Goldman Sachs is rolling out synthetic total-return swap products tied to hyperscaler credit, giving institutional investors a new toolkit for navigating a tech bond market that’s gotten considerably more interesting, and not in the comfortable way.
What Goldman is actually selling
Total return swaps, or TRS, let investors gain synthetic exposure to an asset’s performance, both price changes and income, without actually owning the underlying security.
Goldman began pitching TRS on software-sector corporate loans back in March, positioning the products as a hedge against AI’s potential to disrupt traditional software business models. The timing wasn’t accidental. Secondary market liquidity for software loans had been thinning, and significant new deals were scarce.
By July, the bank, alongside JPMorgan, had expanded the offering into a more structured product: a basket of 18 US high-yield AI-focused issuers tradable via bonds or TRS. Position sizes range from $50 million to $250 million, with the basket carrying an average yield of 7.45% and a spread of 319 basis points.
JPMorgan’s parallel effort focused on investment-grade hyperscaler bonds from names like Microsoft and Amazon, complemented by AI-related junk bonds.
Why CDS spreads are moving
The widening in hyperscaler CDS spreads tells a story about supply and risk perception. Oracle’s CDS, for instance, widened to approximately 200 basis points in 2026, a meaningful jump that reflects the market’s evolving view of tech credit risk.
The catalyst is straightforward: tech companies have been issuing debt at a staggering clip to fund AI infrastructure. Bond issuance from the sector exceeded $194 billion in the first half of 2026 alone, with full-year projections reaching $250 billion.
Goldman’s synthetic trading playbook
Goldman’s credit desk has been building out its synthetic trading capabilities as a strategic priority. The TRS push fits neatly into a broader effort to position the bank as the go-to venue for investors who want to express credit views without the friction of buying and selling actual bonds.
The flexibility cuts both ways. Investors can go long, betting that tech credit holds up as AI spending eventually generates returns, or short, wagering that balance sheets are being stretched too thin. The $50 million to $250 million position range is calibrated for institutional players, not retail.