If you only look at the chart, Nvidia is simply correcting after a boom. But behind the numbers lies a structural shift: a company that spent years dictating the terms of the artificial intelligence market found itself held hostage by decisions made in Washington, not Silicon Valley, for the first time.
Where the $1 trillion in losses came from
Nvidia shares fell approximately 30% from their November 2024 peak of $153, resulting in a loss of approximately $1.1 trillion in market value. There are several triggers, but the key one is not market-related.
After an unexpected ban on H20 chips in the Chinese market, Nvidia revealed it was writing off over $5.5 billion due to excess inventory and obligations to suppliers. Analyst Ed Mills of Raymond James noted: "Restrictions on H20 chips were a surprise, given the apparent approval of the product by the Biden administration."
What H20 meant for Nvidia — and why its ban hurts
H20 was a specially designed "lighter" chip for the Chinese market — a product Nvidia created specifically to circumvent previous U.S. export restrictions. According to SEC filings, Nvidia has still not received any licenses to supply advanced Blackwell chips to China. In other words, the most profitable market outside the U.S. is effectively closed to its entire flagship product line.
The U.S. government states that it may prove impossible to create a competitive product for the Chinese data center market that would receive regulatory approval — effectively meaning Nvidia's exclusion from this segment.
Competitors seized the moment — but fell too
After the ban announcement, AMD shares fell more than 7% — the company said the restrictions would cost it up to $800 million. Broadcom and Qualcomm lost over 2%, Intel — over 3%. The AI chip market responded as a unified system, not as a set of separate players.
DeepSeek, tariffs, Huawei: three blows coming one after another
Nvidia shares fell in waves: first, the "DeepSeek shock" crashed them 17%, then — concerns about the AI industry overall, then — Trump's tariff policy. Nvidia faced particular pressure due to global supply chains and significant sales in international markets: tariffs directly affect both costs and demand.
The company also faces the risk of additional export restrictions — not only regarding China, but also regarding supplies through Singapore and Vietnam.
GPU as a commodity: a signal that was ignored
The rental price of a single B200 GPU on cloud platforms reached a peak of $6.11 per hour on May 30, 2026, and by June 21 had fallen to $4.22 — minus 31% in three weeks. This is a market indicator of real demand for AI computing, and it showed cooling earlier than major investors reacted.
"JPMorgan and Bernstein analysts estimated the cumulative impact of export restrictions and competition from Chinese manufacturers at $5.5–16 billion in lost revenue over the next financial cycle."
Estimate by JPMorgan and Bernstein analysts, according to indmoney.com
Nvidia still controls the GPU market for AI — but for the first time, its growth depends not on how good its chips are, but on whom it's allowed to sell them to. This is a different company than the one that soared thanks to ChatGPT.
If the Trump administration expands export restrictions on next-generation chips — Rubin — beyond China to the Middle East and Southeast Asia, Nvidia's market capitalization will not recover to 2024 levels in the next two years. The question is not whether there will be demand for AI — there will be. The question is who will be permitted to serve it.
