RockstarMarkets
All news
Markets · Narrative··Updated 1h ago
Part of: Semiconductor Cycle

China rejects Nvidia chips as US approves H200 export deals; semiconductor sentiment weakens

China declined to purchase US advanced semiconductors despite US approval to export Nvidia H200 chips to 10 Chinese companies, signaling Beijing's pivot to domestic alternatives. The move pressured Nvidia down 2.2% and broader chip stocks Friday as traders reassess the value of China export permitting amid intensifying US-China technology competition.

R
Rocky · RockstarMarkets desk
Synthesised from 8 wires · 30 mentions in the last 24h
Sentiment
-42
Momentum
68
Mentions · 24h
30
Articles · 24h
20
Affected sectors
Related markets

Key facts

  • US approved H200 chip exports to 10 Chinese companies; China declined to purchase
  • Nvidia declined 2.2% Friday on China chip export rejection news
  • AMD down 3.3% Friday amid broader semiconductor sector weakness
  • Chinese domestic semiconductor programs accelerating; SMIC and local fabs ramping capacity

What's happening

A geopolitical tug-of-war over semiconductor access emerged as the US approved the export of Nvidia's H200 chips to 10 Chinese companies, only for Beijing to signal it would not pursue the purchases and instead double down on domestic alternatives. The apparent reversal caught traders off-guard and raised questions about the utility of recent export licensing approvals. Nvidia declined 2.2% Friday on the news, with broader semiconductor sector weakness following suit as AMD and Broadcom also felt selling pressure. The dynamic underscores the precariousness of relying on China as a growth market for US chipmakers in a climate of deteriorating bilateral relations.

Context matters here: the Trump-Xi summit in Beijing concluded with limited concrete deliverables, and while statements emphasized a 'very strong' relationship, specific trade concessions remained elusive. The chip export approval appeared timed to signal openness to commerce, but China's cold shoulder response suggests Beijing is prepared to absorb short-term tech gaps by investing heavily in homegrown semiconductor capacity. State-backed programs and domestic fabs are accelerating, with companies like SMIC ramping production and domestic AI chips being positioned as viable alternatives within 18-24 months. Analysts noted that geopolitical decoupling, not free trade, is now the baseline expectation.

The broader implication for the semiconductor sector is margin compression and demand uncertainty. Nvidia derives meaningful revenue from Chinese entities, and any sustained denial of access pressures forward guidance. AMD faces similar headwinds. Meanwhile, Broadcom, which supplies networking chips to data-center operators globally, saw some pullback as traders worried that Chinese datacenter builds could shift to domestic suppliers. Raymond James research flagged that Arista Networks is now competing more aggressively on switching supply into Chinese deployments, a sign that market share dynamics are shifting even in permitted products.

Bull-case counters argue that US semiconductor strength ultimately rests on technological superiority, a moat that Chinese fabs have not yet breached. If AI inference and training continue to drive capex, US chips will remain competitive on raw performance. However, the bear case, that regulatory bifurcation and geopolitical risk are now permanent features of chip demand forecasting, seems to be winning the near-term narrative, especially with earnings volatility and margin concerns already depressing valuations ahead of NVDA's May 21 earnings.

What to watch next

  • 01Nvidia Q1 earnings call May 21: forward guidance on China demand
  • 02US-China bilateral trade negotiations: any new export licensing restrictions
  • 03Chinese domestic chip revenue and fab utilization data releases
Mention velocity · last 24 hours
Coverage from these sources
Previously on this story

Related coverage

More about $NVDA

Topic hub
Semiconductor Cycle: AI Capex, Memory and the SOX Trade

Live coverage of the AI semiconductor cycle — NVDA, AVGO, AMD, ASML, memory demand, capex run rates and overbought signals.