宏观危机的三把刀:贸易冲突、AI过热、政治割裂

比推Published on 2025-10-22Last updated on 2025-10-22

作者:arndxt

编译:AididiaoJP,Foresight News

原文标题:宏观困境:贸易战、AI泡沫与政治裂痕


2025 年是一个经济周期的转折点。

市场陷入了一个悖论。

在软着陆乐观情绪的表面平静之下,全球经济正悄然断裂,沿着贸易政策、信贷扩张和技术过度扩张的界线展开。

全球经济的下一次错位不会源于单一的失败,既非源于关税,也非源于人工智能债务,而是源于政策、杠杆和信念之间的反馈循环。

我们正见证一个超级周期的晚期阶段,在这个阶段,技术支撑着增长,财政民粹主义取代了贸易自由主义,货币信任正慢慢侵蚀。

繁荣尚未结束,但已开始断裂。

本周的波动以小见大。

波动率指数经历了自四月份以来的最大飙升,因美中关税担忧重燃,随后在周末前因特朗普总统确认拟议的 100% 进口关税将是「不可持续的」而出现回落。股票市场松了一口气;标普 500 指数企稳。但这种宽慰是表面的,更深层次的叙事是政策工具枯竭和乐观情绪过度拉伸。

稳定的幻觉

7 月美欧贸易协定本意是锚定一个脆弱的体系。

然而,它现在却在气候监管争议和美国保护主义的影响下逐渐瓦解。华盛顿要求美国公司豁免于 ESG 和碳披露规则,突显了日益扩大的意识形态分歧:欧洲的去碳化 vs. 美国的放松管制。

与此同时,中国对稀土出口的新限制,包括禁止含有微量中国来源金属的磁体,暴露了全球供应链的战略脆弱性。美国的回应:威胁对中国进口商品征收 100% 的关税,这是一个具有全球后果的政治姿态。尽管后来这一威胁被收回,但它提醒市场,贸易已变成武器化的金融,与其说是经济理性的杠杆,不如说是国内情绪的杠杆。

世界贸易组织警告称,到 2026 年货物贸易将急剧放缓,这反映了一个现实:公司不再满怀信心地投资于供应链,而是带着应急计划进行投资。

人工智能超级周期

与此同时,在人工智能经济中,第二个叙事正在展开,它更为微妙,但可能更具后果性。

我们正从生产性扩张跨越到投机性金融,其中「供应商融资激增,覆盖率变薄」。超大规模企业现在利用资产负债表进行杠杆扩张的速度,已经超过了收入所能验证的速度,这是周期末期亢奋的典型信号。

这并非新鲜事。自 1790 年以来的 21 次主要投资热潮中,有 18 次以崩溃告终,通常是在融资质量恶化时。今天的人工智能资本支出狂热类似于 1990 年代末的电信泡沫:真实的基础设施收益与信贷推动的投机纠缠在一起。特殊目的实体、供应商融资和结构性债务这些曾一度吹大抵押贷款支持证券的工具正在回归,这次披着「计算能力」和「GPU 流动性」的外衣。

讽刺之处在于?人工智能繁荣是具有生产性的,只是分布不均。微软通过传统债券为扩张融资,显示出信心。CoreWeave 通过特殊目的实体融资,显示出压力。两者都在扩张,但一个建立的是持久能力;另一个建立的是脆弱性。

波动性症状

波动率指数的飙升反映了更深层次的市场不安:政策不确定性、集中的股票领导地位以及在繁荣估值表象下的信贷压力。

当美联储现在发出在增长放缓之际降息的信号时,这不是刺激,而是风险管理。两年期国债收益率跌至 2022 年以来的最低水平,这告诉我们投资者正在为信心的通缩定价,而不仅仅是利率。市场可能仍然为每一次鸽派转向欢呼,但每一次降息都在削弱增长是自我维持的这一幻觉。

综合:贸易、科技与信任

关税政治与人工智能亢奋之间的连接主线是信任,或者更确切地说,是信任的侵蚀。

政府不再信任贸易伙伴。

投资者不再信任政策的一致性。

公司不再信任需求信号,因此他们过度建设。

金价突破 4000 美元,与其说是关于通胀,不如说是关于这种信念的侵蚀:对法币体系、对全球化、对机构协调的信念。它是对冲,但不是对冲价格,而是对冲政策熵。

前路

我们正在进入一个「断裂的繁荣」:一个名义增长和市场高点与结构脆弱性并存的时期:

