--- title: "Denglin Technology's Hong Kong GPGPU IPO: A Third-Place, 1%-Share Chipmaker Prices at HK$231-288" date: 2026-11-03 category: Capital & Markets site: NeuroAI canonical: https://neuroai.site/a/na-capital-denglin-gpgpu-ipo language: en --- # Denglin Technology's Hong Kong GPGPU IPO: A Third-Place, 1%-Share Chipmaker Prices at HK$231-288 > Suzhou chip designer Denglin Technology lists on Hong Kong's Chapter 18C (第18C章) un-commercialized (未商业化) route, selling 6.125 million H shares at HK$231-288 for about HK$1.48 billion, with five cornerstone investors (基石投资者) taking 40.66% of the offer but only about 4% of the enlarged share capital. Forty-seven Hong Kong-listed shares already trade above HK$100. Denglin Technology's offer price starts at HK$231. The Suzhou chip designer has 278 employees, and in the market that matters most to it — enterprise AI GPGPU (通用图形处理器) — it ranked third in 2025 with a 1.0% share. That gap between a scarce listing slot and a thin slice of the market is the tension running through this offering. ## Key takeaways - Denglin Technology (登临科技) is selling **6,125,000 H shares** on the Hong Kong Main Board at **HK$231.00-288.00**, about **5%** in the Hong Kong public tranche and **95%** international, plus a **15%** over-allotment option. Net proceeds land near **HK$1.48 billion (≈US$189M / RMB1.25B)** at the HK$259.50 midpoint. - Five cornerstone investors (基石投资者) signed for **HK$646 million (≈US$82M / RMB548M)** — **40.66%** of the offer shares, but only about **4%** of the enlarged share capital once listed. - Revenue went from **RMB 27.9M (2023)** to **RMB 171.3M (2025)**, then hit **RMB 200.4M in the first four months of 2026**, above the whole of 2025. Adjusted profit (non-HKFRS) turned positive at **RMB 20.6M** in that window. - The cash story is weaker: operating cash flow was **negative RMB 146.7M**, receivables jumped from **RMB 121M** to **RMB 267M**, and total debt rose from **RMB 533M** to **RMB 987M** by end-August. - The listing implies **HK$14.05-17.52 billion (≈US$1.79-2.23B / RMB11.9-14.8B)** of market capitalisation — on our arithmetic, roughly **1.5 to 1.9 times** the RMB 7.19 billion post-money valuation from its January 2026 private round. ## A small float under a heavy cornerstone lock The arithmetic of the deal is unusual. Those 6.125 million H shares are about 10.07% of the enlarged share capital, and the Hong Kong public tranche is only 306,250 shares — roughly 0.5% of the company. That is close to the smallest public float a Main Board deal can carry, a direct consequence of listing under Chapter 18C (第18C章) as an un-commercialized (未商业化) specialist technology company, the route for issuers that have not yet cleared the revenue bar but can show a credible path to it. The lock-up does the stabilizing work instead. Five cornerstone investors (基石投资者) will buy HK$646 million of stock: Huayuan Management Consulting (Hong Kong) at RMB 300 million, Jutian Investment (鉅天投资) at RMB 200 million, the Suzhou Consortium (中方财团) at RMB 30 million, ZGD Capital at US$2 million and Everbright Wealth Management (光大理财) at US$1.8 million. Each tranche is locked for six months. The composition matters more than the total. Jutian is the wholly-owned investment vehicle of A-share listed packaging and test house Tongfu Microelectronics (通富微电) — a chip company buying into a chip company with a supply-chain logic. The Suzhou Consortium is a local state platform. ZGD Capital is a Singapore-registered family office. Everbright Wealth Management is a bank wealth-management subsidiary. One investor writes a cheque because it wants the supply chain; another because it has to hold the asset. ## What the prospectus shows The first four months of 2026 are the strongest part of the story. Revenue of RMB 200.4 million was 34.6 times the RMB 5.8 million of the same period in 2025, and above the RMB 171.3 million booked for the whole of 2025. Unit sales went from 4,199 chips in 2023 to 12,144 in 2025 and 10,344 in the first four months of 2026, with the domestically produced KS series contributing RMB 166 million, or 82.9% of revenue. Losses narrowed in parallel: RMB 332.0 million in 2023, RMB 296.5 million in 2024, RMB 281.0 million in 2025, and RMB 16.5 million in the four months to April 2026. Add back share-based payment and listing expenses and the picture flips to an adjusted net gain of RMB 20.6 million, the first positive figure in the track record. The reasons for caution sit in the same document. Gross margin was 50.9% in the four months to April 2026, down 7.8 percentage points from 58.7% a year earlier. Cash used in operations was RMB 146.7 million. Receivables rose from RMB 121 million at the end of 2025 to RMB 267 million at the end of April, and only 20.3% of that April balance had been settled by end-August. Total debt climbed from RMB 533 million to RMB 987 million in eight