Biren Technology reported RMB 1.235544 billion in revenue for the six months ended June 30, 2026—up 1,997.6% from RMB 58.903 million in the first half of 2025. The result shows how much room has opened for Chinese AI-accelerator companies, but it is not yet a Nvidia-sized breakthrough: Biren remained loss-making, and the percentage surge started from a very small base.
The company’s first-half results were announced on August 28, 2026. Gross margin reached 42.7%, while the loss for the period was RMB 377.232 million. In other words, revenue is moving sharply upward, but Biren is still spending heavily to build a competitive accelerator business.
The number behind Biren’s 1,997.6% growth
The percentage is mathematically correct, but the starting point matters. Biren’s first-half 2025 revenue was RMB 58.903 million. Against that base, first-half 2026 revenue of RMB 1.235544 billion represents a 1,997.6% year-over-year increase.
That is a dramatic change in business activity, not proof that Biren has caught up with the largest accelerator suppliers. The more useful reading is that the company has moved from a relatively small revenue base into a period of much stronger commercial demand.
| Metric | Period or scope | Reported value | Reader context |
| Revenue | Six months ended June 30, 2026 | RMB 1.235544 billion | Biren’s reported first-half revenue |
| Revenue comparison | Six months ended June 30, 2025 | RMB 58.903 million | The small base behind the percentage surge |
| Year-over-year revenue growth | First half of 2026 versus first half of 2025 | 1,997.6% | Calculated from the two reported RMB figures |
| Gross margin | First half of 2026 | 42.7% | Gross profit after the cost of sales |
| Loss for the period | First half of 2026 | RMB 377.232 million | Biren remained loss-making |
| Cited market-share estimate | China’s AI-accelerator market in 2025 | Below 3% | A separate market estimate, not a measure of revenue growth |
How export controls created an opening for domestic accelerators
The reported market explanation is straightforward: restrictions on Nvidia and AMD supply to China reduced access to some U.S. AI accelerators, while demand for domestic alternatives increased. Chinese industrial policy also encouraged local hardware development. Those conditions gave companies such as Biren more room to pursue sales of GPUs and AI-computing systems.
That does not assign a specific dollar amount of Biren’s revenue to export controls. The company’s growth also reflects its own deliveries and product execution. The low first-half 2025 base amplified the final percentage, too.
This is the important distinction: export controls can widen the opportunity without guaranteeing that every domestic chip designer will convert that opportunity into durable market share. Customers still need usable hardware, software support, and enough production volume.
Biren is growing fast, but it is still a small player
Biren’s cited share of China’s AI-accelerator market remained below 3% in the 2025 estimate. That figure uses a different market definition from other published estimates of Nvidia’s position in China, so the percentages should not be combined into a single market-share calculation.
The practical takeaway is less dramatic than the headline: Biren is gaining commercial traction, but it is not yet operating at Nvidia’s scale. Nvidia’s CUDA software ecosystem remains a major competitive advantage, while Biren is building its own software stack to reduce that gap.
For AI-infrastructure customers, the question is therefore not simply whether Biren’s revenue is rising. It is whether the company can turn that revenue into a broader platform that developers can use, operators can deploy, and manufacturers can supply consistently.
The roadmap behind the revenue
Biren has completed the high-end GPGPU chips BR106, BR110, and BR166. Its reported development roadmap includes BR20X, BR30X, and BR31X.
The hardware is paired with BIRENSUPA, a software stack covering programming models, compilers, libraries, frameworks, and developer tools. More than 1,000 supported AI models have been reported for the platform. Biren’s reported product work also includes LightSphere, an optical-interconnected GPU SuperPod system; training throughput gains of more than 30% have been reported for mixture-of-experts models.
Those figures describe the ingredients of a broader platform, not an independent performance verdict. A chip company needs more than silicon to challenge an incumbent: it needs compatible tools, dependable deployment, and customers willing to build around the platform.
Manufacturing capacity is the next test
The next test is whether Biren can obtain enough production capacity to turn demand into sustained shipments. The company’s roadmap spans accelerators, interconnects, and software, but each part depends on execution at scale.
That makes manufacturing a strategic bottleneck. Strong revenue in one six-month period can show that customers are buying; it cannot by itself show that supply will keep pace or that growth will continue beyond the period reported.
For now, Biren’s first-half 2026 results tell a more interesting story than a simple “2,000% growth” headline. China’s domestic AI-accelerator market has gained room as access to foreign hardware tightened, and Biren is large enough to benefit from that shift. But the company is still loss-making, its market position remains modest, and its next step depends on converting a favorable opening into a scalable hardware-and-software business.
