Global markets are no longer treating China’s semiconductor ambitions as a distant threat.
They are pricing them in now.
The latest selloff began in the one corner of the market that investors had treated as almost untouchable: the artificial-intelligence hardware trade. Nvidia dropped about 5% Monday, SanDisk fell roughly 11%, Micron weakened, ASML was hit hard in Europe, and the selling then rolled into Asia, where South Korea’s KOSPI plunged nearly 10% and triggered a circuit breaker. Japan’s Nikkei fell about 4% to 4.4%, according to Reuters and AP.
That is not a normal rotation. It is a warning shot at the structure holding up the global equity rally.
For the last three years, the market’s strongest assumption has been simple: American AI leaders would dominate, Dutch lithography would remain the bottleneck, South Korean and U.S. memory suppliers would enjoy pricing power, and China would remain technologically boxed in by export controls. That assumption is starting to crack.
Reuters reported Monday that China has begun manufacturing domestically developed immersion deep ultraviolet lithography machines, a key chipmaking tool long dominated by ASML. The machines are expected to be delivered this year to major Chinese chipmakers including SMIC, Hua Hong Semiconductor and ChangXin Memory Technologies, according to The Information’s reporting cited by Reuters. ASML declined to comment.
The market reaction was immediate because the strategic meaning is larger than the first machines themselves.
Reuters noted that the Chinese systems still lag ASML on performance and reliability, need more testing, and will begin with limited production — about five DUV machines this year and roughly 20 in 2027. That means this is not the instant death of ASML, nor does it mean China has suddenly caught up to the West at the most advanced edge of chipmaking. But the report does show something more important: China is building around the sanctions wall rather than waiting for it to come down.
That is why markets are right to react.
The risk is not that China replaces ASML overnight. The risk is that China has started reducing its dependence on the exact choke points Washington and its allies have used to slow Beijing’s chip industry. Once China proves it can build credible domestic alternatives, even imperfect ones, investors have to rethink the entire premium attached to the Western semiconductor supply chain.
This is already spreading beyond ASML.
ChangXin Memory Technologies, or CXMT, became a flash point after a huge Shanghai market debut that pushed China’s semiconductor ambitions back into the center of global trading. Reuters said CXMT’s rise, combined with the domestic DUV report, left investors wondering whether China’s chip industry is moving from catch-up mode toward contender status.
That is the part investors cannot ignore. For years, China was treated as a demand story for foreign chipmakers. Now it is becoming a supply story, a competition story and a political-risk story at the same time.
Memory stocks are especially vulnerable. If China gains ground in DRAM and other memory categories while AI demand is still pricing in scarcity, the market has to ask whether today’s margins are sustainable. That is why names tied to memory and storage were hit so hard. SanDisk’s double-digit drop was not just about one company. It was the market repricing the idea that AI hardware scarcity can last forever.
Nvidia’s problem is different but just as dangerous.
The company’s shares fell after reports that it was discussing a roughly $250 billion financing backstop for an OpenAI data-center project and another possible arrangement tied to hundreds of billions of dollars of chip purchases. Reuters said the discussions put a spotlight on the increasingly circular nature of AI build-out spending. The Wall Street Journal reported that Nvidia’s stock fell after news of the backstop intensified concern about the company investing in its own customers, some of whom need financing to buy its chips.
That is the market’s second problem.
China is attacking the supply-chain assumption from one side. Circular AI financing fears are attacking the demand-quality assumption from the other.
The AI boom has been sold as a clean demand story: customers need compute, Nvidia sells chips, hyperscalers build data centers, and revenue compounds. But if the chip seller is increasingly helping finance the buyers, investors are right to ask how much of the demand is organic and how much depends on a financial loop that works only while valuations keep rising.
Axios described the concern plainly: Nvidia is using its financial power to support customers that are burning large amounts of cash, while those customers are expected to buy more Nvidia hardware. Axios also reported that Nvidia shares lost about a quarter-trillion dollars in market value Monday and that credit-default-swap pricing on Nvidia bonds showed stress as the circular-deal concerns spread into the debt market.
That is not a small tremor. That is the credit market tapping the equity market on the shoulder.
Now add the Federal Reserve.
