CNBC ETF Edge focused on the surge in semiconductor and memory-chip stocks, especially Micron and the Roundhill Memory ETF, and then shifted to the broader debate over prediction market ETFs and their regulatory hurdles.
Watch on YouTube ›Get the market thesis, key claims, assets, contradictions, and follow-up questions from any financial video — then unlock a version personalized to your portfolio, watchlist, and favorite speakers.
Host Contessa Brewer, filling in for Dominic Chu, spoke with Dave Maza of Roundhill Investments and Drew Pettit of Citi Research about two main themes: the AI-driven rally in semis/memory chips and the prospects for prediction market ETFs. On semis, both guests argued that memory chips are now a key bottleneck in the AI buildout, with demand from data centers, hyperscalers, and Nvidia-related infrastructure creating a supply/demand imbalance that could persist for years because new fabrication capacity takes time to come online. Maza emphasized that memory has historically been cyclical but is being rerated because demand is becoming more durable and contract durations are lengthening; he also highlighted related “bottleneck trades” such as optical networking names and industrial enablers for data center construction. …
Tactically, the momentum is still with memory-chip and AI infrastructure names, with pullbacks likely to be bought as long as earnings revisions stay positive. Near-term risk comes from overheated positioning, regulatory delays on prediction-market products, and any oil-driven shock that dents sentiment.
Over the next few months, the base case is continued outperformance of memory and adjacent AI bottlenecks if data-center demand, pricing power, and analyst revisions hold up. The view would weaken if supply normalizes faster than expected or if hyperscaler capex decelerates materially.
Structurally, the transcript argues that AI has made memory and related infrastructure a more durable capital-allocation theme than the old cyclical semiconductor playbook. If prediction markets survive the regulatory gauntlet, they could also mark a broader expansion of what investors can package and trade inside ETFs.
Memory chips are now the biggest bottleneck in the AI buildout.
Dave Maza explicitly says investors are waking up to this bottleneck and ties it to the rally in memory-related stocks.
The memory-chip supply/demand imbalance could last into 2027 or 2028 because new fabs take three to five years to build.
This is the core duration argument for the trade remaining supported.
Micron has shifted from a consumer-oriented memory business to a much more data-center-driven business.
Used as an example of why the cycle may be changing and contracts becoming longer term.
What is driving the recent strength in semis, especially memory chips and DRAM?
Dave Maza says investors are realizing the biggest AI bottleneck is memory chips, where supply and demand are badly mismatched. He says demand from data centers and AI is pushing the business away from its old consumer-driven cyclicality.
Why do you think momentum in DRAM and semis can continue?
Drew Pettit says the price gains are backed by earnings momentum, with the best earnings revisions in the U.S. and globally coming from this group. He points to Micron and other DRAM names as adding meaningful earnings support to the market.
Are these stocks still reasonably priced despite the huge run-up?
Drew Pettit says yes, because earnings expectations have risen sharply enough to offset the price move. He argues that if earnings are now expected to be much higher over the cycle, the stocks can still screen as reasonably priced.
Unlock the full claims, asset map, scores, related transcripts, follow-up questions, and AI chat — shaped around your portfolio, watchlist, favorite speakers, and risks.
