
> 
> 3) Following the Jensen
> 
> Culper flagged Jensen Huang’s close contact with Megaspeed executive Alice Huang and quoted an employee saying Alibaba reps repeatedly appeared around Jensen’s visits. 
> 
> Jensen has since acknowled
> — @Lihong062 on X: https://x.com/Lihong062/status/2120385158066880907

**What would prove it wrong:** Nvidia or the authorities show that Southeast Asian buyers are not feeding China, or Nvidia's sales keep growing with no sign of a hole.

## FAQ

### How do I bet that if over 20% of Nvidia's China-linked sales ran through Southeast Asian middlemen, they are at risk as Beijing backs its own chips, so Nvidia is overvalued?

This tested idea holds CAMBRICON (Cambricon) and bets against NVDA (Nvidia) and SMCI (Super Micro Computer), with 15% in cash. Make a copy to run it on practice money first, at real prices.

### What would prove it wrong?

Nvidia or the authorities show that Southeast Asian buyers are not feeding China, or Nvidia's sales keep growing with no sign of a hole.

### Can I make one from my own idea?

Yes. Write what you believe in plain words at https://f.xyz/create. fxyz asks a couple of questions, builds the portfolio and shows how it would have done over the last 3 months.

### Can I turn a post on X into a theory?

Yes. Reply to the post and tag @fxyzai. It tests the view in the post and replies with the 3-month result and a link.

### What can it trade?

Stocks (Tesla, Apple, Nvidia), crypto (Bitcoin, Ethereum, Solana), pre-IPO companies (OpenAI, Anthropic), indices (Nasdaq 100, S&P 500), commodities (gold, silver, oil, copper) and currencies (EUR/USD, USD/JPY). One theory can mix them, and bet for or against any of them.

### How does it react to news?

On each check it reads the last two days of news on its biggest holdings and says whether the idea still holds. It can make one small change at a time: trim or drop one holding by up to 5 points. On autopilot the change waits 10 minutes so the owner can skip it; otherwise it asks first.

### How do I run it with real money?

For real money, tap Make it mine, open your copy and choose Use real money, with an exchange connected: Hyperliquid, Lighter, Lighter (Robinhood), Backpack, Pacifica or Binance. The money stays in your own account there. fxyz can place trades, but it can never withdraw.

### Can the AI get it wrong?

Yes. fxyz's AI turns the idea into holdings, and it can get that wrong: add a coin or stock that doesn't fit, or miss one that does. Check the holdings before you put money in. In your own copy, ask its chat to add or drop any of them.

### What do the 3-month numbers mean?

What $100 would have done holding this mix over the last 3 months, at real prices. It is a what-if, not a record: past or simulated results are no promise of future ones, and nothing here is investment advice.

Test your own idea: https://f.xyz/test
