AI Bond Binge Tops $410 Billion, Flips Credit Risk Hierarchy as Regulators Warn of AI Cyberattack Threat

Since SoftBank's roadshow for up to $20 billion in junk bonds was reported on September 8, the AI debt story has picked up two new threads. Bond investors are starting to price some emerging-market companies as safer than American tech giants. And global regulators are warning that AI models are advancing fast enough to increase the speed, scale, and affordability of cyberattacks against the financial system.

Start with the pricing flip. According to Bloomberg, SK Hynix bonds maturing in 2031 now yield just nine basis points more than comparable Amazon.com debt, down to a third of where that gap stood a year ago. Wall Street's biggest borrowers are now getting less trustworthy in the bond market than a South Korean chipmaker. Bloomberg calls it a historic first: risk premiums on emerging-market corporate debt indexes are converging with U.S. benchmarks.

PIMCO Calls It Indigestion

Marc Seidner, PIMCO's Chief Investment Officer for Non-Traditional Strategies, says the volume of debt hitting the market has caused visible strain. According to reporting picked up by BigGo Finance, Seidner says Amazon, Alphabet, and other top-rated tech names have issued hundreds of billions of dollars in bonds this year alone, producing a "crowding-out effect" that forces corporate debt to compete directly with government bonds for the same pool of capital.

Seidner believes that competition is a major reason the 10-year Treasury yield spiked to around 4.75% earlier this month, hitting the top of its multi-year range. Global corporate financing tied to data centers and AI infrastructure has topped $410 billion this year, according to the same reporting.

SoftBank, led by Masayoshi Son, is one of the most aggressive borrowers in that mix. The firm is reportedly seeking a $10 billion loan to refinance debt tied to its OpenAI investment, on top of the $20 billion bond roadshow and 1 trillion yen in retail bonds it's lining up to repay a $40 billion bridge loan. SoftBank plans to inject roughly $65 billion into OpenAI by October, according to people familiar with the matter cited by BigGo Finance.

Foreign Money Is Choosing Corporate AI Debt Over Treasurys

Africa Business Insider, citing Yardeni Research, reports that private foreign investors bought a net $390 billion of U.S. corporate bonds over the past year, more than the $329 billion they bought in Treasury notes and bonds. That Treasury figure is down more than 40% from a year earlier. Alphabet's $25 billion bond offering last month priced at a higher yield than the company had previously paid, a sign investors are demanding more even from top-rated tech borrowers.

Robin Wigglesworth, writing for The New York Times and republished by RealClearMarkets, put it bluntly: AI is "no longer primarily a stock market story. It is a story of debt, and that should worry us." Manias fueled by credit, he notes, tend not to end well.

Regulators Flag a Different Risk: AI Itself

Separately, the Financial Stability Board sent G20 finance ministers and central bank governors a letter dated August 28 warning that AI models are advancing fast enough to increase the speed, scale, and affordability of cyberattacks against the financial system. FSB chair Andrew Bailey, who also runs the Bank of England, wrote that "many jurisdictions do not have the protocols in place to manage the development, release, and deployment of advanced frontier AI models, heightening risks for the financial sector and beyond."

The letter landed ahead of the G20 Finance Ministerial hosted in Asheville, North Carolina. Epoch Times reported that cyberbreaches involving models built by Anthropic and OpenAI occurred over the preceding weeks, though the extent and targets weren't detailed in that reporting. Commerce Secretary Howard Lutnick said in a June 3 statement that a new Commerce Department review process for AI models from firms like Google, Microsoft, and xAI would not slow innovation. Congress still hasn't passed any comprehensive AI regulation.

If Bailey is right that jurisdictions lack basic protocols for frontier AI deployment, that's a hole in financial-system defenses, not a hypothetical one. Critics of heavy-handed AI regulation have a fair counter: over-regulating now could choke off the investment that's driving actual U.S. economic growth, and Washington has a poor record of writing tech rules that don't become box-checking theater. The system currently has neither hard rules nor evidence of an actual attack that's crashed markets. It's a warning about exposure, not a body count.

PIMCO isn't entirely bearish, either. Despite flagging the "indigestion," the firm's strategists reportedly believe today's higher yields are an attractive long-term entry point for fixed-income investors. That's a real counterweight to the doom framing: higher yields cut both ways, rewarding savers and bond buyers even as they raise borrowing costs elsewhere.

None of this changes the math on Washington's own borrowing. The federal government is competing with Big Tech for the same capital pool, at the same time it's doubling bond buybacks to hold yields down, as reported September 8. Higher Treasury yields mean pricier mortgages and car loans regardless of which side of the AI debate wins. The open question is which snaps first: AI capex growth outpacing what bond markets will absorb, or regulators finally writing rules for frontier models before, not after, a cyberattack tests the system Bailey is warning about.