AI Chip Crash: Why Cheap Chinese Models are Shaking the Market

South Korea’s Kospi Triggers Circuit Breakers as Global AI Stocks Crater

The correction caught fire following intense competitive pressure from low-cost Chinese alternatives.

South Korea’s benchmark Kospi index cratered more than 10 percent during the session, forcing a temporary trading suspension. Memory chip titans took the heaviest direct hits. Samsung shares nosedived 12 percent, while SK Hynix slumped 12.7 percent. The contagion quickly spread westward, dragging down Micron, Advanced Micro Devices, and Intel.

The Economics of Multibillion-Dollar Bets and Cheap Chinese Models

At the heart of the rout lies a sudden wave of skepticism regarding the economic viability of multibillion-dollar AI investments. This unease was sharply accelerated by the rise of economical Chinese models like Kimi and DeepSeek.

Sentiment soured almost overnight following reports that Nvidia might guarantee up to $250 billion in OpenAI loans to fund a sprawling data center in Ohio. Traders quickly grasped a harsh reality: if artificial intelligence utility becomes cheap, chipmakers and cloud providers risk severely compressed profit margins.

These punishing pullbacks centered squarely on equities that had surfed the crest of the recent tech rally.

UBS Research Paints a Divergent Picture of Commercial Adoption

Yet a research note published by UBS prior to the rout tells a fundamentally different story about actual enterprise adoption. UBS analyst Karl Keirstead interviewed more than 15 artificial intelligence firms, including Perplexity and Cursor.

His findings revealed that business operations are expanding at a breakneck pace, with select companies doubling or even tripling their revenues over a six-month window. Rather than pulling back, organizations are aggressively managing costs. They now route straightforward, routine tasks to affordable open-source models while keeping expensive proprietary engines reserved for complex challenges.

Economic history offers a clear parallel. When foundational utilities like electricity or mobile telecommunications experience falling costs, overall consumption surges rather than contracts. Lower pricing unlocks an expansive web of commercial applications, driving aggregate sector demand upward.

Hardware Giants and Cloud Hyperscalers Outlast Token-Pricing Pressures

Market positioning now dictates who survives this structural shift. UBS identifies hardware manufacturers and major cloud operators as the primary structural beneficiaries.

Nvidia retains a commanding grip on the ecosystem because open-source models overwhelmingly run on Nvidia graphics processing units. Simultaneously, hyperscale cloud giants Amazon, Microsoft, and Google stand to gain as enterprise clients deploy multiple AI models at once across their sprawling infrastructure.

The pain falls elsewhere. Leading foundation model developers like OpenAI and Anthropic face slowing momentum as routine workloads flee to cheaper options. Ultimately, corporations controlling fully integrated ecosystems—spanning silicon, cloud capacity, proprietary models, and direct distribution channels—are proving best insulated from pure token-pricing pressures.