Melius Research's Ben Reitzes Says Chip Stocks Selloff Was The 'Exclamation Point' — 'Semis Are Ripping, And I Think They Should Be'

During an interview with CNBC, Reitzes said investor concerns over AI returns on investment, open-source models and forced selling had clouded the outlook for chipmakers.

• Reitzes said the recent correction was fueled by concerns that open-source AI models could compress profit margins at leading AI labs and eventually reduce infrastructure spending.
• He said those fears, coupled with volatility in Korea's semiconductor market, led to indiscriminate selling across the sector.
• Reitzes called the selloff a buying opportunity, citing "outstanding" checks at OpenAI and Anthropic that support Nvidia CEO Jensen Huang's view that open and closed AI models will coexist.

The recent correction in semiconductor stocks created a compelling buying opportunity, according to Melius Research’s Head of Technology Research Ben Reitzes, who described the selloff as the “exclamation point” of the latest AI pullback and said the sector is now rebounding on stronger fundamentals.

During an interview with CNBC, Reitzes said investor concerns over AI returns on investment, open-source models and forced selling had clouded the outlook for chipmakers. But with semiconductor stocks recovering, he believes those fears have eased. “Semis are ripping, and I think they should be,” he said.

The Direxion Daily Semiconductor Bull 3X (SOXL) ETF was up more than 18% at the time of writing, while the Direxion Daily Semiconductor Bear 3X (SOXS) ETF fell nearly 18%.

Forced Selling Created A Buying Opportunity, Says Reitzes

Reitzes said the recent correction was fueled by concerns that open-source AI models could compress profit margins at leading AI labs and eventually reduce infrastructure spending.

He said those fears, coupled with volatility in Korea's semiconductor market, led to indiscriminate selling across the sector.

However, he also noted that the selloff created a buying opportunity. Recent checks at OpenAI and Anthropic remain “outstanding,” he said, reinforcing the view championed by Nvidia Corp. (NVDA) CEO Jensen Huang that both open and closed AI models will coexist.

As companies such as Microsoft Corp. (MSFT), Amazon.com Inc. (AMZN) and Palantir Technologies Inc. (PLTR) improve model-routing capabilities, Reitzes expects infrastructure demand to remain strong regardless of which models ultimately win.

Reitzes Says Memory Got ‘Reinvented’ Thanks To AI

Reitzes argued that memory stands to benefit disproportionately from the next phase of AI adoption because larger memory pools improve model performance, enable longer context windows and make AI applications more responsive.

“Memory got reinvented in 2023 when AI happened,” he said, adding that the proliferation of open-source models running locally on PCs and other edge devices could send RAM demand “berserk.”

While additional supply is expected to arrive in 2028, Reitzes said “the commentary we're going to hear in memory over the next six quarters is going to be pretty darn good.”

Reitzes also pointed to the newly announced High Bandwidth Flash (HBF) technology from SanDisk Corp. (SNDK) and SK Hynix Inc. (SKHY) as evidence that memory technology continues to evolve alongside AI demand.

On Tuesday, the companies unveiled the industry’s first HBF standard specifications, describing the technology as a new memory layer between high-bandwidth memory (HBM) and solid-state drives (SSDs) that combines high-speed data transfer with significantly greater storage capacity.

They said the technology is designed to help overcome AI's growing “memory wall” by addressing both bandwidth and capacity constraints as inference workloads scale.

AMZN’s Andy Jassy Did Everyone A ‘Favor’, Says Reitzes

Reitzes said Amazon CEO Andy Jassy clearly articulated the long-term economics behind the company’s AI investments.

“Andy Jassy did everybody a favor,” he said, arguing that Amazon moved beyond vague commentary and explained how AI spending is expected to generate attractive returns over time.

Reitzes dismissed criticism that AI lacks a measurable return on investment and argued that Jassy's confidence in building a trillion-dollar AI business reinforces the case for sustained hyperscaler capital spending.

The Invesco QQQ Trust (QQQ) is up 27% over the past 12 months, while the VanEck Semiconductor ETF (SMH) is up 97%.

For updates and corrections, email newsroom[at]stocktwits[dot]com.<