Xi Jinping is in Washington for the first time in 11 years. Expectations are on the floor, oil is falling and the Nasdaq 100 is at a record. The number I care about most is a zero buried in Nvidia’s guidance, and it’s why I think this summit is far more one-sided than it looks.
Nvidia’s revenue forecast for this quarter is $108 billion. Tucked inside it is my favourite number in markets this week: zero.
That’s the data centre revenue from China that the company has assumed. A clean, deliberate zero. Customers headquartered in China still paid Nvidia $7.9 billion last quarter, about 8% of sales, but that was mostly gaming and workstation chips. H200 shipments into China were under 1% of data-centre revenue.
Jensen Huang is on the guest list for this week’s White House state dinner for Xi Jinping. The two governments keeping that zero at zero are sitting at the same table.
Nobody is promising fireworks. Macquarie’s Larry Hu expects the summit to “focus on what already exists”. Fair enough. My argument is that low expectations plus a zero in the guidance makes this one of the most lopsided bets on the board. I’ll declare my bias now: I’m bullish on Nvidia, and this week makes me more so.
The Free Option
Finance officially recognises one free lunch, diversification. Nvidia’s China line is starting to look like a second.
Think of it as a call option: the right, without the obligation, to profit if something happens. Normally, you pay a premium for that. Here the premium is nil, because management has already told the market to expect nothing. KeyBanc reckons Chinese firms want roughly 1.5 million H200s, around $30 billion of revenue. Washington approved licences in May for about ten Chinese buyers, up to 75,000 chips each, with the US Treasury taking a 25% cut.
So why is it still zero? Because the blockade moved. US licences require the chips to be used inside China. Beijing has told its tech giants to keep Nvidia out of domestic data centres. The two rules cancel each other out perfectly, which is the most efficient thing either government has produced this year.
Here’s what I find most bullish. You don’t have to ban people from buying something they don’t want. The fact that Beijing has to instruct Alibaba and ByteDance not to buy the H200 says more about Nvidia’s pricing power than any earnings call. And the demand leaks anyway: Chinese firms are already renting Nvidia compute in Southeast Asian data centres.
The honest caveat. Every option has theta, the slow decay in value while you wait. This one’s theta is called Huawei. Each month of stalemate is a month for Chinese chips to get good enough, and DeepSeek already builds models to run on domestic silicon. The window is real, and it’s narrowing.
I’d still take the trade. In June, I argued we aren’t in an AI bubble in the 2000 sense, and a company guiding China to zero is part of why I’m comfortable with that. It can’t disappoint on China. It can only surprise.
Boeing’s Maybe
Boeing is the reminder of how these summits actually pay out.
In May, Trump announced from Beijing that China would buy 200 Boeing jets, plus “a promise of up to 750 if they do a good job”. Boeing’s shares fell about 4%. The market had been briefed on 500, so 200 landed as a downgrade in a winner’s rosette. Buy the rumour, sell the Boeing.
Read the vocabulary, too. Boeing called it “an initial commitment”, and in aerospace, a commitment doesn’t make the official order backlog. It’s the industry’s version of “we should grab a coffee sometime”. Trump flagged September for the next tranche, so expect a number out of this week. Whatever it is, expect it to be bigger at the podium than in the filings.
A Deadline With the Shelf Life of Yogurt
What’s actually at stake is the Busan truce, last October’s pause of tariffs and export controls. It expires on 10 November. Washington has offered to extend it by three to six months. Beijing wants to run it to the end of Trump’s term. Talks the weekend before the summit didn’t close the gap. Nothing says long-term strategic partnership like a deadline with a yogurt’s shelf life.
Look at the date. It lands one week after the US midterms, which I wrote about earlier this month. Neither government wants a tariff cliff in the same week as an election result. That’s the strongest argument I can find for an extension actually happening.
The backdrop softened going into the summit. Brent fell to $98.43 on Wednesday, its sixth straight daily decline and the longest losing streak since last August, while US crude has shed more than 10% in five days. In that midterms piece, I predicted crude would drift back into the eighties by late October. I’m not taking a victory lap from $98. The direction is right, though.
Not everyone’s relaxed. Oil prices in China have hit record highs, and BNP Paribas Wealth Management’s Grace Tam still put the odds of a Fed rate rise in October at 53%, a coin toss in a suit. Goldman Sachs sees the main upside risk to crude as strikes on Gulf infrastructure rather than Chinese demand. There’s talk Trump will ask Xi to have a word in Tehran. Former PLA colonel Zhou Bo thinks “China would love to play this role”. I suspect it would love the bargaining power more.
The Strait of Baotou
In June, I argued electrification is the best form of energy security, because an electron doesn’t need safe passage through somebody else’s strait. I still believe that. This summit is where the argument hits the limit.
The electron travels free. The magnet doesn’t. Wind turbines, EV motors and the servers in data centres like the ones I grew up next to in Slough all depend on rare earth permanent magnets. According to the International Energy Agency, China produces about 91% of its refined rare earths. Even if every announced project outside China gets built, the IEA expects non-Chinese output to meet well under a fifth of magnet demand by 2035.
We’ve spent years trying to escape the Strait of Hormuz and walked straight into what I’d call the Strait of Baotou, after the Inner Mongolian city at the centre of China’s rare earth industry.
The same truce governs it. China’s suspension of its harshest rare earth export controls also expires on 10 November. Flows are still running around 50% below pre-restriction levels, and magnet shipments to the US fell again last month. Georgetown’s Evan Medeiros calls China’s grip on critical metals a “royal flush”. Hard to bluff when the other side is holding your magnets. Literally.
For climate finance, this is the line item missing from every green bond prospectus. The market prices carbon risk with real care. It barely prices magnet risk at all, and the date that risk comes due is being negotiated in Washington this week.
The Last Time Xi Came to Newham
Now East London has seen this kind of podium promise before.
In October 2015, at the height of the so-called golden era, Xi Jinping and David Cameron attended the ceremony singing China’s CITIC Construction onto the £1 billion Royal Albert Dock scheme in Newham. The Beijing developer ABP had been chosen by Boris Johnson two years earlier to build “London’s third business district”: 20,000 jobs and £6 billion for the economy.
Twenty-one buildings were finished. Most sat empty. City Hall tore up the agreement in 2022, PwC was appointed liquidator that July, and local press started calling the site a ghost town. The lesson for the week is simple. Newham learned, expensively, the gap between the number announced at a podium and the number that turns up on site. Boeing’s shareholders priced exactly the same gap in May.
What It Means for Your Portfolio
If you have a pension, you almost certainly own this summit already. Nvidia sits near the top of virtually every global tracker, and the Nasdaq 100 has just set a record on the back of cheaper oil. Everbright’s Kenny Ng warns against “blindly chasing highs”. Sensible advice. My read is narrower: the downside is largely priced by low expectations, and the upside is concentrated in names with China optionality. Chipmakers first, Boeing a distant second.
Prediction, written before the handshakes and held loosely. The truce gets extended, somewhere between Washington’s six months and Beijing’s end-of-term ask, and both sides call it a win. Boeing gets a headline number that shrinks on contact with a filing. No real H200 breakthrough, because the veto sits in Beijing, but warm enough language to lift Nvidia for a day or two. And the newly announced AI dialogue “mechanism” gets sold as the deliverable, with a follow-up meeting due inside two months, conveniently after both deadlines.
Gideon Rachman wrote this week that Trump and Xi are “the wrong men at the wrong time”. On the big questions, he is probably right. Markets don’t need the big questions answered, though. They need the truce not to expire.
Watch the chips this week. Worry about the magnets on 10 November.
