The domestic AI landscape has reached a significant new milestone. Following the release of GLM-5.2, Knowledge Atlas (02513) has achieved another remarkable feat. According to reports, by July 2026, the company's Annual Recurring Revenue (ARR) has reached $1 billion. This makes Knowledge Atlas the first local large language model (LLM) developer in China to reach this commercial benchmark, achieving a market expectation of $1 to $1.5 billion ARR by year-end a full six months ahead of schedule. Its astonishing growth rate has fundamentally reshaped market perceptions about the pace of LLM commercialization.
When compared to the commercialization journey of the global benchmark firm, Anthropic, the growth speed of Knowledge Atlas is particularly groundbreaking. Based on official data from Anthropic and financial reports from Knowledge Atlas, it took Anthropic 15 months to grow its ARR from $100 million to $1 billion, whereas Knowledge Atlas accomplished the same feat in just five months. Insiders reveal that between January and July 2026, the ARR of Knowledge Atlas increased fifteenfold.
From an industry-wide perspective, the current $1 billion ARR scale of Knowledge Atlas is now equivalent to the estimated combined ARR of three other leading domestic LLM firms: DeepSeek, Kimi, and MiniMax. This achievement effectively dispels long-standing market skepticism that "domestic companies cannot replicate Anthropic's commercial success story." Previously, the industry consensus was that North America's mature enterprise subscription ecosystem and abundant capital reserves for computing power were unique advantages for Anthropic, making it difficult for local model developers to replicate its explosive revenue path. The performance of Knowledge Atlas has overturned this assumption.
Core Drivers of Growth
This explosive growth is primarily attributed to the company's early and steadfast commitment to two core capability pathways: Coding and Reasoning. Knowledge Atlas was among the first domestic model developers to heavily invest in the Coding route. The company's founder has previously stated that Coding and Reasoning are core model capabilities that can deeply synergize with AI Agents and continuously unlock commercial value. Guided by this strategy, Knowledge Atlas has maintained an iteration cycle of approximately two months, consistently releasing its flagship GLM series models with code capabilities ranking among the world's best.
On June 17th, the company fully released its new-generation flagship model, GLM-5.2, which achieved performance levels surpassing top overseas models. It ranked first among globally available models in the Code Arena blind test and topped the open-source model leaderboard on the Artificial Analysis comprehensive ranking.
Significance of the Milestone
Breaking the $1 billion ARR barrier marks the official entry of China's homegrown AI large models into a new phase of large-scale, stable profitability. It validates that the domestic market also possesses the fertile ground necessary to nurture high-revenue LLM enterprises, demonstrating that Chinese AI can indeed carve out an independent, high-growth commercialization path. The development logic of the AI industry is currently at an inflection point, with competition gradually shifting from pure technical benchmarking contests to the pursuit of genuine, sustainable business value for enterprises.
It is foreseeable that as the AI industry fully enters a cycle of commercial implementation and value realization, ARR will progressively become one of the key metrics for evaluating the long-term core competitiveness of large model companies.
Comments