对人工智慧的庞大资本投资，从天文数字般的估值与融资目标中可见一斑。各大科技公司正花费数千亿美元建设数据中心基础设施，以满足其 AI 雄心。SpaceX 在其首次公开募股中瞄准了 1.77 兆美元的估值，而 Alphabet 则轻松筹集了 850 亿美元的新股本以支持其 AI 支出习惯。像 OpenAI 和 Anthropic 这样的初创公司也沿著实质上相同的路径发展，后者在迈向自身兆元以上估值的过程中，不断从 Amazon.com、Alphabet 及风险投资家那里安排新的融资计划。

然而，这种前所未有的支出面临著供应限制和市场质疑。尽管 Cerebras 的执行长 Andrew Feldman 指出，由于数据中心限制，目前的投资仍落后于实际的客户需求，但基础设施的建设却受到政治上的反对，以及能源、水、劳动力和晶片采购困难的阻碍。Broadcom 在周三公布了令人失望的 AI 晶片销售预测后，其股价在周四上午下跌了高达 15%，这凸显了与资本快速动员相关的潜在市场波动与风险。

此外，AI 热潮的财务可持续性仍然存在不确定性，因为未履行的合约订单高度集中在未盈利的私有企业手中。Cerebras 报告了 250 亿美元的未履行订单，这主要由 OpenAI 的多年承诺所驱动。Google 拥有 4620 亿美元的订单，其中约 40% 归功于其主要投资对象 Anthropic 的承诺，而 Microsoft 则拥有 6270 亿美元的订单，主要面向其支持的 OpenAI。因此，数兆美元已被承诺用于尚未完全实现的 AI 承诺，迫使投资者密切审查企业招股书，以评估财务回报是否能证明这些庞大支出的合理性。

The massive scale of capital investment in artificial intelligence is highlighted by astronomical valuations and funding targets. Major technology firms are spending hundreds of billions of dollars on data center infrastructure to fuel their AI ambitions. SpaceX is targeting a valuation of $1.77 trillion in its upcoming initial public offering, while Alphabet successfully raised $85 billion in new equity to support its intensive AI expenditure. Startups like OpenAI and Anthropic are following similar trajectories, with the latter frequently securing funding from tech giants like Amazon.com and Alphabet as it moves toward its own trillion-dollar-plus valuation.

However, this unprecedented spending faces supply constraints and market skepticism. While Cerebras Chief Executive Officer Andrew Feldman argues that investment lags behind actual customer demand due to data center constraints, infrastructure construction is hindered by political unpopularity and shortages in energy, water, labor, and chips. Highlighting potential market volatility, Broadcom saw its shares plunge by up to 15% on Thursday morning following a disappointing forecast for its AI chip sales, underscoring the risks associated with the rapid mobilization of capital.

Furthermore, the financial sustainability of the AI boom remains uncertain because backlogs are highly concentrated among unprofitable, privately held entities. Cerebras reports a backlog of $25 billion, primarily driven by multiyear commitments from OpenAI. Google maintains a backlog of $462 billion, of which 40% is tied to Anthropic, while Microsoft holds a backlog of $627 billion, largely backed by OpenAI. Consequently, trillions of dollars are pledged to unfulfilled AI promises, forcing investors to closely scrutinize corporate disclosures to assess if financial returns can justify these massive capital outlays.