Shuntai Technology: Raw Material Prices Show Varying Degrees of Growth, Company Responds with Multi-Angle Comprehensive Measures 尚太科技：原材料价格存在不同程度增长 公司多角度综合应对

The 6.1 million monthly active user decline might signal more pain for Doubao than a typical earnings season. This isn't merely a "trial period ending"—it's a resounding slap aimed at every company that tried to leap straight from treating AI large models as infrastructure to selling them as paid products. Never before has user willingness to pay loomed so large over the industry, a true Sword of Damocles.

Behind these numbers lies a harsh reality: once the novelty fades, most users are still far from reaching the rigid "must pay" threshold for AI dependency. They’re willing to pay for efficiency tools—like the professional version of CapCut or Microsoft's Copilot—because those directly optimize workflows. But a general-purpose chatbot? What would users pay a few yuan a day to discuss? Ask it to draft a weekly report that won’t get approved, or generate AI art that quickly grows monotonous? The vagueness of use cases and the abundance of alternatives (including free or more niche options) form the toughest barrier to paid models. Doubao’s "stress test" might have foreshadowed the preliminary collapse of monetizing general chat-based AI.

Interestingly, just as the industry frets over user conversion, Anthropic is calling to "slow down." This creates a sharp irony: on one side, the urgent anxiety over monetization; on the other, the deep fear of technology spiraling out of control. Anthropic’s warning about "self-improvement" risks feels like the prologue to a sci-fi film, while all current companies are haggling over ticket prices. This disconnect highlights the AI industry’s core contradiction: we worry it will become too powerful to control, yet complain it’s not powerful enough to make users willingly pay. Are we raising a monster, or keeping a pet that isn’t fat enough?

Looking closer to home, Shangtai Technology’s "comprehensive response" to rising raw material costs sounds like a standard answer from any manufacturing firm. But viewed within the AI supply chain, it carries deeper meaning. As a company specializing in anode materials, it is the silent foundation of the "arms race" in AI computing hardware. Any upstream fluctuation ripples through computing power, servers, and eventually hits the cost sheets of every large model. Doubao’s anxiety may not just be about user churn, but also about potentially soaring training and inference costs ahead. As revenue growth stagnates on one side and costs climb on the other, AI service providers are being squeezed from both ends.

Meanwhile, the appointment of E Fund’s chairman as a part-time president of the Asset Management Association of China paints another picture. Having a capital giant’s helmsman join the core of an industry self-regulatory body is often seen as a sign of market maturity and standardization. However, in the AI field, this "maturity" might be ill-timed. The risk-averse nature of financial capital naturally conflicts with the adventure, trial-and-error, and long-termism that the AI industry demands. Can an industry heavily governed by stability-focused capital titans nurture groundbreaking, short-term-return-agnostic breakthroughs? This is a question worth pondering.

Thus, today’s AI news mosaic reveals an industry grappling with "adolescent growing pains." It’s desperate to prove its commercial value (Doubao’s setback) while being plagued by grander ethical risks (Anthropic’s warning); the physical industrial base it relies on is under pressure (Shangtai’s costs), and its capital ecosystem may turn more conservative (the AMAC personnel shift). All contradictions are erupting at once. This is no longer a rosy narrative on a PPT, but a real game of survival, pricing, and boundaries. The departure of those 6.1 million Doubao users isn’t an ending—it’s a stark reminder: the road to AGI must first convince users that it’s worth paying to walk along.

月活610万的下降数字摆出来时，豆包可能正经历比财报季更刺痛的阵痛。这已经不是简单的“试用期结束”，而是一记响亮的耳光，打在所有试图将AI大模型从“基础设施”直接跳到“付费产品”的玩家脸上。用户的付费意愿，从未像今天这样，成为悬在行业头顶的达摩克利斯之剑。

这个数字背后，是一个冰冷的现实：当新鲜感褪去，大多数用户对AI的依赖还远远没有达到“付费”的刚性阈值。他们愿意为效率工具付费，比如剪映的专业版或微软的Copilot，因为那直接对应着工作流的优化。但一个通用的聊天机器人？用户愿意每天花几块钱和它聊什么？是让它写一篇不会被采纳的周报，还是生成几张审美疲劳的AI绘画？需求场景的模糊和替代品的众多（包括免费或更垂直的方案），构成了付费模式最坚硬的壁垒。豆包的这次“压力测试”，或许提前宣告了通用聊天AI付费模型的初步溃败。

有趣的是，就在行业为用户转化率焦头烂额时，Anthropic却在疾呼“放慢脚步”。这形成了绝妙的讽刺：一边是商业变现的迫切焦虑，另一边是对技术失控的深切恐惧。Anthropic警告的“自我改进”风险，像一部科幻片的片头，而眼前所有公司都在为门票价格争论不休。这种割裂感凸显了AI行业的深层矛盾：我们一边担忧它强大到无法控制，一边又抱怨它不够强大到能让用户心甘情愿掏钱。我们是在培育一个怪物，还是在饲养一只不够肥的宠物？

视线拉回国内，尚太科技面对原材料上涨的“综合应对”，听起来像所有制造业公司的标准答案。但如果把它放在AI产业链里看，就意味深长。这家主营负极材料的公司，正是AI算力硬件“军备竞赛”中沉默的基石。上游的任何波动，都会沿着算力、服务器、最终传导到每一个大模型的成本表上。豆包们焦虑的或许不只是用户流失，还有未来可能持续走高的训练和推理成本。当一边是增长乏力的收入，另一边是可能攀升的成本，夹在中间的AI服务商，其利润空间将被双重挤压。

与此同时，易方达董事长出任中基协兼职会长，则是另一幅图景。资本巨头的掌舵者进入行业自律组织的核心，这通常被解读为行业走向成熟与规范的信号。但在AI领域，这种“成熟”可能来得不是时候。金融资本的谨慎天性与AI行业所需要的冒险、试错和长期主义，存在天然张力。一个由求稳的资本大佬深度参与治理的行业，能否孕育出颠覆性的、不计短期回报的突破？这或许是个值得玩味的问题。

所以，今天的AI资讯拼图，展现了一个正在经历“青春期烦恼”的行业。它急于证明自己的商业价值（豆包的挫折），同时被更宏大的伦理风险所困扰（Anthropic的警告）；它依赖的实体工业基础正面临压力（尚太科技的成本），而它的资本世界又可能变得更加保守（中基协的人事变动）。所有矛盾都在此刻集中爆发。这不再是PPT上美好的未来叙事，而是一场关于生存、定价和底线的现实博弈。豆包那610万用户的离去，不是结束，而是一次残酷的提醒：通往AGI的道路，首先得让用户觉得，这条路值得花钱同行。