PIMCO AI Capex Concentration
4 steps1 pieces4 Sources
Pacific Investment Management Company (PIMCO) warns that Artificial Intelligence capital expenditures are creating hidden concentration risks across asset classes, challenging the traditional diversification benefits of stocks, bonds, and infrastructure. In its "Charting the Year Ahead: Investment Ideas for 2027" report published on November 17, 2026, PIMCO highlights that nearly a quarter of 2026 USD investment-grade non-financial debt issuance was AI-related. This surge in borrowing by hyperscalers, whose capital spending increasingly outpaces cash flow, means investors holding tech stocks, corporate bonds, and data center funds may be unintentionally over-exposed to the same underlying risks.
The issue has drawn broader industry attention. On November 3, 2026, Blackstone President Jon Gray stated at the Hong Kong Monetary Authority’s Global Financial Leaders’ Investment Summit that the AI buildout requires more equity financing as debt investors face growing concentration risks. PIMCO CEO Emmanuel Roman also warned that the scale of AI financing—estimated at around $10 trillion over five years—could strain markets, noting approximately $550 billion in AI debt issued year-to-date.
Political developments further complicate the outlook. Following the U.S. midterm elections, incoming officials have signaled potential restrictions on data center development. PIMCO’s Lotfi Karoui noted on November 16, 2026, that while these frictions may slow the pace of construction, they are unlikely to reduce the ultimate scale of the AI buildout, potentially prolonging compute scarcity and favoring diversified hyperscalers with strong balance sheets over leveraged infrastructure providers.
To address this, PIMCO recommends analyzing total portfolio risk rather than relying on asset-class labels. Investors should consider diversifying into agency mortgage-backed securities (MBS), emerging market debt, or commodities, which offer different risk profiles. Key metrics to watch include the ratio of AI-related debt issuance to total corporate supply, the spread compression between hyperscaler-tenanted data center bonds and senior unsecured debt, and the implementation timeline of local data center moratoria.