Reuters Finds $3.2 Billion in NVIDIA Chip Projects Nvidia Says It Doesn't Have
A Reuters investigation found $3.2 billion in Chinese AI projects using restricted NVIDIA chips, four quarters after NVIDIA guided investors to expect zero China data center revenue.

Why it matters
- Reuters found 23 billion yuan ($3.2 billion) in Chinese AI projects using restricted NVIDIA H100, H200 and H800 chips.
- NVIDIA guided to zero China data center compute revenue for four straight quarters; Q1 FY2027 H20 China shipments were zero versus $4.6 billion a year earlier.
- IDC estimates NVIDIA held 34% of Chinese AI-accelerator shipments in H1 2026.
- Q2 FY2027 revenue hit $96.22 billion, up 105.8% year over year; Q3 guidance of $108.0 billion assumes no China revenue.
- NVDA closed at $162.85 on November 13, 2026, down 12.75% for the week.
A Reuters investigation published November 13, 2026 found more than 23 billion yuan — roughly $3.2 billion — in Chinese AI computing projects built around restricted NVIDIA processors, four quarters after NVIDIA started guiding investors to expect nothing from China. The report landed the same week CBS News covered Jensen Huang's resignation as chief executive, and it landed on a stock that closed at $162.85 on November 13, down 12.75% for the week and 26.41% for the month. The gap between what NVIDIA tells investors and what Chinese buyers are doing is the story here. It matters because NVIDIA has already baked zero China revenue into its forecasts, which means any legal reopening would add to the number — and any enforcement crackdown adds risk to projects already running.
What did Jensen Huang actually say about China?
In 2025, Huang said NVIDIA's China market share had fallen from 95% to zero after U.S. export restrictions. The company's filings backed him up.
The paper trail is consistent:
- In Q3 FY2026, NVIDIA said H20 sales were "insignificant in Q3, reflecting export restriction impacts on China-targeted products."
- In Q1 FY2027, it shipped no H20 compute products to China, compared with $4.6 billion a year earlier.
- On the August 26 earnings call, CFO Colette Kress said: "Given ongoing geopolitical uncertainty, there is no China data center compute revenue in our forward outlook."
- NVIDIA's August filing said the company was effectively excluded from China's data-center computing market.
Kress also disclosed that licensed Hopper 200 shipments to China made up less than 1% of Data Center revenue, and she called them "dilutive to corporate gross margins." In plain terms: the legal channel that remained was small enough to be a rounding error, and it actually hurt margins. That's the official version — a market of 1.4 billion people reduced to less than 1% of one segment.
What did Reuters actually find in China?
Reuters assembled more than 150 Chinese procurement notices, investment plans, contracts and corporate reports. Together they cover over 23 billion yuan, about $3.2 billion, in AI computing projects built around restricted NVIDIA processors — including H100, H200 and H800 chips.
There's a measurement caveat worth understanding. The $3.2 billion figure includes supporting infrastructure, so it captures what Chinese buyers are spending on whole projects, not what NVIDIA booked as chip revenue. Some of the projects are running, some are under construction, and some are still planned.
Reuters also cited IDC estimates that NVIDIA held 34% of Chinese AI-accelerator shipments in the first half of 2026. That counts shipments into China, including hardware NVIDIA may not have sold directly or legally. Reuters has not linked these projects to the separate Megaspeed claims, so these are two distinct threads of scrutiny, not one case.
Read together, the two pictures aren't necessarily contradictory. Huang's "zero" described NVIDIA's legal access to China — what it can book, ship and disclose. Chinese demand for its chips continued through channels the company doesn't control. The takeaway: the market didn't die, the revenue did.
Why doesn't losing China show up in NVIDIA's numbers?
Huang's own words explain the mechanism. Talking about fiscal 2028, he said: "Even though our demand is much greater than 70%, our supply allows us to confidently deliver 70%." He added: "The unconstrained number would be a lot higher."
That's the key. NVIDIA is supply-constrained, not demand-constrained. Chips that can't go to China get absorbed by buyers elsewhere, so a lost market doesn't leave a hole in revenue — it just caps what the company could otherwise sell.
The recent results show how little the China exit dented the trajectory:
- Q2 FY2027 revenue: $96.22 billion, up 105.8% year over year, ahead of the $92.07 billion consensus.
- Non-GAAP EPS: $2.22, beating the $2.09 estimate.
- Data Center revenue: $89.02 billion.
- Q3 guidance: $108.0 billion plus or minus 2%, assuming no China Data Center compute revenue.
"Now, compute is revenue," Huang said on that call. When your product is compute itself and demand outruns supply by a wide margin, one market can vanish from the spreadsheet without flattening growth. The one-line version: NVIDIA lost China and doubled anyway.
Why is China back on investors' radar now?
Two things changed in the same week. First, the Reuters investigation put restricted NVIDIA chips on the record inside Chinese projects — running, under construction and planned. Second, Huang resigned, leaving NVIDIA without a permanent CEO, an event 24/7 Wall St. noted came with a historical warning: the last company that did that left shareholders 17 points behind the market.
The investment math cuts both ways. Because guidance already assumes nothing from China, any licensed reopening would be pure upside to the forecast. But the same reporting that documents the demand also documents the exposure: Reuters says the leadership change and U.S. scrutiny now imperil those Chinese projects. A company that has told investors to expect zero China revenue is now tied, through its own chips, to projects that enforcement could shut down.
That's the bind. The $3.2 billion Reuters found isn't on NVIDIA's income statement, and NVIDIA can't legally claim it. But it proves the demand exists, and it proves restricted hardware is flowing. Watch whether U.S. regulators treat the Reuters findings as a compliance problem for NVIDIA or purely a Chinese procurement story — that determines whether this week's stock slide marks a bottom or a new risk line in the filings.
Source: 247wallst.com
More from Sophie Lindqvist
Show full bio
Staff writer covering regulation, policy and their impact on artificial intelligence and machine learning at AI In Context.
625 articles
Related articles
- Nvidia's Board Opens Independent Probe Into CEO Jensen Huang
- Report: Jensen Huang's Affair Linked to $7.5 Billion in Chips Reaching China
- Nvidia Faces $400 Billion Two-Day Wipeout Over CEO Affair Allegations
- Nvidia Loses $430 Billion as CEO Affair and China Chip Claims Escalate
- Kalshi Traders Put 55% Odds on Jensen Huang Exiting Nvidia by 2029