人工智能投资驱动 GDP 的方式,就像 19 世纪的铁路一样。

贸易保护主义刺激了本地生产,同时消耗了全球流动性。

金融波动在欣快和政策恐慌之间摇摆。

在这个阶段,风险是累积的。

每一次收回的关税、每一次资本支出公告、每一次降息都延长了周期,但压缩了其最终的瓦解。问题不在于人工智能或贸易泡沫是否会破裂,而在于当它们破裂时,两者已经变得多么地相互交织。


Twitter:https://twitter.com/BitpushNewsCN

比推 TG 交流群:https://t.me/BitPushCommunity

比推 TG 订阅: https://t.me/bitpush

说明: 比推所有文章只代表作者观点,不构成投资建议

Trending Cryptos

Related Reads

Media: Anthropic Aims for Revenue Up to $200 Billion by 2028 Ahead of IPO

Anthropic, the AI company behind Claude, is reportedly projecting its revenue could reach between $190 billion and $200 billion by 2028, according to sources cited by Reuters. This forecast, significantly higher than its reported $47 billion annual run-rate in May 2026, is being used by investors and bankers to value the company ahead of a potential IPO. Anthropic has confidentially filed with the SEC, with listing details dependent on market conditions and regulatory approval. The valuation reportedly applies a revenue multiple (EV/Revenue) to these future 2028 figures, an approach common for high-growth tech firms but unusual for a forecast two years ahead of an IPO. This reflects confidence in Anthropic's rapid scaling; its annual run-rate grew tenfold yearly for three years leading into 2026, jumping from ~$9B in late 2025 to over $47B by May 2026. Investors are betting the company's massive spending on GPUs, model training, and infrastructure will eventually outpace costs, leading to expanding margins. Public companies like Palantir, Cloudflare, and SpaceX, valued at high multiples of their forward revenue, serve as benchmarks. While this forward-looking method has precedents, some analysts question the sustainability of such valuations and the actual productivity gains from AI. Despite these concerns, earlier reports suggested Anthropic could be valued at over $1 trillion for its IPO.

cryptonews.ru18m ago

Media: Anthropic Aims for Revenue Up to $200 Billion by 2028 Ahead of IPO

cryptonews.ru18m ago

Open Blockchain Business: Who Gets the Fees and Why a Foundation is Needed

Business of Open Blockchain: Who Gets the Fees and Why a Foundation is Needed In public blockchains, user transaction fees do not simply become "revenue" for a single entity. The flow of money is complex and protocol-dependent. In Ethereum, fees consist of a base fee (burned, permanently removing ETH from circulation) and a priority tip (paid to the validator). In Solana, half the base fee is burned and half goes to the validator, while priority fees go entirely to validators. Thus, "fees" are not synonymous with "project revenue"; they may go to network operators or be destroyed. Foundations like the Ethereum Foundation do not take a cut of user fees. They are funded by their own treasuries, which can generate income (e.g., from staking rewards) and finance ecosystem development, research, and security. Their role is to fund public goods that are hard to monetize directly. Developers of open-source blockchain code can monetize through services, not the code itself. For example, Optimism provides its technology freely but sells enterprise support and managed services (OP Enterprise). Additionally, networks built on Optimism's tech, like Coinbase's Base, share a portion of their revenue with the Optimism ecosystem via agreements, creating an economic link. Token holders do not automatically receive fees as dividends. In Optimism, a portion of protocol revenue is now used to buy back and treasury-lock OP tokens, creating an indirect link between network usage and token economics. In Ethereum, fee burning reduces ETH supply but doesn't distribute funds to all holders; stakers and validators receive direct rewards. For evaluation, analysts should look beyond total fee figures. Key questions are: Who paid and for what? Where did the money go (to validators, operators, treasury, or burning)? What were the recipient's costs? Is there a mechanism linking this income to the token? The same fee amount impacts different blockchain economies in vastly different ways depending on these factors.

cryptonews.ru20m ago

Open Blockchain Business: Who Gets the Fees and Why a Foundation is Needed

cryptonews.ru20m ago

Trading

Spot

Hot Articles

Discussions

Welcome to the HTX Community. Here, you can stay informed about the latest platform developments and gain access to professional market insights. Users' opinions on the price of AI (AI) are presented below.

活动图片