与汤森路透和诺蒂卡共驾资本市场

Capital MarketsRadar ReportQ4’25 Terms-level BenchmarkingAnalysis & Industry Trends Q4’25 CEO Letter After the Spike, MarketsFind Their Level Last year,Noetica's AI surfaced a market shift that no one else was talking about: counterparties trading structuralprotections for issuer-friendly EBITDA flexibility. Our customers saw that trend early, andBloombergcovered thephenomenon weeks later. That trade-off framework is now the established baseline—validated by full-year 2025 datashowing year-over-year increases across both investor protections and issuer flexibility. But Q4’s data reveals something more important: After Q3's dramatic spike in investor protections, the market didn'tcontinue escalating. It stabilized. Beneath that stabilization, next-generation LME blockers are emerging exactly as wepredicted inAmerican Banker, signaling the market's continued evolution toward more sophisticated protectiondesign. This pattern of bilateral negotiation under novel risk conditions isn't confined to LMEs—we're seeing the samedynamic play out across capital markets, particularly in AI infrastructure financing, where lenders are buildingprotection frameworks from scratch for risks in real time. Q3 Spike, Q4 Stabilization: The Market Finds Its Level Q3 2025 saw dramatic surges in protective terms. Lien subordination protections jumped 23 percentage points to84% of deals. J. Crew language (either form) hit 44%. Anti-PetSmart terms reached 28%. The market appeared to behardening rapidly—and lenders appeared spooked. But Q4 reveals the real story.Lien subordination consent requirements moderated to66%—still well above historicalnorms, but down from Q3's peak. J. Crew language pulled back to35%. Anti-PetSmart terms declined to16%. Themarket found its level. This isn't creditor retreat—it's equilibrium. Q3's spike was overcorrection, an unsurprising response to regulatoryuncertainty and heightened risk awareness. Q4's moderation shows the market calibrating to sustainable norms ratherthan maintaining panic-driven maximums. Critically, the most fundamental protections remained strong. Pro rata sharingactuallyincreased from 81% in Q3 to84%in Q4, and erroneous payment terms appeared in91%of deals in Q4. The core architecture holds. What'sadjusting are the deal-specific calibrations. We’re seeing a functioning market pricing risk appropriately through terms. With even the most common protections still below universality, every deal involves real give-and-take. Credit quality,sponsor relationships, and market conditions all influence which protections make the final document. This negotiatingspace is healthy, and information remains more important than ever. Next-Generation Protections: The Market Learns The most telling signal in our Q4 data isn't in the percentages—it's in the evolution. As we wrote inAmerican Bankerthis year, sophisticated market participants are developing next-generation LME blockers that respond to specificliability management innovations. The Pluralsight blocker, which prevents transfers of material IP toanynon-guarantor subsidiary (not just unrestrictedsubs), appeared in11%of deals in Q4. This protection closes the gap revealed by the 2024 Pluralsight deal, wherematerial assets moved to a non-guarantor restricted subsidiary—an escape route traditional J. Crew language didn'tcover. Envision blockers, Wesco/Incoravote manipulation protections, and LMEomniblockerseach appeared in 1%of deals.These terms are creeping into the market, but the data tells us that they remain rare. Early adopters are testing language that will likely proliferate as the market learns from each LME case study. Lenders are splitting into two camps: those playing whack-a-mole with increasingly specific blockers in reaction toeach new LME tactic, and those trying to shut down the entire game with comprehensiveomniblockers. Bothapproaches are experimental—but their emergence proves the market is actively innovating beyond first generationLME blockers. Data Center Financings: Same Challenge, Different Collateral The explosive demand for compute and storage to support the growth of the AI economy shows credit marketsgrappling with the same fundamental challenge: how do counterparties structure protections for risks that are evolvingfaster than standardized frameworks can accommodate? In Q3,CoreWeave closed a $2.6 billion delayed draw term loan facility dedicated to delivering services under a long-term agreement with OpenAI—a structure that concentrates credit risk on a single customer relationship and collateralthat depreciates as GPU technology evolves. This isn't traditional project finance. Lenders can't rely on the same frameworks they use for power plants or toll roads.Instead, they're negotiating bespoke protections like contract realization ratios (measuring actual billings againstprojections) and material contract event triggers—risks that don't fit neatly into standard covenant packages. Just asfirst-gene