Robert Cohen: Fixed Income the Ballast, Not Another AI Bet | Adjusted for Risk

In this episode of Zephyr's Adjusted for Risk podcast, Robert Cohen from Double Line Asset Management discusses the current macroeconomic environment and fixed income strategy. Cohen identifies two primary macro themes: "war and AI." On the war side, Middle East conflict has pushed oil above $100 per barrel, creating inflation concerns that have shifted Fed expectations from rate cuts to potential hikes. Cohen estimates core CPI could exceed 4% and remain above 3% for an extended period. On the AI side, hyperscalers are spending nearly $1 trillion this year with potentially $5 trillion over the coming years, creating concentrated equity market risk comparable to the 1800s railroad era. This massive capital spending is currently stimulative for the economy and equity returns, but builds risks as valuations remain high and credit spreads compress.

Cohort emphasizes that fixed income should serve as a "stay rich" asset class, not a "get rich" asset class, and should steer clear of AI risk exposure. Double Line measures AI exposure across their portfolios, which remains in single-digit percentages across corporate credit, securitized products, and other sectors. While the firm sees growing AI-related financing, they remain cautious about ROI expectations and maintain discipline in credit selection, avoiding deals that appear unsustainable from day one. Cohen notes this is different from the dot-com era only in that credit markets have shown institutional memory and maintained a steep credit curve with appropriate yield premiums for speculative credits (like Coreweave at 8-9% yields) versus hyperscalers (at 5% yields).

On corporate credit specifically, Cohen identifies an upgrade cycle in investment grade debt and maintains that credit fundamentals remain strong. High yield has achieved its highest credit quality in history, now predominantly double-B rated rather than single-B, with limited AI or technology risk. The weakness appears in the triple-C bank loan market and lower-quality transactions from the 2021 M&A boom, where spreads around 20% reflect the hangover from historically low rates. Cohen observes that credit as a percentage of GDP has been declining since the pandemic, suggesting excesses haven't built up despite new issuance. He attributes the lack of expected M&A activity in 2024-2025 to higher rates and geopolitical volatility, plus a shift in advantage from financial sponsors to strategic buyers who use less leverage.

Regarding securitized credit, Cohen argues the asset class matters less than the underlying assets. He identifies data center risk in CMBS and ABS as the same AI risk present in corporate credit, so merely shifting sectors doesn't provide diversification. He favors non-agency mortgages due to tight lending standards post-financial crisis and high equity cushions from elevated home prices, plus CMBS after the post-pandemic office space calamity when investors regained credit discipline. Bank loan CLOs present risks due to underlying credit quality issues, reinforcing the need for active management rather than broad-brush approaches.

On duration, Cohen maintains a lower duration stance, focusing on the 10-year area rather than the long end, citing risks from inflation, fiscal deficits, and potential crowding out. He questions whether long duration assets will hedge equity volatility as they have historically, noting that 2022 showed everything correlating with both rates and equities falling. Clients generating fixed income should achieve returns through shorter duration credit selection rather than extending duration for yield. Cohen recommends a multi-sector approach to fixed income allocation, allowing managers to adjust allocation dynamically as conditions change rather than forcing single-sector bets. He concludes that given current overweight equity positions and high valuations, most investors are probably underweight fixed income and should rebalance toward fixed income as it finally offers meaningful yields.