Marathon's Richards: A 15% Default Rate Is Coming for Leveraged Software

What's New

Bruce Richards, Founder and CEO of Marathon Asset Management, delivered the most pointed credit call at Bloomberg Invest 2026, drawing a detailed parallel between the energy sector default cycle of 2016 to 2018 and what he believes is the coming default wave in direct lending to highly leveraged software companies. Richards argues that AI-driven pricing disruption, combined with eight to ten times leverage ratios and near-zero free cash flow after debt service, makes a 15% default rate in direct lending software portfolios not a possibility but a mathematical certainty in 2027 and 2028 when maturity walls hit.

Why It Matters

The private credit industry has spent weeks arguing that its software exposure is either limited or manageable. Richards provides the forensic counterargument: 23% of direct lending portfolios are in software, versus only 1% of private US companies being software companies, and the leverage ratios on those loans are two to three times higher than publicly syndicated equivalents. The free cash flow problem is structural. Companies carrying eight to ten times leverage have no capital left after debt service to invest in the AI transition required to survive it.

Big Picture Drivers

• Energy parallel: Horizontal drilling destroyed the pricing structure for oil and gas in 2014 to 2015, causing 15% back-to-back default rates in 2016 to 2018 for companies that were overleveraged going in. AI is doing the same thing to software pricing, and the leverage is even higher in direct lending.

• Leverage disparity: Public software companies carry approximately 0.5 turns of leverage. Broadly syndicated loan market software carries approximately 5.2 turns. Direct lending software portfolios carry 8 to 10 turns. That disparity is the entire story.

• Free cash flow trap: A company with 10 times leverage and uncertain revenue has no free cash flow after debt service to reinvest in becoming an AI-native business. It cannot reposition. It can only default.

• Capital vacuum: New capital is not flowing into software direct lending at the leverage levels that currently exist. The only capital available is amend-and-extend from existing lenders trying to avoid crystallizing losses. That is a delay, not a resolution.

• HALO framework: Richards introduced Hard Assets Low Obsolescence as his investment thesis for the current environment. Asset-backed lending against aircraft, engines, turbines, and mission-critical real economy assets with long-term contracted cash flows is the antidote to software overexposure.

By The Numbers

• 3% of the high-yield bond market that is software exposure

• 13% of the broadly syndicated loan market that is software exposure

• 23% of the direct lending market that is software exposure, versus 1% of all private US companies

• 8 to 10x leverage on software companies in direct lending, versus 0.5x for public software companies

• 15% projected default rate for direct lending software portfolios, expected to arrive in 2027 to 2028

Key Trends to Watch

• Maturity walls in 2027 to 2028 will force a reckoning as 2020 to 2022 vintage software loans reach their five and seven-year maturities without refinancing capital available at current leverage levels.

• Oil price impact on growth is Richaeds' most immediate macro concern, with his model suggesting $80 oil cuts US GDP growth by 1%, $100 oil brings it to 1% growth, and $120 sustained triggers recession conditions.

• HALO assets including aircraft, engines, cranes, turbines, and commercial real estate lending are seeing strong demand as institutional investors shift away from software concentration and toward hard collateral with contracted cash flows.

• Public software names may represent value at current levels since unlike private portfolio companies they carry minimal leverage and have free cash flow to invest in AI transition, making them better positioned to survive creative destruction.

Memorable Quotes

• "23% of the direct lending loan marketplace is software. And so it has that exposure regardless of where you want to put the chair on the deck." Richards cutting through the industry's attempts to recategorize or redefine software exposure away. Health care software is still software. Education software is still software. The exposure is 23% and no amount of relabeling changes the math.

• "If you're ten times leveraged, all your cash flow is going to debt service. You have no free cash flow to reposition the company, make the investment to transition to an API-first software business." The most clinical articulation of why leverage, not AI disruption itself, is the actual kill mechanism. The disruption creates the need to pivot. The debt load makes the pivot impossible.

• "Hard assets, low obsolescence. It's hard assets that are hard to replicate. Real economy assets that we can lend against." Richards introducing his HALO framework. The contrast with software, which is infinitely replicable at near-zero marginal cost, is intentional. Physical collateral with mission-critical utility is the opposite of what is under stress.

• "The public companies are going to be the winners in that space, not necessarily the private equity companies." A deliberately counterintuitive conclusion. The firms with the financial flexibility to invest in the AI transition are the ones that were never loaded up with ten turns of leverage in the first place.

The Wrap

Richards is not predicting a systemic crisis. He is predicting a localized, time-delayed, mathematically predictable default cycle in a specific corner of direct lending, driven by three factors that are already locked in: technological pricing disruption, excessive leverage, and a capital vacuum that will prevent refinancing. The HALO framework is his practical response, shifting Marathon toward assets with physical collateral, contracted cash flows, and genuine obsolescence resistance. The firms and allocators who make that same distinction now, rather than after 2027, will have a significant advantage in the cycle ahead.