Last Updated -

April 28, 2026

Moore Threads

Company Profile and Market Insights

Explore the business model, global strategy, and market performance including insights into its position in China.

Key facts

Founded 2020 • SSE: 688795 • Q1 2026 results (Mar 31, 2026 quarter)

¥737.6m

Revenue (Q1 2026)

¥29.4m

Net profit (Q1 2026)

¥(54.3)m

Adjusted net loss (Q1 2026)

¥(1.49)b

Operating cash flow (Q1 2026)

¥369.0m

R&D investment (Q1 2026)

50.0%

R&D intensity (Q1 2026 revenue)

About

Moore Threads is a Beijing-based Chinese GPU company founded in 2020 by former Nvidia China executive Zhang Jianzhong. The company focuses on full-function GPUs and describes itself as a provider of accelerated computing infrastructure and one-stop solutions for digital transformation across industries. Its stated mission is to become a global GPU leader, and its technology stack is built on the in-house MUSA architecture, which combines AI compute, graphics rendering, physics simulation, scientific computing, and high-definition video processing. Over a short period, Moore Threads has expanded from chip design into a broader platform spanning chips, boards, clusters, and software, and it completed its STAR Market listing in late 2025.

‍

That scale-up is starting to show in the numbers. Moore Threads generated RMB 1.51 billion in revenue in 2025 and still spent RMB 1.31 billion on R&D, which highlights how aggressively it is investing to build a domestic GPU and software ecosystem. In Q1 2026, revenue rose 155.35% year over year to RMB 737.6 million, net profit turned positive at RMB 29.4 million, and R&D spending reached RMB 369.0 million, equal to 50.03% of revenue. Taken together, these results point to a company that is moving from pure technology build-out toward early commercial scale in China’s local AI compute market.

Business Model and Market Position

Moore Threads follows a fabless model. It keeps GPU architecture, chip design, board design, and system validation in house, while outsourcing wafer manufacturing, packaging, testing, and board assembly to specialist partners. The company generates revenue from GPU cards, AI all-in-one systems, and intelligent computing clusters, and it sells through both direct sales and distributors. In 2025, direct sales accounted for 72.86% of revenue, distributors for 27.14%, and 99.87% of revenue came from China.

‍

Cloud AI infrastructure is the core business.
Moore Threads has moved beyond selling standalone chips and now focuses on cloud-side AI compute products such as training and inference cards, integrated servers, and large clusters. In 2025, cloud products generated RMB 1.461 billion of revenue, equal to about 97.0% of total revenue. That makes the company’s commercial model heavily tied to AI infrastructure build-outs, large model training, inference workloads, and related data center demand.
‍
The second pillar is enterprise graphics and metacomputing.
Moore Threads uses the same GPU base to serve cloud desktops, real-time rendering, digital twins, and professional visualization. The MTT S3000 is positioned around graphics rendering, video processing, and deep learning, and the annual report highlights cloud desktop and cloud rendering as established deployment areas. This broadens the addressable market beyond pure AI training and gives the company a more diversified usage profile than an AI-only accelerator vendor.
‍
Edge and terminal products are strategic, but still small.
The company is also building products for edge AI and end devices, including AI notebooks and intelligent modules. Management describes this layer as a heterogeneous stack that combines GPU, CPU, NPU, and VPU resources for local AI workloads. Financially, the segment is still early. Edge and terminal products contributed roughly 1.7% of 2025 revenue, so the investment case still depends far more on cloud infrastructure than on client devices.
‍
Software is a key part of the moat.
Moore Threads has built its own MUSA architecture and software stack, with support that spans AI acceleration, graphics rendering, video processing, and scientific computing. It has also extended that stack from chips to boards to clusters. In practice, this means Moore Threads is selling a domestic compute platform, not only a GPU chip, which matters in China’s market where software compatibility and cluster orchestration are as important as raw silicon performance.

‍

Moore Threads’ market position comes from breadth. In its 2025 annual report, the company says it is one of the few domestic vendors already shipping full-function GPUs at commercial scale. By the end of 2025, it had released five GPU architectures spanning cloud, edge, and terminal use cases. The same report says its MTT S5000 sits in the first tier for domestic AI infrastructure and that Moore Threads has already commercialized thousand-card and ten-thousand-card clusters, with the Huagang architecture designed for clusters above 100,000 cards. That places Moore Threads in a stronger position than many local peers on product range, software depth, and cluster delivery.

‍

The latest numbers support that positioning. In Q1 2026, revenue rose 155.35% year over year to RMB 737.6 million and net profit turned positive at RMB 29.4 million. For now, that looks like early proof that Moore Threads’ cloud-first business model is translating into real commercial scale.

Performance in China

Moore Threads is still almost entirely a China story. In 2025, RMB 1.504 billion of its RMB 1.506 billion total revenue came from the domestic market, equal to about 99.9% of sales, and China revenue rose 244.62% year over year. That momentum continued into Q1 2026, when revenue reached RMB 737.6 million, up 155.35%, and net profit turned positive at RMB 29.4 million.

