← 返回 Avalaches

儘管樂觀主義者認為科技巨頭擁有充足的現金和強大的信用評級來支持其資料中心投資,但這些公司正採用越來越有創意的方式來籌集資金。例如,Meta 與合作夥伴成立合資企業來開發資料中心,並透過簽署長期租賃協議來發行債券,這種方式成功將巨額債務從 Meta 自身的資產負債表中隱藏起來。

其他科技巨頭也紛紛效仿這種租賃結構,導致隱藏的財務義務大幅增加。根據高盛的分析,目前科技巨頭的總租賃承諾已高達一點五兆美元,其中約一兆美元因為租賃尚未開始,根據美國通用會計準則的規定,並未顯示在公司的財務報表上。(關鍵數字:1.5, 1)

除了龐大的租賃承諾外,這些科技公司還積累了大量購買運算能力、晶片和電力的承諾,總額接近一兆美元。雖然只要人工智慧資料中心的投資最終能帶來回報,這些就不成問題,但這些隱藏的債務和採購承諾最終仍將轉化為必須償還的實質財務負擔。(關鍵數字:1)




Although optimists argue that tech giants have ample cash and strong credit ratings to support their data center investments, these companies are using increasingly creative methods to raise capital. For instance, Meta formed a joint venture with a partner to develop a data center and issued bonds backed by long-term lease agreements, effectively keeping massive debt off its own balance sheet.

Other tech giants are quickly adopting similar lease structures, leading to a massive surge in hidden financial obligations. According to Goldman Sachs, total lease commitments in the sector have reached 1.5 trillion dollars, with about 1 trillion dollars omitted from financial statements because the leases have not yet commenced, taking advantage of US GAAP accounting rules.

In addition to lease commitments, these technology companies have also accumulated nearly 1 trillion dollars in purchase obligations for computing power, chips, and electricity. While this may not be an issue if AI data center investments eventually pay off, these hidden debts and purchase commitments represent substantial financial burdens that must ultimately be paid.
2026-08-11 (Tuesday) · 646b03da00c8d68f3f58eb39a211d58ba4edbd4e

Attachments