Nvidia CEO Jensen Huang has criticized the United States’ restrictions on selling its H20 graphics processing units (GPUs) to China.

Speaking at the Computex 2025 trade show in Taipei, Huang expressed concerns about the impact of these restrictions on Nvidia and the broader AI ecosystem.

The restrictions, initially introduced by the Trump administration, have significantly affected Nvidia’s operations.

Huang said that the company wrote off approximately US$5.5 billion in inventory and lost an estimated US$15 billion in potential sales to China.

Huang argued that limiting access to the H20 chips would not stop China from advancing its AI capabilities.

.source-ref{font-size:0.85em;color:#666;display:block;margin-top:1em;}a.ask-tia-citation-link:hover{color:#11628d !important;background:#e9f6f5 !important;border-color:#11628d !important;text-decoration:none !important;}@media only screen and (min-width:768px){a.ask-tia-citation-link{font-size:11px !important;}}

🔗 Source: South China Morning Post

🧠 Food for thought

1️⃣ Export controls historically yield mixed results for national security objectives

US restrictions on China’s tech sector have repeatedly demonstrated the challenges of achieving strategic goals through export controls.

The US government previously targeted Fujian Jinhua, a state-owned Chinese chipmaker, with export restrictions in 2018, citing national security concerns similar to today’s Nvidia case 1.

These earlier restrictions were also designed to slow China’s technological advancement, yet China’s semiconductor industry continued growing, with the country investing heavily in domestic capabilities through initiatives like “Made in China 2025” 1.

Nvidia’s situation reflects a persistent challenge for policy makers: despite export controls, China has demonstrated resilience by developing workarounds, such as optimizing less powerful chips for AI applications as noted by Brookings Institution research 2.

The Semiconductor Industry Association has long warned that narrowly targeted controls are necessary to avoid unintended consequences on US competitiveness, suggesting that overly broad restrictions can weaken American leadership 3.

Jensen Huang’s criticism that blocking H20 chips won’t stop China’s AI development aligns with historical patterns showing that technological advancement often finds alternative paths around regulatory barriers.

2️⃣ Financial ripple effects of tech restrictions extend far beyond direct sales

Nvidia’s experience illustrates how export controls create cascading financial impacts throughout technology ecosystems.

The company’s $5.5 billion inventory write-off and potential $15 billion in lost sales to China’s $50 billion market represent immediate financial damage, affecting not just Nvidia but its shareholders and supply chain partners.

The mainland Chinese market constituted approximately 14% of Nvidia’s total revenue in its most recent fiscal year, highlighting the significant financial stake US tech companies have in maintaining access to Chinese markets.

Beyond direct sales impacts, Huang’s concern about the long-term harm to Nvidia’s AI ecosystem, particularly its CUDA programming interface dominance, suggests these restrictions threaten strategic technological positioning that took years to build.

Industry analysts at the RAND Corporation have noted that international competition significantly shapes AI policy outcomes, with restrictions potentially affecting global economic dynamics well beyond the targeted technology 4.

These broader financial consequences align with the Semiconductor Industry Association’s warning that poorly designed regulations could lead to the loss of strategic markets and ultimately weaken US competitiveness in the semiconductor sector 3.

3️⃣ The AI innovation race creates policy dilemmas between security and competitiveness

The Nvidia case highlights the fundamental tension between protecting national security and maintaining technological leadership.

China has overtaken the US in AI and machine learning patents since 2021 despite previous export restrictions, demonstrating the difficulty of slowing technological advancement through policy alone 2.

The global AI market is projected to reach $3.5 trillion by 2033, creating enormous economic incentives for nations to develop competitive advantages regardless of export controls 2.

The policy challenge is further complicated by the “full-stack” nature of AI that Huang described, where restricting one component (like GPUs) may prove ineffective when alternative approaches can be developed.

Experts gathering at the US Chamber of Commerce emphasized that 70% of businesses are expected to adopt AI by 2030, creating pressure for policies that balance security concerns with innovation potential 5.

This dilemma reflects the core challenge identified by the Information Technology Industry Council, which recommends that US leadership in AI requires strategic policy development focused on both innovation and competitiveness rather than primarily restrictive approaches 6.

Recent Nvidia developments

……