虽然此里程碑意味著AI产业开始收回投资资本，但目前的利润空间仍旧十分狭窄，光是折旧费用就耗去了营收的三分之二以上。即便市场忧虑晶片迭代过快会导致设备迅速贬值，但数据显示如辉达 H100 等旧款晶片租用价格依然坚挺，反映出市场对算力的强劲需求。
此外，报告指出开发者正转向使用开源模型和以 DeepSeek 为代表的中国模型，使 Google、OpenAI 和 Anthropic 在 OpenRouter 平台上的流量份额一年内从 72% 骤降至 33%。这项转变迫使顶尖的模型开发商必须提供更多附加服务与生态绑定，以合理化其溢价定位。
According to the latest report from research firm Exponential View, artificial intelligence (AI) revenues have reached a critical tipping point, indicating that the hundreds of billions of dollars spent by tech giants on infrastructure may be economically sustainable. Global AI sales, excluding China, reached $25 billion in the first quarter of 2026, exceeding the estimated $21 billion in depreciation costs for the second consecutive quarter.
While this milestone suggests that AI companies are beginning to cover their capital expenditures, margins remain extremely thin, with depreciation charges still consuming more than two-thirds of revenue. Despite concerns that rapid chip innovation could accelerate hardware obsolescence, data shows older chips like Nvidia's H100 retain high rental value due to strong demand.
Furthermore, the report highlights a shift among developers toward open-weight and Chinese models like DeepSeek, causing the combined traffic share of Google, OpenAI, and Anthropic on OpenRouter to plummet from 72% to 33% in a year. This transition forces leading foundation-model companies to offer value-added services to justify their premium pricing.