Chip stocks are surging again as investors reconsider whether the latest wave of Chinese AI models poses a threat or actually strengthens the long-term investment case.
The rally has been driven in part by renewed debate around Kimi K3, an open-source AI model developed by China-based Moonshot AI that competes with leading closed U.S. models on certain benchmarks.
Kimi K3 achieves this despite operating under significant financial and technological constraints, which initially raised fears about AI hardware overspending across the industry.
Those fears appear to be fading quickly, with investors now viewing competitive open-source models as confirmation that AI infrastructure demand remains robust rather than inflated.
Nancy Tengler, CEO of Laffer Tengler Investments, compared the current moment to last year’s market selloff triggered by China’s open-source DeepSeek model, which she said created “a tremendous opportunity” for investors who bought the dip.
Shares of memory maker Micron Technology (MU) surged 12.2% on Tuesday, while Sandisk jumped 14.4% and SK Hynix’s American depositary receipts climbed 13.8%.
Storage makers Western Digital (WDC) and Seagate Technology (STX) also posted sharp gains, rising 12.5% and 11.1%, respectively, as the broader chip sector caught a strong bid.
Bank of America analyst Vivek Arya said China’s recent open-source model releases support his “bullish thesis on memory” and specifically on Micron’s stock going forward.
Arya noted that Chinese companies are charging developers far less than Western firms to access their open models, but he views that “as reflective of business-model choices,” and not of the underlying cost of hardware.
The critical insight from Arya is that even efficiency-focused Chinese models still “require the same or more memory as their model weights and active parameters increase,” making memory demand a structural rather than cyclical story.
Kimi K3 itself illustrates this dynamic directly, boasting 2.8 trillion parameters in total, with 50 billion of those parameters active at any given time.
A model’s parameters are the variables it uses to learn and recognize patterns during training, functioning as a form of memory that scales with model sophistication.
That scaling dynamic is precisely why analysts like Arya believe memory chipmakers stand to benefit regardless of whether AI development is led by open-source or proprietary closed models.
The latest rally suggests the broader market is coming around to that view, treating each new wave of AI model releases as a demand catalyst rather than a reason to reduce exposure to chip stocks.
