🐯虎嗅•Freshcollected in 32m
Moore Threads Eyes Hong Kong Listing
💡China’s GPU challenger has 56 billion yuan in liquidity—but rising inventory, debt, and software risk remain.
⚡ 30-Second TL;DR
What Changed
First-half revenue reached 1.736 billion yuan, up 147% year over year.
Why It Matters
The story highlights the capital intensity and supply-chain risks facing China’s GPU challengers. For AI infrastructure builders, Moore Threads’ progress will affect the availability of domestic alternatives, but its profitability, software ecosystem, inventory turnover, and large-scale delivery remain unresolved.
What To Do Next
Benchmark your inference and training workloads on Moore Threads hardware with MUSA compatibility before committing to it as a production GPU alternative.
Who should care:Enterprise & Security Teams
Key Points
- •First-half revenue reached 1.736 billion yuan, up 147% year over year.
- •Adjusted for non-recurring gains, the company still posted a 151 million yuan loss.
- •Cash and cash-management products totaled approximately 5.625 billion yuan.
- •Inventory rose from 1.332 billion to 3.55 billion yuan, while operating cash flow was negative 2.169 billion yuan.
- •Moore Threads plans an H-share listing less than a year after its 75.76 billion yuan-equivalent-equivalent? A-share fundraising, ahead of a 39.55% lock-up expiration.
🧠 Deep Insight
AI-generated analysis for this event.
🔑 Enhanced Key Takeaways
- •Moore Threads' H-share application is widely interpreted by analysts as a strategic move to hedge against potential further tightening of US export controls on high-end semiconductor manufacturing equipment and IP access.
- •The company's significant inventory buildup is largely attributed to the aggressive stockpiling of HBM (High Bandwidth Memory) and advanced packaging capacity to ensure production continuity amidst geopolitical volatility.
- •The upcoming lock-up expiration involves a significant portion of shares held by early-stage venture capital investors who are seeking liquidity, putting pressure on the company to demonstrate international market valuation.
- •Moore Threads has been actively expanding its 'MUSA' architecture ecosystem, recently securing partnerships with major Chinese state-owned cloud providers to integrate their GPUs into domestic AI training clusters.
- •The negative operating cash flow is exacerbated by the company's heavy R&D investment in software stack optimization, specifically targeting CUDA compatibility to lower the barrier for developers migrating from NVIDIA platforms.
📊 Competitor Analysis▸ Show
| Feature | Moore Threads (MUSA) | NVIDIA (H20/B20) | Huawei (Ascend) |
|---|---|---|---|
| Market Focus | Domestic AI/Graphics | Global AI/HPC | Domestic AI/Cloud |
| Software Stack | MUSA (CUDA-compatible) | CUDA (Industry Std) | CANN (Proprietary) |
| Supply Chain | Domestic/Foundry-dependent | TSMC-dependent | Domestic/SMIC-dependent |
| Pricing | Competitive/Subsidized | Premium/Restricted | Enterprise/Bundled |
🛠️ Technical Deep Dive
- MUSA Architecture: A unified architecture designed to handle both graphics rendering and general-purpose parallel computing (GPGPU).
- Software Stack: Features the MUSIFY toolchain, which aims to automate the translation of CUDA kernels to MUSA instructions to reduce porting friction.
- Memory Strategy: Utilizes advanced packaging techniques to integrate high-speed memory, though it faces challenges in achieving the same bandwidth-per-watt as NVIDIA's latest HBM3e implementations.
- AI Optimization: Focuses on FP8 and INT8 precision acceleration to optimize performance for large language model (LLM) inference tasks within the constraints of domestic manufacturing nodes.
🔮 Future ImplicationsAI analysis grounded in cited sources
Moore Threads will face increased scrutiny from Hong Kong regulators regarding its reliance on domestic foundry partners.
The exchange requires transparent disclosure of supply chain risks, which are currently heavily concentrated in China's domestic semiconductor ecosystem.
The company will likely pivot toward a 'software-first' revenue model to offset hardware margin compression.
High inventory costs and R&D spending necessitate higher-margin software services and ecosystem licensing to achieve long-term profitability.
⏳ Timeline
2020-10
Moore Threads is founded by former NVIDIA executives in Beijing.
2022-03
Company releases its first-generation MUSA-based GPU, the Sudi (S2000/S3000).
2023-10
Moore Threads is added to the US Entity List, restricting access to certain US technologies.
2025-09
Moore Threads completes a major A-share fundraising round totaling 75.76 billion yuan.
2026-08
Company reports H1 2026 revenue of 1.736 billion yuan and announces H-share listing plans.
📰
Weekly AI Recap
Read this week's curated digest of top AI events →
👉Related Updates
AI-curated news aggregator. All content rights belong to original publishers.
Original source: 虎嗅 ↗