‍

Key strategic drivers in China include:

‍

AI infrastructure demand is the main growth engine.
Moore Threads is gaining traction in local AI data center build-outs, with its annual report highlighting deployments for internet, research, government, and energy workloads. The company also says its MTT S5000 and KUAE clusters have already been adapted for major Chinese model ecosystems including DeepSeek, Qwen, Wan, and GLM.
‍
Commercial adoption extends beyond model training.
In enterprise and public-sector computing, Moore Threads has pushed its S3000 into the VDI cloud desktop market. The annual report says the company has already landed multi-province commercial projects with China Mobile, while its solution materials position the platform around high-concurrency cloud desktops, rendering, and localized digital office workloads.
‍
Its local position is improving, but competition is intense.
China’s AI accelerator server market is shifting toward domestic suppliers. Reuters reported that Chinese GPU and AI chip vendors captured nearly 41% of that market in 2025, which supports Moore Threads’ addressable opportunity. At the same time, the field is getting more crowded, with Huawei leading among domestic vendors and Nvidia still setting the benchmark on software depth and installed base. Moore Threads’ answer is ecosystem build-out. By the end of 2025, its Moore Academy platform had gathered more than 450,000 developers and learners and reached over 200 universities.

Growth and Future Prospects

Moore Threads entered 2026 from a stronger base than a year earlier. The company completed its STAR Market listing in December 2025 and raised nearly RMB 8 billion, which strengthened its funding base for the next phase of product rollout and ecosystem investment. Q1 2026 then marked an important commercial milestone: revenue rose 155.35% year over year to RMB 737.6 million, net profit reached RMB 29.4 million, and management said the growth was driven by products reaching scaled deployment.

‍

At the same time, Moore Threads is still in a heavy build-out phase. In Q1 2026, R&D spending was RMB 369.0 million, equal to 50.03% of revenue, adjusted net profit excluding non-recurring items stayed negative at RMB 54.3 million, and operating cash flow was negative RMB 1.49 billion as procurement rose with production expansion. That makes the recent profit inflection meaningful, though not yet proof of a fully mature earnings profile.

‍

Key growth drivers include:

‍

A shift from chip vendor to full-stack AI infrastructure provider.
Moore Threads is expanding from standalone accelerators into integrated servers, supernodes, and large computing clusters. Its 2025 annual report says the MTT S5000 targets large model training, inference, and high-performance computing, while its ten-thousand-card KUAE cluster has already been deployed and placed into service. The same report says the company is advancing technologies for clusters above 100,000 cards and has published the MTT C256 supernode architecture plan.
‍
A new product cycle built on the Huagang architecture.
In 2025, Moore Threads released its fifth-generation Huagang architecture. The company says it supports FP4 to FP64 precision, lifts compute density by 50%, and improves computing efficiency by 10 times. Future products based on this platform include the Huashan AI chip for training and inference and the Lushan chip for graphics rendering, which gives Moore Threads a clearer multi-year roadmap across AI and graphics workloads.
‍
A deeper software and developer ecosystem.
Moore Threads is trying to reduce one of the main barriers for domestic GPU vendors, which is software adoption. By the end of 2025, its Moore Academy platform had gathered more than 450,000 developers and learners and reached over 200 universities. The annual report also says the company had established routine Day-0 adaptation support for major Chinese model families including DeepSeek, Zhipu, MiniMax, Moonshot, and Alibaba models.
‍
A favorable domestic substitution trend in China.
Reuters reported that Chinese GPU and AI chip vendors captured nearly 41% of China’s AI accelerator server market in 2025, helped by a new wave of AI infrastructure spending and stronger policy support for domestic supply chains. That trend expands Moore Threads’ addressable market, especially in state-linked and enterprise compute projects where local sourcing matters.

‍

Challenges ahead include:

‍

Profitability is still early.
Moore Threads generated RMB 1.51 billion of revenue in 2025, but still reported a net loss of RMB 1.00 billion and negative operating cash flow of RMB 2.96 billion for the year. The company also states in its annual report that it remains unprofitable and continues to invest heavily in architecture, products, and software.
‍
Customer concentration remains high.
The annual report explicitly warns that customer concentration has stayed elevated over the past three years. That means revenue growth still depends heavily on a relatively small set of accounts and project wins.
‍
Execution pressure is high in a fast-moving market.
Moore Threads says missing the pace of product iteration or shifts in customer demand would weaken competitiveness and slow growth. External competition is also intense. Reuters says Huawei led Chinese vendors in 2025 shipments, while Nvidia still held a 55% share of China’s AI accelerator server market.

‍

Overall, Moore Threads now looks less like an early-stage concept and more like a company entering commercial scale. The next step is clear: turn fast revenue growth, cluster deployment, and ecosystem gains into repeatable profits and stronger cash generation. If it executes on Huagang-based products and keeps winning domestic AI infrastructure projects, its position in China’s local GPU stack should strengthen over the next few years.

This Company Profile was written by Dominik Diemer

Dominik Diemer blends an investor mindset with execution discipline.

He is a SAFe Program Consultant (SPC) and Lean Portfolio Management (LPM) practitioner at DMG MORI Digital, working as a SAFe Release Train Engineer and internal consultant in the Lean-Agile Center of Excellence (LACE).

His focus is prioritization, flow, and dependency management that turns strategy into outcomes. With experience across Bertelsmann and the Founders Foundation, he bridges corporate and startup thinking.

He also invests privately in private equity deals, sharpening his view on business models, value drivers, and go-to-market.

StockCounterParts reflects that lens.

Dominik Diemer

Founder of StockCounterparts.com