Cambricon Q1 Profit Surges 185% to RMB 1.01B
Cambricon's 185% profit boom reveals China AI chip surge despite US export curbs.
30-Second TL;DR
What Changed
Q1 revenue: 28.85億元, +159.56% YoY
Why It Matters
Cambricon's results underscore China's accelerating AI infrastructure buildout, offering alternatives to restricted Western chips. This could lower costs for domestic AI training and inference.
What To Do Next
Benchmark Cambricon MLU290 chips against Nvidia A100 for your next inference workload.
Key Points
- •Q1 revenue: 28.85億元, +159.56% YoY
- •Q1 net profit: 10.13億元, +185.04% YoY
- •Driven by explosive growth in AI semiconductor sales
- •Strongest quarterly performance amid China AI push
Deep Insight
AI-generated analysis for this event — not the original article.
Enhanced Key Takeaways
- •Cambricon's Q1 2026 performance marks a significant pivot from its historical reliance on government subsidies and R&D grants toward sustainable, product-driven revenue growth.
- •The surge in profitability is largely attributed to the mass adoption of the MLU590 series chips among domestic cloud service providers and state-backed AI research centers seeking alternatives to restricted high-end Western GPUs.
- •Despite the profit surge, Cambricon continues to face significant supply chain constraints due to ongoing US export controls, forcing the company to optimize its 7nm and 5nm chip yields through domestic foundry partnerships.
Competitor Analysis
- Cambricon (MLU590)
- Domestic Cloud/Data Center
- Huawei (Ascend 910B)
- Domestic Cloud/Government
- NVIDIA (H20)
- China-specific Export Model
- Cambricon (MLU590)
- Proprietary MLU Core
- Huawei (Ascend 910B)
- Da Vinci Architecture
- NVIDIA (H20)
- Hopper Architecture
- Cambricon (MLU590)
- Cambricon Neuware
- Huawei (Ascend 910B)
- CANN / MindSpore
- NVIDIA (H20)
- CUDA
- Cambricon (MLU590)
- High (Domestic)
- Huawei (Ascend 910B)
- High (Domestic)
- NVIDIA (H20)
- Restricted/Limited
| Feature | Cambricon (MLU590) | Huawei (Ascend 910B) | NVIDIA (H20) |
|---|---|---|---|
| Primary Market | Domestic Cloud/Data Center | Domestic Cloud/Government | China-specific Export Model |
| Architecture | Proprietary MLU Core | Da Vinci Architecture | Hopper Architecture |
| Software Stack | Cambricon Neuware | CANN / MindSpore | CUDA |
| Availability | High (Domestic) | High (Domestic) | Restricted/Limited |
Technical Deep Dive
- •The MLU590 series utilizes a multi-die chiplet architecture designed to improve yield rates for advanced nodes under domestic manufacturing constraints.
- •Enhanced support for FP8 and INT8 precision formats to accelerate Transformer-based Large Language Model (LLM) inference workloads.
- •Integration of proprietary 'Cambricon-X' interconnect technology to facilitate high-bandwidth, low-latency communication between clusters of AI accelerators.
- •Optimized memory controller design to support HBM3, reducing bottlenecks during massive parameter model training.
Future ImplicationsAI analysis grounded in cited sources
Timeline
- 2016-03Cambricon Technologies is founded as a spin-off from the Chinese Academy of Sciences.
- 2019-12Company launches the MLU270, its first dedicated cloud AI inference chip.
- 2020-07Cambricon completes its IPO on the Shanghai Stock Exchange STAR Market.
- 2023-04Company shifts strategic focus heavily toward high-performance training chips for LLMs.
- 2025-11Cambricon reports record-breaking adoption of its MLU590 series in domestic data centers.
Event Coverage
Weekly AI Recap
Read this week's curated digest of top AI events →
AI-curated news aggregator. All content rights belong to original publishers.
Original source: 36氪 ↗
This is a summary, not the original. Read the source, or get the weekly briefing.
The weekly digest
One email a week. Unsubscribe anytime.