Nvidia chief executive Jensen Huang used an interview at his company’s GTC Berlin conference to push back on the Nvidia memory glut debate: fears that the memory chips feeding AI accelerators will be in oversupply by 2029. “You can always use more,” Huang told CNBC’s Arjun Kharpal, according to CNBC.
Huang aimed the remark at a sell-off that has hit memory makers over the past week. Investors are debating whether the AI build-out will peak before the supply chain finishes expanding. If it does, a wave of capacity could arrive without enough buyers. Huang’s answer is that memory is not a commodity that loses its usefulness when capacity rises. HBM, the high-bandwidth memory stacked beside Nvidia’s GPUs, will remain in use for years, he said. Nvidia also plans to keep buying large amounts of LPDDR, the phone-grade memory it has moved into data centers.
What Huang actually said
Huang framed the debate as a demand question rather than a supply question. He said Nvidia’s use of LPDDR has helped it deliver performance that is twice as energy efficient as anything else, and that the company will continue to consume abundant amounts of memory. He did not give a forecast for prices or a specific supply target.
The memory comments came during a wide-ranging interview that also covered China and export controls. In a separate interview at the same conference, Huang said demand for neoclouds was “off the charts,” CNBC reported. Nvidia is trying to expand its AI business in Europe, where it has committed nearly $10 billion to companies, universities and research institutions, CNBC reported. The company’s messaging is that the build-out is still early, even after two years of explosive spending.
Why the Nvidia memory glut fear spread
The sell-off began after an opinion column in a South Korean publication on Oct. 15. It urged the government to prepare for an AI slowdown. The column cited two experts who expected a flood of new memory supply by 2029. That would turn the market into a “red ocean” and possibly end the current supercycle, according to CNBC’s translation.
The warning landed on a sector already nervous about capacity. Memory is a core input for Nvidia’s chips, and an oversupply can push prices down and squeeze margins at SK Hynix, Samsung and Micron. China’s ChangXin Memory Technologies added to the anxiety with a new design architecture. The design could improve efficiency and help a Chinese supplier win share from the established players.
Elon Musk also weighed in, predicting that China would produce enormous amounts of RAM in 2028, Quartz reported. The combination of a supply warning, a Chinese capacity story and a stretched AI trade was enough to knock memory stocks lower even before Huang spoke.
Huang did not dismiss the Chinese threat. Asked whether Nvidia would partner with CXMT, he said the company is open-minded. He added that Nvidia works with Chinese companies, as much of the global supply chain does. He also repeated his view that U.S. export restrictions have accelerated China’s push to build its own semiconductor industry. China’s chip sector has been trying to reduce its reliance on American hardware, a theme covered in earlier reporting on Ant Group.
That creates an awkward backdrop for his supply argument. If Chinese memory capacity grows quickly, the world may be able to make more memory than Nvidia’s customers need. The real question is whether they will keep buying at current prices. Huang’s bet is that AI workloads will absorb the extra supply, especially as inference spreads from data centers to devices and smaller systems.
What to watch
The next test is whether memory makers confirm Huang’s demand picture in their earnings and guidance. Investors will also watch Nvidia’s own supply commitments. If those commitments keep rising, it is a sign that the company still expects to be short of components rather than long. For now, Huang has made his position clear: the Nvidia memory glut debate is premature, and Nvidia plans to remain a very large buyer.
I am a person who is positive about every aspect of life.I have always been an achiever be it academics or professional life. I believe in success through hard work & dedication.
Technology Blogger at TechnoSecrets.com
