One Sentence From Jensen Huang Sent Chinese AI Stocks Soaring

One sentence. Billions of dollars in market cap.

When Nvidia CEO Jensen Huang told the crowd at GTC on Tuesday that OpenClaw was "definitely the next ChatGPT," he may not have anticipated the ripple effect that would hit Hong Kong by Wednesday morning. Chinese AI firm MiniMax surged 22%. Zhipu (Knowledge Atlas Technology) jumped 14%. SenseTime added 2.43%. Shanghai-listed cloud firm UCloud Technology climbed 13%.

And it didn't stop there. Huang's separate forecast that purchase orders between Blackwell and Vera Rubin chips could hit $1 trillion by 2027 sent SK Hynix up nearly 9% and Samsung Electronics up 7.53% in Seoul.

One conference. One afternoon. A lot of portfolios moved.

What Is OpenClaw, and Why Does It Matter

OpenClaw isn't a chatbot. It's an AI agent — software that can take instructions and carry out multi-step tasks autonomously: browsing the web, writing and executing code, managing files, interacting with external services. Think of it less as a conversation partner and more as a digital employee that actually does things.

Crucially, it's open-source. That means any company can take the underlying model, adapt it, and deploy it in their own products — no licensing fees, no dependency on a single vendor. Chinese tech firms have moved fast on this. MiniMax and Zhipu both recently launched agentic AI tools built on OpenClaw. SenseTime, which has been pivoting away from its controversial facial recognition roots toward AI software platforms, integrated one of its AI assistants with OpenClaw as well.

Zhipu went further last month, releasing GLM-5, an open-source large language model the company claims approaches Anthropic's Claude Opus 4.5 on coding benchmarks and outperforms Google's Gemini 3 Pro in certain tests. CNBC noted it hasn't independently verified those figures — a caveat worth keeping in mind when parsing competitive AI benchmarks, which have a long history of being selectively framed.

Sponsored

Advertise with Us

[email protected]

Sponsored

Winners, Losers, and the Investment Angle

The clearest winners from Wednesday's moves are the companies that had already positioned themselves inside the OpenClaw ecosystem. MiniMax and Zhipu had done the work before Huang said a word — his endorsement simply accelerated the market's recognition of that bet.

For semiconductor investors, the $1 trillion chip order forecast is the more durable signal. SK Hynix and Samsung both supply high-bandwidth memory (HBM) that powers the AI infrastructure buildout Huang is describing. If that forecast holds even partially, demand for their chips doesn't slow down anytime soon.

The less obvious losers? Closed-source AI companies — primarily OpenAI and Anthropic — whose subscription-based business models depend on users not having comparable free alternatives. If open-source agents can match or approach the performance of paid products, the pricing power of the incumbents erodes. Huang calling an open-source agent "the next ChatGPT" is, intentionally or not, a challenge to OpenAI's market position.

There's also a geopolitical irony worth noting: Nvidia is subject to US export controls that restrict selling its most advanced chips to China. Huang is publicly praising the capabilities of Chinese AI firms that have largely had to work around those restrictions — building competitive models on older hardware, open-source frameworks, and domestic alternatives. That tension isn't going away.

The Bigger Shift: Hardware Walls vs. Software Ecosystems

Moody's recently observed that "China's rapid uptake of artificial intelligence reinforces its position as one of the world's leading AI markets," while noting that adoption remains uneven — large tech firms leading, consumer and industrial sectors following more slowly.

That unevenness matters. The Chinese companies making headlines right now — MiniMax, Zhipu, SenseTime — are concentrated in the AI software and model layer, not hardware. The US strategy of restricting chip exports assumes that compute is the binding constraint on AI development. But DeepSeek's efficient low-cost models last year, and now OpenClaw's agentic ecosystem, suggest that software innovation can partially compensate for hardware limitations.

This doesn't mean export controls are ineffective. It means the relationship between hardware access and AI capability is more complex than a simple bottleneck model suggests. The gap may be narrowing in some dimensions while remaining wide in others.