China's challenge to U.S. dominance in artificial intelligence is increasingly moving from private markets into publicly traded stocks, creating a new way for ETF investors to bet directly on the country's emerging AI model developers.
中國對美國在人工智慧領域主導地位的挑戰，正日益從私人市場延伸至公開上市股票，為ETF投資者提供了一種直接押注中國新興AI大模型開發商的新方式。
EMXETF's newly launched China AI Tigers LLM ETF (NASDAQ:TGRZ) is built around that shift. The fund targets Chinese companies developing foundational AI models rather than casting a wider net across semiconductors, cloud computing, e-commerce and other technology businesses.
EMXETF新推出的China AI Tigers LLM ETF（納斯達克：TGRZ）正是圍繞這一轉變構建的。該基金專注於投資開發基礎AI模型的中國公司，而非廣泛覆蓋半導體、雲計算、電子商務及其他科技業務。
The timing was driven partly by the January listings of Z.ai and MiniMax, according to EMXETF founder Kevin Carter.
據EMXETF創始人Kevin Carter表示，這一時機部分源於Z.ai和MiniMax在一月份的上市。
"The timing really came down to the IPOs — Z.ai and MiniMax both went public in January 2026, which suddenly made the China AI model trade investable," Carter toldBenzinga. He said the ETF was also designed to capture potential future listings from companies such as DeepSeek and Moonshot AI.
「時機的關鍵在於那些IPO——Z.ai和MiniMax均於2026年1月上市，這使得中國人工智能模型類股驟然具備了投資價值，」卡特向Benzinga表示。他還稱，該ETF還旨在捕捉DeepSeek、Moonshot AI等公司未來可能的上市機會。
That pipeline could become increasingly important.Moonshot, developer of the Kimi models, is reportedly eyeing a Hong Kong IPO and recently raised more than $2 billion. It is also in early talks withMicrosoft Corp(NASDAQ:MSFT),Amazon.com, Inc(NASDAQ:AMZN) andAlphabet Inc(NASDAQ:GOOGL) over potential revenue-sharing agreements involving its Kimi K3 model, Reuters reported.
這一上市渠道可能變得愈發重要。據稱，Kimi模型的開發商Moonshot正考慮在香港進行IPO，並 recently籌集了超過$20億的資金。路透社報導稱，該公司還正與微軟公司（納斯達克：MSFT）、亞馬遜公司（納斯達克：AMZN）和Alphabet Inc（納斯達克：GOOGL）就其Kimi K3模型的潛在收入分成協議進行初步洽談。
Carter defines China's "AI Tigers" as companies building foundational AI models, rather than businesses that simply use AI or supply the broader technology infrastructure. Many of these startups have links to Tsinghua University, which he describes as a breeding ground for some of China's leading AI labs. Z.ai, for instance, emerged from Tsinghua research, while Moonshot AI's founderYang Zhilinstudied there under Z.ai co-founder and Tsinghua professorTang Jie.
Carter將中國的「AI老虎」定義爲構建基礎AI模型的公司，而非僅使用AI或提供更廣泛技術基礎設施的企業。他指出，這些初創企業中的許多都與清華大學有關聯，而清華被視爲中國一些領先AI實驗室的搖籃。例如，Z.ai源自清華大學的研究，而Moonshot AI的創始人Yang Zhilin曾在清華大學師從Z.ai聯合創始人兼清華教授Tang Jie。
Carter Sees a Valuation Convergence Trade
Carter看好估值趨同交易
Carter argues investors may still be underestimating how far Chinese models have progressed.
Carter認爲，投資者可能仍低估了中國AI模型的發展水平。
"The biggest misconception is that China is still playing catch-up," he said. "Some Chinese models are now competing near or even at the frontier, and American companies are already building on and using them."
他表示：「最大的誤解是認爲中國仍在追趕。一些中國模型如今已接近甚至達到前沿水平，而美國公司已經在其基礎上進行開發並加以使用。」
More importantly for investors, Carter sees a valuation disconnect. He said no Chinese AI model developer is currently valued above $100 billion, while American AI labs are pursuing dramatically higher valuations.
對投資者而言更重要的是，Carter看到了一種估值錯位。他表示，目前沒有任何一家中國AI大模型開發商的估值超過$1,000億，而美國的AI實驗室則正追求高得多的估值。
"There is arguably aconvergence tradeahead," Carter said.
「可以說，前方存在一種趨同交易機會，」Carter表示。
That doesn't necessarily mean Chinese AI companies are cheap by conventional metrics. Z.ai, for example, generated revenue of 953.9 million yuan ($141.9 million) in the first half of 2026, up 400%, but still posted a 2 billion yuan ($297.6 million) net loss as R&D spending reached 2.1 billion yuan ($312.4 million), per Reuters. MiniMax, meanwhile, more than doubled on its Hong Kong trading debut in January, valuing the company at about $13.7 billion at the close.
這並不一定意味着中國AI公司按照傳統指標來看就便宜。例如，據路透社報導，Z.ai在2026年上半年實現收入9.539億元（$1.419億），同比增長400%，但由於研發支出達到21億元（$3.124億），仍錄得20億元（$2.976億）的淨虧損。與此同時，MiniMax今年1月在香港上市首日股價上漲逾一倍，收盤時公司估值約爲$137億。
A Different Bet from Traditional China Tech ETFs
與中國傳統科技ETF的不同押注
TGRZ, which carries a 0.86% expense ratio, isn't intended to provide broad China technology exposure.
TGRZ的費率比率爲0.86%，其目的並非提供廣泛的中國科技行業敞口。
"This is not a diversified China technology fund; it's a concentrated bet on the companies building China's leading AI models," Carter said.
Carter表示：「這不是一隻多元化的中國科技基金；而是對中國領先AI模型開發公司的集中押注。」
That distinction matters as China's AI strategy increasingly emphasizes open and lower-cost models. DeepSeek demonstrated that frontier-level AI development may not necessarily require U.S.-style spending, while Moonshot's latest Kimi models have further intensified competition.
這一區別至關重要，因爲中國的AI戰略正日益強調開放和低成本模型。DeepSeek已證明，前沿水平的AI開發未必需要採用美國式的高投入模式，而Moonshot最新的Kimi模型則進一步加劇了競爭。
Carter believes that cost advantage could ultimately broaden adoption.
Carter認爲，這種成本優勢最終可能擴大應用範圍。
"China is pursuing a different playbook — open, customizable and dramatically lower-cost AI that can spread far beyond China," he said. "There is tremendous room for both playbooks to grow."
他表示：「中國正在採取一套不同的策略——開放、可定製且成本大幅降低的AI，能夠遠遠超越中國本土市場。兩種策略都有巨大的成長空間。」
TGRZ will rebalance quarterly, giving it opportunities to add newly public AI developers. Carter said companies such as DeepSeek and Moonshot are precisely the type of future listings the ETF was designed to capture.
TGRZ將按季度進行再平衡，從而有機會納入新上市的AI開發商。Carter表示，像DeepSeek和Moonshot這樣的公司正是該ETF設計之初旨在捕捉的未來上市標的。 (Benzinga)