Is AI Growth Slowing Down? Memory Industry Expert Says It’s ‘Too Early’ To Tell

EXCLUSIVE: Is AI Growth Slowing Down? Memory Industry Expert Says It’s ‘Too Early’ To Tell

Benzinga and Yahoo Finance LLC may earn commission or revenue on some items through the links below.

Memory chip stocks had a punishing week. Yet, the shortage that powered their rally has not gone away. That disconnect is whatMS Hwang, research director at Counterpoint Research, sees in the sell-off.

“While most of these negative indicators miss the point or exaggerate, it is still too early to conclude that AI growth is slowing down,” Hwang told Benzinga on Friday.

• A single bad hire can set a startup back years. Here are the 5 hires founders most often misjudge — and why

• Still Learning the Market?These 50 Must-Know Terms Can Help You Catch Up Fast

So far, the damage has been steep:

• SK Hynix Inc. fell 24.3% in Seoul this week, even after a 0.8% gain on Friday. The stock now trades about 57% below its record high.

• Suwon, South Korea-based Samsung Electronics Co. Ltd. lost 17.7% over the week. South Korea’s benchmark KOSPI index dropped 14.5%, despite a 2% rebound on Friday.

• Boise, Idaho-based Micron Technology Inc. closed Thursday at $931.24. That is down 9.5% over five sessions and about 26% below its record of $1,255. Sandisk Corp. fell 11.8%.

• The Roundhill Memory ETF is down 14% for the week, heading for its worst weekly performance since late June.

China Is Not The Real Trigger

The sell-off began after China’s ChangXin Memory Technologies, or CXMT, reported progress on a denser chip design for server memory, the market’s most profitable segment.

Hwang does not think Chinese capacity explains the drop.

“Memory supply is unlikely to have a major impact before 2028,” he said.

China’s progress may surprise investors who doubted it, he added, but “it wouldn’t trigger a sell-off.”

Trending:Avoid the #1 Investing Mistake: How Your ‘Safe’ Holdings Could Be Costing You Big Time

Instead, he pointed to the financial side: inflation, interest rates and funding costs, which could slow the flow of capital into AI.

Rumors added pressure.

These included talk of companies delaying back-end investment and of less high-bandwidth memory, or HBM, per AI system. HBM is the stacked memory that sits next to AI processors.

Broadcom’s Comment Changed The Mood

If one data point explains the sell-off, Hwang said, it is likely Broadcom Inc.’s recent remarks.

At a panel in Singapore on Monday, CEO Hock Tansaid capital, not wafers or memory, would be the company’s biggest constraint as AI development continues.

According to Hwang, that message may have created the impression that it is braking AI spending.