AIs got a gender gap: Women are more skeptical
Summary
Volcano Engine captured roughly half of China’s enterprise model-as-a-service (MaaS) call volume in 2025 and grew its share despite intensified competition, according to IDC data. The company prioritized rapid token consumption and used aggressive pricing enabled by scale and inference engineering (prefill-decode disaggregation, KV cache) to lower per-token costs. Volcano Engine has expanded from model interfaces into agent infrastructure and developer tooling (HiAgent, ArkClaw, OpenViking) to increase stickiness and speed productization. ByteDance’s internal feedback loops and cross-business token usage help iterate models and drive a flywheel of capability, volume, and cost advantage, while the company prioritizes long-term market expansion over short-term margin gains.