All posts

Baidu's Kunlunxin eyes $50B Hong Kong IPO with chip buy-in twist

Manaal KhanJune 29, 2026 at 12:32 AM4 min read
Baidu's Kunlunxin eyes $50B Hong Kong IPO with chip buy-in twist

Key Takeaways

Baidu's Kunlunxin eyes $50B Hong Kong IPO with chip buy-in twist
Source: Tech-Economic Times
  • Kunlunxin targets a $50 billion valuation for its Hong Kong listing, making it one of the largest chip IPOs in recent years
  • Investors must reportedly purchase Kunlunxin chips worth 3-7 times their planned share subscription to participate
  • ByteDance is considering using Kunlunxin chips, while Tencent is already a customer

Baidu's AI chip division Kunlunxin is targeting a $50 billion valuation for its Hong Kong public offering, according to The Information. But there's an unusual condition attached: investors reportedly must buy Kunlunxin chips worth three to seven times their planned share subscription to participate.

The report, citing two sources, could not be independently verified by Reuters. Baidu has not responded to requests for comment. If the valuation holds, Kunlunxin would rank among the largest chip company listings globally in recent years.

Why require chip purchases from IPO investors?

The chip purchase requirement is unconventional. Most IPOs simply ask investors to commit capital in exchange for shares. Requiring them to also become customers creates immediate revenue and signals genuine commercial demand for Kunlunxin's products.

It also filters for strategic investors. A venture fund might buy shares speculatively. A company that commits to purchasing $300 million in chips for a $100 million allocation has operational skin in the game. The structure suggests Baidu wants investors who will integrate Kunlunxin into their supply chains, not just their portfolios.

Advertisement

Who's already buying Kunlunxin chips?

Tencent is already a Kunlunxin customer, according to one of The Information's sources. More significant: ByteDance is reportedly considering using Baidu's chips. If TikTok's parent company signs on, it would validate Kunlunxin as a supplier beyond Baidu's internal operations.

Kunlunxin has been selling externally for about two years now, but the bulk of its shipments still go to Baidu's own search, cloud, and autonomous driving operations. Landing ByteDance would be a major expansion of its addressable market.

Advertisement

Beijing's tech self-reliance push

China's onshore technology IPOs are heading toward their strongest year since 2023. The government has been actively encouraging listings from chip and AI companies as part of its strategy to reduce dependence on foreign semiconductors, particularly American ones.

U.S. export restrictions have made this push urgent. Nvidia's high-end AI chips are banned for sale to Chinese companies, which has accelerated demand for domestic alternatives. Kunlunxin, with its Kunlun II chips built on 7nm process technology, represents one of China's more advanced homegrown options.

Baidu filed confidentially for the Hong Kong listing in January. The company founded Kunlun as an internal unit in 2017 to build chips specifically for AI workloads like search ranking and natural language processing. It spun Kunlunxin into a separately operated company while retaining a controlling stake.

Can Kunlunxin compete with Nvidia?

Directly? No. Nvidia's H100 and newer chips remain the gold standard for training large AI models. But Kunlunxin doesn't need to match Nvidia's raw performance to find a market. It needs to be good enough for inference workloads, available in China, and cheaper.

The company's Kunlun II chip, launched in 2021, achieved 2-3x the performance of its predecessor. For Chinese companies locked out of Nvidia's latest hardware, a domestic chip that handles production AI workloads reasonably well is worth paying for.

The $50 billion valuation, if achieved, would price Kunlunxin at roughly one-third of AMD's current market cap. That's aggressive for a company with limited external sales, but it reflects both China's strategic interest in semiconductor independence and investor appetite for AI plays.

Frequently Asked Questions

What is Kunlunxin?

Kunlunxin is Baidu's AI chip division, founded in 2017 to develop semiconductors for AI workloads. It operates separately from Baidu but remains majority-owned by the parent company.

Why does the Kunlunxin IPO require investors to buy chips?

According to the report, investors must purchase Kunlunxin chips worth 3-7 times their share subscription. This creates immediate revenue and ensures investors have commercial, not just financial, interest in the company's success.

Which companies use Kunlunxin chips?

Baidu is the primary customer for internal use. Tencent is already a Kunlunxin customer, and ByteDance is reportedly considering using the chips.

How does Kunlunxin compare to Nvidia?

Kunlunxin's chips don't match Nvidia's top-tier AI training hardware, but U.S. export restrictions prevent Chinese companies from buying Nvidia's best chips. Kunlunxin serves as a domestic alternative for AI inference and some training workloads.

When will Kunlunxin go public?

Baidu filed confidentially with the Hong Kong stock exchange in January 2025. The exact timing of the IPO has not been announced.

ℹ️

Logicity's Take

The chip purchase requirement is the real story here. Baidu is essentially pre-selling product as a condition of investment, which hedges against the risk that public market investors will flip shares without caring about the underlying business. If ByteDance signs on as a customer, Kunlunxin becomes a genuine merchant chipmaker rather than a captive supplier. Competitors in China's AI chip space include Huawei's Ascend line and startups like Enflame and Biren. None have achieved Nvidia's ecosystem lock-in, which means the market remains fluid. For CTOs at Chinese tech companies, the question isn't whether to evaluate domestic chips but which ones will still be around in five years.

ℹ️

Need Help Implementing This?

If you're evaluating AI chip options for your infrastructure, Logicity's research team can help map vendors, benchmark performance claims, and assess supply chain risks. Reach out at research@logicity.in.

Source: Tech-Economic Times / ET

Advertisement
M

Manaal Khan

Tech & Innovation Writer

Produced with AI assistance and reviewed by the Logicity editorial team. Learn more in our Editorial Policy.

Related Articles