Skip to main content
Back to News
Enflame Technology lists on Shanghai STAR Market at a ~RMB 170.9B (~$24B) close, capping China's Four GPU Dragons IPOs.
Funding
2 min read
CN

Enflame Technology lists on Shanghai STAR Market at a ~RMB 170.9B (~$24B) close, capping China's Four GPU Dragons IPOs.

The AMW Read

Completes the Four GPU Dragons STAR cohort with hard IPO pricing and DSA/inference economics; segment-level for CN AI silicon capital and player-map update.
NoveltySignificance
AI Infra · Player MapCapital CyclesSilicon Substrate

Enflame Technology lists on Shanghai STAR Market at a ~RMB 170.9B (~$24B) close, capping China's Four GPU Dragons IPOs.

On September 11, 2026, Enflame Technology (燧原科技) debuted on the A-share STAR Market, opening at RMB 410 and closing at RMB 397 — roughly 2–3x its RMB 142.18 issue price — for a closing market capitalization of about RMB 170.85 billion. Online demand triggered a clawback that lifted the online allotment to 10.33 million shares, with a final hit rate of 0.0246%. Founded in 2018 by AMD veterans Zhao Lidong and Zhang Yalin, Enflame is the last of China's so-called Four GPU Dragons to complete a STAR listing, after Cambricon, Moore Threads, and MetaX. Per the AI Market Watch index, the company has raised about $746M in total funding across its private rounds (index coverage, not a census).

The listing crystallizes a China-specific AI silicon path: Enflame alone among the Four Dragons uses a domain-specific architecture (DSA) trimmed for tensor/matrix compute rather than CUDA-compatible GPGPU, trading ecosystem breadth for efficiency. Prospectus figures show 2023–2025 revenue of RMB 301M, RMB 722M, and RMB 990M (81% CAGR), with H1 2026 already at RMB 1.12B — ahead of full-year 2025 — and more than 80% of AI accelerator revenue from inference. Tencent supplied about 80% of 2025 sales, including WeChat speech-to-text and other internal workloads. Gross margin rose from 22.6% to 31.8% over three years, but R&D still exceeded revenue and the firm remains unprofitable — a PS multiple near 94x prices growth and domestic-substitute optionality more than current earnings.

For builders and investors, the open questions are training-chip delivery and customer concentration. Fourth-generation train-and-infer L600, announced in July 2025, is still not in mass production; schedule slip would delay any move beyond inference. DSA forces customers through compiler and operator migration, so Enflame must turn its Tencent co-adaptation into repeatable multi-customer wins while peers lock model labs into large Ascend, Moore Threads, and Cambricon purchases. The IPO funds Gen-5/Gen-6 R&D and full-stack software — the test is whether public capital converts a single-hyperscaler DSA bet into broader share beyond the cited ~1.7% domestic footprint.

#Enflame #AIChips #STARMarket #ChinaAI #Inference #GPU

#Enflame#STAR Market IPO#AI chips#DSA#Tencent#China GPU

How This Connects

Based on AI Infra · Player Map

  1. 3h agoNvidia works to ease the electrical power bottleneck slowing AI data center expansionNvidia's
  2. 1w agoPositron AI closes $875M Series C at $5B to tape out LPDDR5X inference ASIC AsimovPositron AI
  3. 1w agoEnflame Technology lists on Shanghai STAR Market at a ~RMB 170.9B (~$24B) close, capping China's Four GPU Dragons IPOs. · THIS ARTICLE
  4. 2w agoCoreWeave Commits $6 Billion to Lancaster, Pennsylvania Data CenterCoreWeave
  5. 2w agoNVIDIA to Invest $3.5 Billion in MediaTek in Expanded AI Chip PartnershipNVIDIA invests $3.5B in MediaTek
  6. 2w agoNvidia commits $3.5 billion to MediaTek in a strategic bid to keep AI chip buildouts inside its own ecosystem.Nvidia

Related News

More news from Enflame Technology

Stay updated with the latest news and announcements from Enflame Technology.

View all Enflame Technology news

Discover AI Startups

Explore 5,000+ AI companies with VC-grade analysis, funding data, and investment insights.

Explore Dashboard