Morgan Stanley expects a growing divide in financing terms between high-quality hyperscalers and lower-quality companies as investors remain wary of the debt funding the artificial intelligence build-out. The firm noted that investment-grade hyperscalers maintain strong balance sheets, whereas junk-rated companies tend to be more leveraged. This disparity follows a period of escalating debt fears as artificial intelligence-related supply flooded the bond market, causing yields to jump and credit default swap spreads to widen for many names.
CoreWeave recently became a primary example of these AI debt concerns when its bond yield topped 16% earlier this month. However, the neocloud company reported a third-quarter revenue beat and raised its full-year guidance on Tuesday, leading to a share price jump of more than 25% on Wednesday. Morgan Stanley strategist Fernanda Lima said, "We see an improving risk-reward for HQ hyperscalers and expect more spread volatility for data center bonds and neoclouds given execution and funding risk."
According to Lima, the credit market will remain open for hyperscalers, which have already demonstrated a willingness to pay more to issue bonds. Since September 30, Microsoft, Meta Platforms, and Oracle have issued new bonds in the U.S., while Alphabet and some of Meta's bonds have gone to the European market. Total issuances across both markets have reached $67 billion during that period.
Lima noted that strong balance sheets and earnings cushions allow high-quality hyperscalers to absorb higher borrowing costs while maintaining attractive prospective returns. While most names have seen spreads tighten since recent deals were priced, Oracle was the only company to see spreads widen. Lima added that for these high-quality names, "risk-reward is as good as it's been all year."
Looking ahead, Morgan Stanley anticipates significant increases in capital expenditures from AI companies, forecasting $1.66 trillion next year—a 75% increase from 2026—and $1.77 trillion in 2028. While Lima expects the remainder of this year to be quiet regarding issuance activity, she anticipates 2027 will bring more movement as project timing shifts.
Potential delays in construction may arise due to a Democratic sweep of Congress and voter support for local data center restrictions. Despite these potential shifts in capacity build-out, Lima argued that hyperscalers are better prepared to navigate uncertainty regarding power and new data center construction. She also suggested that companies may use equity issuances to help fund the gap between massive investments and monetization.
