Shanghai Jiliutech Co., Ltd., an AI infrastructure company founded in February 2023 and led by Tsinghua University PhD graduate Hu Xiaohao, formally submitted its Hong Kong Stock Exchange listing application with Guotai Junan International serving as sole sponsor, according to the company’s prospectus. The company’s post-valuation reached 9.16 billion yuan following its D-round funding in April 2026.
Jiliutech provides AI computing cluster solutions, positioning itself as “China’s largest independent AI computing cluster provider.” The company’s core product, Galaxy, is an AI computing cluster system combining hardware and software platforms. As of December 31, 2025, Jiliutech held approximately 10% market share in China’s 10,000-card cluster service market, with other major players being large internet enterprises and telecommunications operators. By 2025 revenue, the company ranked first among independent AI computing cluster providers in China.
Despite operating for only three years, Jiliutech achieved rapid revenue expansion. From 2023 to 2025, the company’s three-year compound revenue growth rate reached 304.5%, with revenues of 31.8 million yuan (2023), 324 million yuan (2024), and 520 million yuan (2025), according to the prospectus.
However, profitability has been challenged. The company recorded a loss of 2.194 million yuan in 2023, profit of 11.46 million yuan in 2024, and a significant loss of 355 million yuan in 2025. The 2025 loss was primarily attributed to a 367 million yuan increase in fair value losses on financial liabilities with preferential rights, partially offset by 195 million yuan revenue increase from growing AI computing cluster demand and improved product competitiveness.
Gross margins improved across the period: 7.6% (2023), 20.1% (2024), and 21.8% (2025). However, the company’s profitability structure shows a notable weakness: 83.9% of 2025 revenue came from AI computing cluster products with lower gross margins of 7.6%-18.2%, while high-margin AI computing cluster operational services (gross margin 47.7%-56.1%) represented only 16.1% of revenue.
Operating cash flow remained negative: -9.41 million yuan (2023), -21.164 million yuan (2024), and -1.111 million yuan (2025). As of December 31, 2025, the company’s asset-liability ratio reached 136.61%, with current assets of 739 million yuan against current liabilities of 1.188 billion yuan, creating a working capital deficit of 449 million yuan.
Despite profitability pressures, Jiliutech maintained strong capital market momentum. The company completed 11 funding rounds in three years, raising a total of 2.202 billion yuan—averaging one round every four months, according to the prospectus.
Zhipu AI participated in Jiliutech’s angel round in December 2023 and subsequently made multiple follow-on investments through its ecosystem fund Xinglian Capital. As of the prospectus signing date, Xinglian Capital held 7.7% of Jiliutech, making it the largest external shareholder.
The funding trajectory accelerated significantly in 2025-2026. The company completed: A+ round (70.5 million yuan, July 2025), B round (45 million yuan, December 2025), B+ round (146.4 million yuan, February 2026), C round (628 million yuan, April 2026), and D round (1.16 billion yuan, April 2026). Within nine months, funding jumped from the millions to the billions of yuan scale.
Notably, the D-round funding was fully settled on April 21, 2026, just eight days before the company submitted its Hong Kong Stock Exchange prospectus.
According to prospectus disclosures, four early shareholders—Beijing Chenxing, Anjirongchuang, Fangxin Qingcheng, and Huaqing Huiqi—exited with profits by the end of 2025. Founder Hu Xiaohao also transferred shares in two separate transactions, realizing approximately 36 million yuan, according to the filing.
As of April 20, 2026, Hu Xiaohao provided personal guarantees of approximately 240 million yuan for company bank borrowings. The prospectus states that Hu’s personal guarantees are planned to be released before IPO, with most borrowings to be converted to company guarantees post-listing. However, risks remain: if the IPO is delayed or the company’s debt service capacity declines, Hu may face bank claims for repayment.
Jiliutech’s customer base concentrates in large language model companies, cloud service providers, educational and research institutions, and telecommunications operators. Client and supplier concentration remains elevated.
In 2023, the company served four customers. In 2024 and 2025, revenue from the top five customers totaled 321 million yuan and 294 million yuan respectively, representing 98.9% and 56.6% of annual revenue. The largest customer contributed 15.5 million yuan (2023, 48.6% of annual revenue), 191 million yuan (2024, 59% of annual revenue), and 86.3 million yuan (2025, 16.6% of annual revenue).
Procurement from the top five suppliers totaled 47.1 million yuan (2023, 90.7% of annual procurement), 254 million yuan (2024, 75.5%), and 284 million yuan (2025, 54.2%). The largest supplier accounted for 28.3 million yuan (2023, 54.5%), 122 million yuan (2024, 36.2%), and 151 million yuan (2025, 18.8%).
Jiliutech stated in its prospectus that high client and supplier concentration in 2023-2024 is not unusual given AI computing cluster market characteristics. The company cited Frost & Sullivan confirmation that such concentration is typical for companies in early development stages progressively diversifying products, services, and customer/supplier bases. Several top-five customers also served as suppliers and vice versa. According to Frost & Sullivan, overlapping major clients and suppliers is a common phenomenon in the AI computing cluster industry, as some suppliers also require AI computing products and services.
Related News
Matrix Supermind Launches MATRIX-3 Robot, Eyes L2 Autonomy by Year-End
AI Drives Semiconductor 'Super Cycle': Storage Chips Move Center Stage
AI Investment Shifts to 'Soft Infrastructure' and Longevity, VCs Say
SGX FX Adopts Chainlink for On-Chain OTC FX Data Distribution
Chainlink Holds Above Key Support as Institutions Expand Use