与过去外国央行主要将超额储蓄投入美国公债以压低利率不同，当前的资金主力多掌握在避险基金与退休基金等私营机构手中，并高度集中于美股与AI相关资产。由于私营投资者更具收益敏感性且操作迅速，他们甚至将持有的美债作为杠杆担保，使得美国国债殖利率面临上升压力，同时也大幅推高了美欧等国政府偿还庞大债务的财政成本与金融脆弱性。（关键数字：2000）
这场热潮的实体核心——例如拥有全球最密集数据中心园区的维吉尼亚州劳登郡，其财政收入已有超过40%深度依赖数据中心，使其在AI泡沫破裂或遭遇电网负荷挑战时面临巨大的本地经济风险。为了防范潜在的产业崩跌与重演2008年房市危机的惨况，地方政府已著手暂停新建许可，并设立1.2亿美元的收入稳定基金及除役保证金计划，积极推动经济多元化以对冲曝险。
Global capital is flowing heavily into an artificial intelligence boom that mirrors the severe imbalances preceding the 2007-08 subprime crisis. Massive surpluses in export economies like China and Japan coexist with expanding fiscal deficits and ballooning debt in the United States, while windfall profits from Asian chipmakers such as TSMC and Samsung Electronics are channeled into a massive loop funding American tech giants and their $9 trillion AI infrastructure expansion.
Unlike the early 2000s when foreign central banks channeled excess reserves primarily into US Treasuries to depress borrowing costs, current surplus capital is largely controlled by private entities such as hedge funds and pension funds betting heavily on US equities and AI assets. These price-sensitive private investors often leverage Treasury holdings as collateral, driving bond yields higher and escalating debt-servicing burdens for heavily indebted sovereign governments.
At the ground level, regions like Loudoun County, Virginia, host massive data center campuses that generate over 40% of local tax revenues, leaving public finances acutely vulnerable to any potential bursting of the AI bubble or power grid constraints. Mindful of the 2007-08 downturn, local officials have paused approvals for new facilities, established a $120 million revenue stabilization fund, and planned decommissioning bonds while striving to diversify their local economy against sector-specific risks.