On Thursday, China’s Moonshot unveiled Kimi K3, a massive AI model boasting 2.8 trillion parameters, claiming parity with Anthropic’s Claude Fable and OpenAI’s GPT-5.6. This release rattled US chip stocks, raising questions over America’s grip on the AI market.

The Philadelphia Semiconductor Index, a key benchmark for US chipmakers, plunged 12.5% this past week, marking its sharpest decline in over 15 months. Major players like Nvidia, AMD, and Broadcom suffered steep drops despite the index maintaining a year-to-date gain exceeding 60%. Investors now face a dilemma: is the AI rally losing steam as foreign competitors close the gap?

Kimi K3’s Market Impact

Kimi K3 is the largest open AI model ever released, available for free download starting July 27 through Moonshot, an Alibaba-backed startup. Its sheer size and performance disrupted expectations, topping Arena’s coding leaderboard with 1,679 points and surpassing Claude Fable 5. Importantly, its cost efficiency $3 per million input tokens compared to Fable 5’s $10 introduces a competitive pricing dynamic that could pressure existing AI service providers.

However, the backlash was not limited to the US. Chinese AI stocks like Zhipu and MiniMax also tumbled in Hong Kong trading, falling 28% and 16% respectively, reflecting investor anxiety over the sector’s sustainability. The recurring theme of an AI bubble resurfaced, echoing the market shock in January 2025 when China’s DeepSeek led to Nvidia’s single-day loss of $589 billion, a historic event spotlighting geopolitical tensions intertwined with technology rivalry.

Market commentators like Jim Cramer emphasize that beyond technology, trust remains the critical barrier. Persistent mistrust between the US and China complicates data sharing and cooperation, intensifying competitive postures in AI development and chip manufacturing.

Amid this tension, derivatives linked to AI computing power are gaining traction. Crypto-style derivatives on AI compute are now traded by Bernstein, CME Group plans compute futures with Silicon Data, and ICE has announced GPU contracts with Ornn, signaling new financial instruments tied to AI hardware demand.

Moonshot’s founder Zhilin Yang outlined strategies for AI scaling focusing on token efficiency, wider context windows, and parallel agent swarms, underlining technical innovation driving the model’s capabilities.

This material is informational and not financial advice.