Losses Exceed 2.8 Billion Yuan: Another GPU Giant Eyes IPO with 12 Billion Valuation

Shanghai Tianshu Zhixin Semiconductor Co., Ltd. (referred to as "Tianshu Zhixin") disclosed post-hearing documents on December 19, indicating it has passed the Hong Kong Stock Exchange's listing hearing. If successful, Tianshu Zhixin will compete with Biren Technology—which cleared its hearing just two days prior—for the title of "China's first listed domestic GPU company" in Hong Kong's capital markets.

Founded in 2015, Tianshu Zhixin launched its general-purpose GPU design in 2018 and completed its Series A funding. By 2021, it delivered China's first mass-produced general-purpose GPU product, outpacing newer competitors now entering the IPO race.

According to its HKEX filings, Tianshu Zhixin's revenue primarily comes from its training-series GPUs, which also support gross margins, while inference-series sales are gradually rising. The company has expanded into AI solutions, a segment more susceptible to external factors.

Like most AI chip firms, Tianshu Zhixin faces losses. However, it has reduced its net loss margin through cost optimization, product mix adjustments, and operational efficiency improvements.

**Business Expansion and Losses** Tianshu Zhixin is China’s first chip designer to mass-produce both training and inference GPUs using 7nm technology. Since launching its first training GPU, Tiankai Gen 1, in 2021, it has accelerated R&D, releasing inference GPUs (Zhikai Gen 1/1X) in 2022 and Tiankai Gen 2/3 in 2023–2024.

Revenue grew from RMB 189 million in 2022 to RMB 540 million in 2024 (CAGR: 68.8%), with H1 2025 revenue reaching RMB 324 million (60% of 2024’s total). Its business now spans AI training, inference, and computing solutions.

Training GPU sales remained stable, with higher-priced iterations like Tiankai Gen 2 (87.9% of training revenue in 2024) driving average selling prices (ASP) up from RMB 24,400 to RMB 38,600 (2022–2024), though H1 2025 ASP dipped to RMB 30,400 due to Gen 1 clearance sales.

Inference GPU shipments surged from 38 units in 2022 to 9,800 in 2024, but ASP fell from RMB 11,400 to RMB 9,200 amid market competition. The next-gen Zhikai series will optimize for large language models (LLMs).

AI solutions, launched in 2023, saw project counts rise from 6 to 26 in 2024, with ASP fluctuating between RMB 2.6–5.4 million due to third-party integrations.

**Losses and R&D Investment** Net losses totaled RMB 5.54 billion (2022), RMB 8.17 billion (2023), and RMB 8.92 billion (2024), with H1 2025 adding RMB 609 million, cumulating over RMB 2.8 billion. High R&D spending—241.1%, 213.1%, and 143.2% of revenue (2022–2024)—contributed to losses.

**Profitability and Valuation** Training GPUs maintained 53%–61% gross margins, while AI solutions improved from 25.9% (2023) to 45.7% (H1 2025). Overall net loss margin dropped from 292.3% (2022) to 165.4% (2024).

Post-D+/D rounds in 2025 raised over RMB 3.45 billion, valuing Tianshu Zhixin at RMB 12 billion pre-money. Investors include CDH Investments, Princeville Capital, and Sequoia China.

(Note: This content is for reference only and does not constitute investment advice.)