US President Donald Trump has vowed to stop the exports of high-end Nvidia AI chips due to their dual use in defence and several other strategically important industries. In a CBS interview, Trump said that only US companies will have access to Nvidia’s top-end chips. He said: We will not let anybody have them other than the United States. We don’t give the Blackwell chip to other people. His reactions followed Nvidia’s announcement of supplying over 2.6 lakh Blackwell chips to Samsung Electronics in South Korea. If that happens, many countries, primarily China, could be impacted as they heavily rely on them.
In an exclusive interview with Bhaskar English, Ajay Srivastava, former trade official and founder of the think tank Global Trade Research Initiative, explained that unlike China, which runs large-scale, cutting-edge AI systems powered by Nvidia’s A100, H100 and the new Blackwell-series chips. India’s AI programme currently operates just below that level and can still grow effectively using slightly older but powerful chips like Nvidia’s H20 and A800, which the US continues to allow for export. Here’s an edited excerpt of the interview: Q1. If Nvidia restricts top AI chips to the US, what economic and technological repercussions will India face in the short and long term? Ans: If the US continues to restrict Nvidia’s top AI chips, the A100, H100, and new Blackwell-series, for domestic use, the effect on India will be limited. Unlike China, which runs large-scale, cutting-edge AI systems that need these frontier GPUs, India’s AI programme operates just below that level and can grow effectively using slightly older but powerful chips like Nvidia’s H20 and A800, which the US still allows. These deliver strong performance for training language models, powering industrial and public-sector AI, and building affordable national computing grids. Key sectors such as AI startups, telecom, defence, and academia will continue progressing with these export-compliant chips. The ₹1.6 lakh crore semiconductor and IndiaAI Missions will also stay on track, as they focus on expanding compute access, local chip design, and manufacturing, rather than frontier supercomputing. In short, US curbs are aimed at China, not India. India’s AI journey will continue within what Washington allows, just below the global frontier, but still advancing steadily and strategically. Q2. Could this harm Nvidia’s global business? Or will countries be forced to comply because there’s no real alternative yet? Ans: Nvidia’s CEO has acknowledged that the company has lost about 95% of its China market due to U.S. export bans, an enormous blow, given that China once accounted for nearly a quarter of its AI chip sales.
While this will dent Nvidia’s global share, it also opens space for Huawei to catch up, as its Ascend chips are now just one generation behind the cutting edge and can already be fabricated on 7 nm nodes.
Still, most countries, including India, Japan, and the EU, will continue relying on Nvidia’s ecosystem for the foreseeable future, as no alternative yet matches its AI training performance, CUDA platform, or developer support. Q3. If Nvidia limits global supply, will we see a surge in grey-market chip sourcing, price inflation, and GPU smuggling? Ans: Some grey-market activity and price inflation are likely, but only at a small scale. A few resellers and data centers in regions like Hong Kong and the Middle East may trade limited quantities of older Nvidia GPUs at marked-up prices, yet strict US export tracking and licensing make large-scale “GPU smuggling” difficult. Minor leakages may occur, but they won’t meaningfully alter global supply or prices. The mainstream AI market will remain tightly regulated and largely compliant. Q4. Why is the US treating Nvidia chips almost like weapons-grade tech? Is this the new form of digital geopolitics? Ans: Absolutely. The US views Nvidia’s AI chips as strategic assets because they can train military-grade AI models, autonomous weapons, and surveillance systems. The H100 and Blackwell GPUs, built on TSMC’s 4 nm process and CoWoS packaging, are effectively the nuclear reactors of the AI age, small, powerful, and geopolitically decisive. Controlling who can access them has become a new form of digital statecraft, where computing power, not oil or steel, defines global influence. Conclusion
Trump’s stance shows the US intends to keep a tight hold on frontier AI computing power. While it could slow China’s progress, India is unlikely to face a major setback because it operates one step below the frontier and still has access to allowed Nvidia chips. Nvidia may take a global hit, but alternatives are not strong enough yet to trigger a shift. For now, the AI world will move at two speeds: the U.S. at the frontier, and others advancing within permitted limits.
The post Nvidia’s top chips are ‘Nuclear Reactors’ of AI age:Why India might not be in trouble even if Trump bans their exports of AI chips to desi shores appeared first on Tri-Cities India.