months. The top five customers still accounted for 90.4% of four-month revenue, the largest at 38.0%. The demand mix explains both the growth and the fragility. About 80% of GPGPU revenue came from smart-city work in 2023; by the first four months of 2026, enterprise private cloud accounted for 83%. That is a deliberate move away from the public-cloud training cluster market, where advanced export-controlled parts dominate, toward fragmented inference deployments priced on efficiency and local service. ## The market math behind the price Frost & Sullivan, the consultant whose numbers the prospectus carries, sizes China's enterprise-AI GPGPU market at RMB 1.9 billion in 2021 and RMB 11.8 billion in 2025, a 58.1% compound annual growth rate, and forecasts RMB 108.5 billion by 2030 on 61.3% annual growth. Denglin's position inside those numbers is modest by construction. In the enterprise-AI segment, Nvidia held about 83.1% of 2025 revenue and AMD about 12.3%; Denglin's enterprise-AI revenue was roughly RMB 116 million, or 1.0%. The company that will be third on a Hong Kong ticker has about one percent of a fast-growing niche, and the two firms ahead of it are not Chinese. That is what the IPO is being asked to price. At HK$231-288 a share, the listing implies HK$14.05-17.52 billion (≈US$1.79-2.23B / RMB11.9-14.8B). Subtracting net proceeds and converting at prevailing rates puts implied post-money valuation at roughly RMB 10.7-13.6 billion, about 1.5 to 1.9 times the RMB 7.19 billion post-money mark from the January 2026 private round. The public market is being asked to pay up for a company with one percent of its market, a 90% customer concentration and negative operating cash flow. ## What Chapter 18C actually buys The route is not a formality. Under the temporary threshold adjustment Hong Kong Exchange (港交所) published in August 2024, running to end-August 2027, an un-commercialized Chapter 18C issuer must expect a market capitalisation of at least HK$8 billion at listing, against HK$4 billion for a commercialized one. It must also show R&D spending of at least 30% of total operating expenditure if most-recent-year revenue reaches HK$150 million, and 50% if it does not. Denglin clears both comfortably: 2025 revenue of RMB 171.3 million sits above the HK$150 million line, and its R&D spending of RMB 238.3 million was roughly 139% of revenue. What the framework gives a company like this is a listing with no profitability requirement attached; what it does not give is a working capital buffer, which is why the HK$1.48 billion matters more than the market-cap headline. ## Honest limitations The offer price is not final until pricing on 6 November, so every ratio here moves with the clearing price. The cornerstone figure needs its denominator attached: the prospectus puts the five investors at 40.66% of the offer shares and 4.09% of enlarged issued capital, while some press coverage quoted roughly 36.6% by measuring the subscription against gross proceeds rather than share count. Frost & Sullivan figures are consultant estimates reproduced in a listing document, not audited data, and the market-share figures for the two companies ranked above Denglin come from prospectus-based analysis rather than separate disclosure. The comparison with the January 2026 private-round valuation is our own arithmetic on disclosed figures, and the RMB 7.19 billion post-money mark comes from company and media disclosure rather than a filed document. The four-month revenue jump also rests on a window in which recognised revenue can be lumpy — precisely why receivables moved the way they did. Nothing here is a recommendation to subscribe. ## What readers can do now - If you are considering the subscription, watch the final price rather than the range. At the HK$231 floor the implied post-money valuation is about 1.5 times the last private mark; at HK$288 it is about 1.9 times, and Hong Kong's tech names have been repriced alongside a global reassessment of AI capital-spending returns. - If you follow Hong Kong listings, treat the cornerstone lock as what it is: about 4% of the company locked, about 0.5% genuinely tradeable. A thin float cuts both ways on volatility. - If you track Chinese AI hardware, the numbers to watch after listing are whether the KS series holds its 83% revenue share, whether gross margin recovers above the 50.9% printed in the first four months, and whether operating cash flow follows revenue up rather than receivables. - If you read Chapter 18C filings, use the checklist the route imposes: HK$8 billion expected market cap for un-commercialized issuers, a 30% or 50% R&D-to-operating-expense ratio, and a credible disclosed path to the revenue threshold. --- Published by NeuroAI (https://neuroai.site/) — https://neuroai.site/a/na-capital-denglin-gpgpu-ipo Free to quote with attribution and a link to the original.