xFusion’s $9B valuation, IPO plans, and the promise & risk in China’s AI server boom

The decision by xFusion to engage Citic Securities for IPO “tutoring” — a preparatory period focused on familiarizing its executive team with listing requirements — suggests a cautious and structured approach to the IPO process. The timeline from January through April/May 2026 implies xFusion may target listing within Q2 or early Q3 of 2026, assuming regulatory and market conditions are favorable.

Financially, xFusion’s sales exceeding 40 billion yuan in 2024 places it among China’s top domestic AI server providers. Nonetheless, some external sources estimate revenue somewhat higher (≈ 43.5 billion yuan, ~US$6.1 billion), reflecting possible variations depending on accounting criteria or recent updates. These figures underpin its US$9 billion valuation in 2023 – large in absolute terms for a privately held server infrastructure company, though still substantially lower than global hyperscale players, leaving room for both upside and valuation risk.

Ownership by major state telecom capital groups provides xFusion with institutional support, access to large contracts, and alignment with Chinese national tech policy, which is increasingly important amid U.S. export controls on high-end hardware. Also, its global reach (over 100 countries) enhances diversification but could expose it to cross-border regulatory and supply risks.

The IPO context is favorable. Investor enthusiasm for domestically originated AI/chip stock is high, as seen in the recent meteoric gains of Biren, Moore Threads, and MetaX. However, wild stock price appreciation comes with caution: many of these firms are unprofitable and/or have business models and product readiness yet to be stress-tested under global competition. xFusion’s could be viewed as more mature, given its larger revenue base and infrastructure business focus, but execution risk remains, especially with respect to supply chain security, technology pace, margin pressure, and compliance with evolving regulatory regimes.

Strategically, xFusion’s IPO, if successful, could serve three potential roles: (1) anchoring domestic AI infrastructure capability, (2) offering a relatively safer play in the AI server market versus more speculative chip designers, and (3) positioning for government contracts and imports substitution. Key open questions include: how profitable is xFusion currently (versus just revenue); how much of its technology is domestically sourced versus reliant on foreign suppliers; and which stock exchange it will list on (A-shares vs. HK). Also, its valuation will be tested by comparison with the premium valuations demanded by chip startups, which often lack revenue but prize growth narratives.