Reports suggest that Baidu may be preparing to spin off an asset that could be worth more than itself.

According to a June 29 report from The Information, Baidu's AI chip company, Kunlunxin, is planning a Hong Kong listing with a target valuation of approximately $50 billion. When contacted for comment, Baidu did not respond.

Further rumors indicate that Tencent has become a customer of Kunlunxin, and ByteDance is also considering adopting its AI chips. While there has been no official response, an informed source has stated that the news regarding ByteDance is inaccurate.

Despite this, market attention remains firmly fixed on the potential IPO itself. If successful, the listing would not only be one of the largest AI chip IPOs in Hong Kong this year, but Kunlunxin's target valuation would also exceed the current market capitalization of its parent, Baidu.

As of the close on June 30, Baidu's Hong Kong-listed shares were trading at HK$109.6, giving the company a total market value of around HK$298.3 billion. In comparison, a $50 billion valuation for Kunlunxin would be approximately 1.3 times the size of Baidu's market cap.

An industry analyst noted that the current valuation surge for AI chip companies stems from demand for domestic alternatives due to computing power shortages, technological advancements, and market growth expectations. As data center construction accelerates and pilot orders transition to bulk deployments, the commercial viability of these firms is strengthening, leading capital markets to assign higher growth premiums.

Reaching a Lofty Valuation Requires Moving Beyond Baidu

Public records show Kunlunxin originated from Baidu's internal smart chip and architecture department. It completed an independent financing round in April 2021, spinning off from Baidu to operate on its own.

Since then, Kunlunxin has accelerated its market and capital expansion. Corporate records indicate the company has progressed through to a Series D financing round within three years of its Series A in June 2022. Investors include CPE Yuanfeng, IDG Capital, Legend Capital, and BYD, though specific funding amounts for each round have not been disclosed.

In April of this year, Kunlunxin completed filing for listing tutoring with the Beijing Securities Regulatory Bureau, targeting an initial public offering on Shanghai's STAR Market. The new reports of a Hong Kong listing suggest the company may be pursuing a dual-track "A+H" listing strategy.

On the product front, Kunlunxin has launched several AI chips, including the Kunlunxin II, P800, and M100, which cover scenarios from model training to inference. At a recent developer conference, a Baidu executive stated that the P800 chip has completed large-scale validation, with multiple 10,000-card clusters based on the P800 delivered since 2025.

The critical question is whether a $50 billion valuation is realistic for a domestic AI chip company yet to go public. One financial analyst argued that viewed purely as a fabless chip designer, $50 billion seems expensive. However, if evaluated as a composite asset encompassing Baidu's AI cloud, domestic computing power substitution, inference infrastructure, and a software/hardware ecosystem, the valuation becomes more debatable.

He suggested that valuing domestic AI chip firms requires a multi-dimensional approach, considering factors like the transition from prototypes to mass deployment, customer diversification beyond related parties, software ecosystem maturity, and cost per token.

Kunlunxin's key advantage is that it didn't start in a lab but grew within Baidu's ecosystem encompassing search, intelligent cloud, large models, and the PaddlePaddle framework, and is now expanding to external clients. The $50 billion figure represents a "strategic scarcity valuation," not a traditional financial one. Its realization hinges on proving it is not just an internal Baidu supplier but a foundational AI computing infrastructure company for China.

Another analyst echoed that while profitability remains weak industry-wide, leading to a focus on price-to-sales ratios, Kunlunxin must consistently demonstrate scalable revenue generation and further reduce its reliance on the Baidu ecosystem to justify a high valuation.

Linking Investment to Purchase Orders

While capital is once again assigning high valuations to AI companies, the criteria are shifting. Orders, customers, and cash flow are becoming crucial metrics for reassessing value.

Media reports suggest that during its IPO roadshow, Kunlunxin has implemented an "investment + procurement" mechanism, prioritizing investors willing to purchase its chips. Some participating institutions are reportedly required to procure chips worth three to seven times their investment amount.

An analyst views this model as essentially using equity investment to lock in long-term orders, a practice not entirely unprecedented in the semiconductor sector. For AI chip companies with high R&D costs and long commercialization cycles, securing customers and orders in advance can boost market confidence in their commercial capabilities.

However, he noted that binding procurement value so closely to investment size is relatively rare in domestic IPOs. While it may enhance order certainty in the short term, genuine market demand must support it long-term; otherwise, the market may reassess the quality of revenue and the valuation.

Another expert emphasized that the market's ultimate concern is not the volume of orders generated during the IPO but whether they reflect real market demand. The key determinants for achieving the $50 billion valuation will be whether chips are genuinely deployed, customers make repeat purchases, and orders convert into stable cash flow.

For Kunlunxin, the market is repricing not just AI technology, but the ability to turn AI into a sustainably profitable business.

The Evolving Logic Behind AI Investment

Over the past few years, large language models were the hottest AI investment theme. The market focused on model capability, parameter scale, and technological leadership. However, as competition enters the commercialization phase, this logic is changing.

On one hand, model capabilities improve while prices fall, shifting competition from pure technology to commercial implementation. On the other hand, demand for computing power continues to grow with the proliferation of AI agents, inference models, and generative AI applications.

Rather than a simple shift from models to chips, capital is moving from betting on individual model brands to betting on the entire AI production system. While models can be discounted and applications iterated, computing power infrastructure remains the foundational layer of the entire industry chain. As AI applications develop, demand for computing power intensifies, making the sustained growth of this demand the core investment thesis.

But chips are not the final destination. Market focus continues to evolve. According to an angel investor, while attention was on GPUs last year, capital this year is already flowing into other AI infrastructure segments like memory and inference optimization, indicating a search for directions with more sustainable growth potential within the AI industry chain.

Kunlunxin's potential IPO arrives precisely during this window of industry value reassessment.