Morgan Stanley has emerged as a leading beneficiary of the AI bond issuance surge, earning $2.3 billion in underwriting fees during the first half of this year, up from $1.4 billion six months prior, according to Yahoo Finance on local time. The bank's success stems from its early positioning in structuring Big Tech creditworthiness and long-term computing contracts into investable bond products, with AI-related bond issuance reaching $236 billion through May—four times the volume from the same period last year. Morgan Stanley now ranks second in related fee revenue behind JPMorgan, surpassing Goldman Sachs, after underwriting $65 billion in AI bonds in the final months of last year alone.
Morgan Stanley's fee income reached $2.3 billion in the first half of this year, compared to $1.4 billion six months earlier, driven by its role in AI-related bond underwriting. The bank advanced to second place in the sector behind JPMorgan and ahead of Goldman Sachs. This growth followed Morgan Stanley's underwriting of $65 billion in AI bonds during the final months of last year, positioning the firm ahead of the market surge. AI bond issuance totaled $236 billion through May, representing a fourfold increase from the same period in the previous year.
TeraWulf, a former Bitcoin mining company now constructing AI datacenters, completed a $3.2 billion bond offering that attracted $10 billion in orders at a 7.75% yield. The speculative-grade issuer secured strong demand due to Google's guarantee of a $3.2 billion lease agreement with FluidStack, TeraWulf's tenant at its New York campus. Under the arrangement disclosed in SEC filings, Google assumes lease payments if FluidStack defaults, in exchange for the right to acquire approximately 14% of TeraWulf's equity. Morgan Stanley served as sole underwriter for the transaction. Cypher Mining structured a similar deal using the same framework.
Meta sold an 80% stake in its Hyperion Datacenter project to private equity firm Blue Owl after initially serving as sole owner of the funding initiative. Morgan Stanley underwrote a $27 billion bond issuance to finance the datacenter construction, marking the largest single structured project finance debt offering in US history. The bonds remain off Meta's balance sheet as an off-balance-sheet obligation. Meta's approach mirrors the TeraWulf model but operates at significantly larger scale.
Market appetite for AI bonds has declined from nearly five times the issuance volume in February to less than two times oversubscription this month. Oracle's credit default swap premiums—insurance costs against bond default—reached their highest levels since 2009 at the end of last year. Despite these cooling signals, AI-related spending shows no signs of immediate slowdown, according to market observers.
Morgan Stanley estimates that AI datacenter construction will require $2.9 trillion by 2028, but Big Tech companies' cash reserves can cover only half of that amount. Bonds previously fueled US railroad expansion and the 1990s telecommunications boom, and now serve as the primary funding mechanism for AI infrastructure. Yahoo Finance noted that the entities pricing these bonds will ultimately determine how quickly the AI future materializes, as every semiconductor and chatbot operates on borrowed capital.
What role did Morgan Stanley play in AI bond issuance through May?
Morgan Stanley earned $2.3 billion in underwriting fees during the first half of this year by structuring AI-related bonds, advancing to second place in the sector behind JPMorgan. The bank underwrote $65 billion in AI bonds in the final months of last year and served as sole underwriter for TeraWulf's $3.2 billion offering and lead underwriter for Meta's $27 billion Hyperion Datacenter bond.
How did Google's involvement affect TeraWulf's bond offering?
Google guaranteed a $3.2 billion lease agreement between TeraWulf and FluidStack, agreeing to assume lease payments if the tenant defaults. This guarantee enabled TeraWulf, a speculative-grade issuer, to attract $10 billion in orders for its $3.2 billion bond at a 7.75% yield. In exchange, Google secured the right to acquire approximately 14% of TeraWulf's equity.
What funding gap exists for AI datacenter construction by 2028?
Morgan Stanley estimates AI datacenter construction will require $2.9 trillion by 2028, but Big Tech companies' existing cash reserves can finance only half of that total, creating a $1.45 trillion funding shortfall that bond markets must address.
Related News
Morgan Stanley Forecasts Memory Chip Shortage Through 2028, Recommends Buying Stocks
JP Morgan Calls Semiconductor Stock Bottom Amid 13% SOXX ETF Decline
Samsung Electro-Mechanics Q2 Operating Profit Forecast Jumps 93% on AI Server Demand
Morgan Stanley Overtakes Goldman Sachs as Top AI Debt Bank in 2026
NVDA, AVGO Top Morgan Stanley Chip Stocks Picks Amid Sell-Off