Jensen Huang rejects 2029 memory-glut fears as chip stocks sell off

After a week of selling in memory stocks, Nvidia CEO Jensen Huang pushed back on predictions that the industry faces a surplus by 2029. In an October 22, 2026 CNBC interview, he argued that AI systems would keep using large amounts of memory—a bullish demand outlook as investors worry about new supply and stronger Chinese competitors.

The warning that rattled memory stocks

The immediate backdrop was an October 15 opinion column in a South Korean publication. According to CNBC’s translation, two experts cited in the column predicted a flood of new memory supply and warned that the industry’s strong cycle could give way to a downturn by 2029. Their concern was about future market conditions, not an oversupply already in place.

Memory shares fell over the following week amid fears of oversupply and weakening demand. The stakes extend to SK Hynix, Samsung and Micron: memory is a core component of Nvidia’s chips, and excess supply could push prices lower and squeeze the companies that manufacture it.

China’s ChangXin Memory Technologies, or CXMT, added another source of investor anxiety. Its announcement of a new memory design architecture raised concerns that greater efficiency could help it take market share from those three established suppliers. That put potential Chinese competition alongside the broader question of how much memory the market will need.

Huang’s answer starts with demand

Speaking with CNBC’s Arjun Kharpal on the sidelines of Nvidia’s GTC Berlin conference, Huang said he saw no sign of a 2029 glut. His answer emphasized continued use rather than a prediction that supply expansion would stop. He said high-bandwidth memory, or HBM, would remain in use for years to come.

He also pointed beyond HBM to LPDDR, the kind of memory used in mobile phones. Huang claimed Nvidia was the first company to put it into data centers and said the approach enabled performance twice as energy-efficient as anything else available. That comparison is Nvidia’s claim, not an independently verified result.

An open door to Chinese suppliers

Asked whether Nvidia planned to partner with CXMT and use its memory, Huang said the company was “open-minded.” He noted that Nvidia works with Chinese companies and that Chinese businesses are present across global supply chains. The answer expressed a willingness to consider the supplier; it was not an announcement of a partnership or purchase.

Huang praised China’s capabilities in fundamental infrastructure technology and manufacturing, saying they had to be taken seriously. Those remarks addressed the competitive concern behind the sell-off: CXMT’s progress could matter not just as additional supply, but as a challenge to the suppliers investors already know.

He also blamed U.S. restrictions for speeding China’s semiconductor development. The United States restricts exports of its most advanced semiconductors to China, including top-end Nvidia chips. Huang argued that those rules accelerated China’s domestic manufacturing initiatives and the industrial support directed toward the sector. That was his assessment of the restrictions’ consequences.

Huang nevertheless said American technology companies were prepared to compete with anyone. His position leaves room for both competition and sourcing from China: he took its manufacturing strength seriously while remaining open to using its products. A concrete CXMT agreement would be a further step beyond the openness he expressed in this interview.