Reuters reported that markets see roughly a 38% chance of a U.S. rate hike this week, while a separate Reuters-based market report said traders were pricing a 62% chance of no change and a 38% chance of a 25-basis-point hike at Wednesday’s decision. CME says its FedWatch tool tracks rate probabilities implied by 30-day Fed funds futures.
That is a serious repricing for a market built on long-duration growth stocks.
AI stocks are not cheap defensive assets. They are priced for years of future growth, massive capital spending and sustained access to financing. A surprise hike would hit exactly the part of the market that has carried indexes higher: expensive tech, crowded semiconductors, data-center infrastructure, crypto and anything else dependent on easy liquidity.
That is why Bitcoin is reacting too.
BTC was trading near $63,477 in the latest market snapshot, down about 2.8% on the day, after falling as low as $63,054 intraday. Whether a particular venue briefly printed below $63,000 or not, the message is the same: crypto is not trading like a safe haven. It is trading like a risk asset caught in the same liquidity squeeze hitting AI and semiconductors.
The next 72 hours matter because the market is facing three tests at once.
First, China has forced investors to confront the possibility that the West’s chip bottleneck is not permanent. Second, Nvidia’s financing discussions have revived fears that the AI boom is being partly propped up by circular capital flows. Third, the Fed is entering one of its most uncertain meetings in years, with rate-hike odds high enough to matter but low enough to shock traders if the hike actually arrives.
Then comes Big Tech.
Microsoft, Meta, Apple and Amazon are all set to report earnings this week, with investors focused less on headline revenue and more on AI spending, cloud demand, data-center costs and whether the hyperscalers can justify the enormous capital expenditures already embedded in market valuations. Reuters said those reports will help round out the picture on hyperscaler spending plans. Zacks and other earnings previews have Microsoft and Meta scheduled for Wednesday, July 29, and Apple and Amazon for Thursday, July 30.
That creates an unusually fragile setup.
If Big Tech confirms that AI spending is still accelerating, investors will ask whether the returns justify the spending. If spending slows, investors will ask whether Nvidia’s order book and the memory boom have already peaked. If the Fed hikes, valuations compress. If the Fed holds but sounds hawkish, the relief may be limited. If China’s chip progress keeps dominating headlines, every bounce in ASML, Nvidia, Micron, SK Hynix and Samsung becomes harder to trust.
This is why the China story is not being overblown.
Markets are not crashing because one Chinese company suddenly built a perfect ASML replacement. They are selling because the entire AI trade depends on a chain of assumptions that all need to hold at the same time. China must stay behind. AI demand must stay real. Financing must not look circular. Memory shortages must persist. Hyperscalers must keep spending. The Fed must not tighten into the trade.
On Tuesday, too many of those assumptions were hit at once.
That is why the selloff in Asia matters so much. South Korea is not just another market. It is home to Samsung Electronics and SK Hynix, two pillars of the global memory and AI hardware supply chain. Japan is not just another market either. Its chip-equipment and materials ecosystem sits deep inside the semiconductor production map. When KOSPI trips a circuit breaker and the Nikkei drops more than 4%, the market is not reacting to noise. It is repricing the AI supply chain itself.
The comfortable narrative was that China remained boxed in while U.S.-aligned chip leaders collected the profits.
The new narrative is more dangerous: China is adapting, domestic alternatives are emerging, Western bottlenecks may weaken over time, and the AI boom is beginning to look more financially stretched just as rates may rise again.
That does not mean the AI trade is dead. It does mean the market has been forced to admit it is not invincible.
For months, investors treated every dip in chips as a buying opportunity. This one is different because it challenges the foundation of the trade. The issue is no longer only valuation. It is dependency: dependency on ASML, dependency on Nvidia financing, dependency on hyperscaler capex, dependency on cheap money, and dependency on the belief that China cannot catch up fast enough to matter.
China just made that belief harder to defend.
And when the most crowded trade in the world starts losing its cleanest assumptions, the selloff does not need a single trigger. It only needs doubt.
Right now, doubt is spreading from Shanghai to Seoul, Tokyo, Amsterdam, New York and Bitcoin.
That is why markets are shaking.
And this time, they have a reason.
