Chinese GPU maker Moore Threads reported a dramatic revenue rebound for 2025, driven by surging demand for high‑performance processors in the AI boom. The company posted operating revenue of RMB 1.506 billion, a 243% year‑on‑year increase, while narrowing its net loss to RMB 1.024 billion from RMB 1.618 billion the previous year.
Management attributed the improved top‑line and gross‑profit performance to accelerated product development and market acceptance after launching the MTTS5000 — a full‑function GPU designed for both training and inference. The firm says the MTTS5000 has reached market‑leading performance levels and entered scaled production, a milestone that the company frames as validation of its architecture and commercial viability.
The results are important in two connected ways: they reflect the immediate pull of AI workloads on GPU demand, and they underscore China’s intensifying effort to build a domestic high‑performance GPU supply chain. Moore Threads is one of a handful of Chinese firms attempting to move beyond inference‑only chips to fully capable training GPUs, an area long dominated by foreign rivals such as Nvidia and, to a lesser extent, AMD.
Yet sizable challenges remain. Despite the revenue jump, Moore Threads is still loss‑making, and translating product wins into sustainable profitability will require steady scale, robust software and ecosystem support, and access to advanced manufacturing inputs. The company will also have to fend off incumbents and prove that its cards can be integrated by cloud providers, data‑centre operators and software frameworks at scale.
For international observers, Moore Threads’ progress is a useful barometer of how quickly China’s domestic GPU sector can close performance and ecosystem gaps. The firm's mass production of a training‑capable GPU suggests accelerating capability, but commercial traction in data centres and among AI developers will be the true test of whether China can reduce reliance on Western suppliers for high‑end compute.