Knowledge Atlas released its 2025 annual results after market close on March 31, marking its first financial report since going public. The company announced total annual revenue of 724 million Chinese yuan, representing a year-over-year increase of 131.9%, establishing Knowledge Atlas as the large language model company with the largest revenue scale in China. The company's consolidated gross margin for the year was 41%, significantly exceeding industry standards. According to the financial data, the core growth driver for Knowledge Atlas in 2025 was the full-scale expansion of its MaaS API platform, which achieved an Annual Recurring Revenue of 1.7 billion Chinese yuan (approximately 250 million USD), a sixty-fold increase compared to the previous year. Notably, through extreme engineering optimization on the inference side, Knowledge Atlas substantially reduced the unit cost per token, leading to a major improvement in business profitability. The gross margin for the MaaS platform increased nearly fivefold to 18.9%, far surpassing the industry average. During the subsequent annual earnings conference call, Knowledge Atlas CEO Zhang Peng introduced the new concept of "Token Architect," representing the company's first definition of AI productivity value as "Intelligent Call Volume × Intelligent Quality × Economic Conversion Efficiency." Zhang Peng also stated that enhancing the upper limit of intelligence is the sole "first principle" in the era of AGI for large models. Knowledge Atlas will further explore breakthroughs in the upper limit of intelligence in 2026, aiming to lead paradigm innovation and consistently deliver high-quality tokens as a foundational platform.
Comments