🟢 [E3773] Anthropic's legal workflow plugin for contract review, NDA triage, and compliance sent shares of Thomson Reuters and RELX (LexisNexis) down 18% in a single day. AI can now read files, organise folders, and execute multi-step tasks independently, eliminating need for dozens of expensive SaaS subscriptions.

🟢 [E3780] Goldman Sachs estimates IT spending will shift toward AI rather than traditional software and productivity tools. Profit pool expected to shift from traditional SaaS to AI agents, with TAM projections showing AI agents growing from near-zero in 2025 to $50bn+ by 2030 while traditional SaaS declines.

🟢 [E3782] Claude Code enables engineers to run multiple AI instances simultaneously, becoming project managers working with compute to create code at 18x speed. With source code available on GitHub, Claude can replicate any SaaS product rapidly. Currently 4% of all GitHub code is AI-generated, posing severe threat to incumbents.

🟢 [E3781] Public software valuations declined over 30% as recurring revenue is no longer viewed as durable defensive moat. EV/FCF multiples average 28x versus ex-COVID 5-year average of 38x. Market is bifurcating into 'AI Compute Winners' (Chips/Infrastructure) and 'Legacy Software Losers' with vicious repricing.

🟢 [E3788] Average number of SaaS applications per US organisation fell from peak, reflecting white-collar labour culling since 2023. The AI trade expanded to financial services firms providing tax advice as Claude launched financial services/data platform. Wealth management stocks tumbled on Altruist's Hazel AI tax-strategy tool.

🟢 [E3785] Ironshield argues investors fear 'proprietary' software interfaces are becoming obsolete. If users can manage entire CRM or HR processes through Claude or ChatGPT interface, the 'sticky' dashboards of legacy providers lose value. Loans to even quality software companies like Cloudera, Dayforce, and Rocket Software dropped 7 cents on dollar.

🟡 [E3792] While software sector faces existential AI disruption, companies able to adapt will survive. Cybersecurity demand will only grow as agentic threats become difficult to assess. Management teams could cut internal engineering costs via AI and change pricing to include agentic agents instead of human seats.

🟢 [E3772] Software sector faces existential AI disruption as Claude Code and agentic AI enable workflow automation and rapid software replication. Traditional SaaS economics built on per-seat pricing are challenged as AI automates tasks, directly shrinking revenues of companies like Salesforce, ServiceNow, and Workday. Pure seat-based pricing predicted to be obsolete by 2028.

🟢 [E3783] Software accounts for 1 in 5 liability management exchanges (LMEs). The equity cushion has evaporated as loans originally at 40% LTV are now at 80% LTV. In the last four weeks alone, an estimated $10-15B+ in new software loans fell into distressed territory yielding SOFR + 10%.

🟢 [E3789] Key differentiator from 2016 HY energy crisis: current software stress is sector-wide, not concentrated in small subset. The entire software sector's resilience is in question. US more exposed to current revaluation given technology focus. Lenders who extended 9x-10x Debt-to-EBITDA are now effectively equity holders.

🟢 [E3790] European leveraged loans have only 7% software exposure and Pan-European High Yield has ~4% technology (~2% software), creating relative value opportunity for European stressed credit investors. The lower exposure means Europe faces less systemic contagion risk from software credit deterioration.

🟢 [E3791] Ironshield positions as European credit specialists noting this presents 'very interesting opportunity for European credit investors with little exposure to the asset class, from both a long and a short perspective' given Europe's lower software concentration versus US credit markets.

🟢 [E3774] Software is the single biggest concentration risk in speculative-grade credit. Across HY, leveraged loans, and US private credit (~$4T total), Software is $597B (14%) and Technology is $681B (16%), totalling 30% combined. Deutsche Bank warns cracks will appear first in Private Credit and BDCs which have >20% software exposure.

🟢 [E3775] US distressed software loans rose by $15B in one month, from $10B in December 2025 to $25B in January 2026, representing 11% of total US software loans. Software now accounts for 31% of distressed loan names despite being only 13% of the market, signalling intensifying credit deterioration.

🟢 [E3776] Payment-in-Kind (PIK) usage within Software-focused BDCs has reached record 11.3%, 250+ basis points higher than the already elevated BDC index average of 8.7%. Elevated PIK election rates suggest growing borrower difficulty in servicing cash interest, with 6% classified as 'bad PIK' (restructurings).

🟢 [E3777] BDCs have around 20% of portfolios tied to software sector, but Raymond James analyst Robert Dodd notes actual exposure is 'meaningfully higher than it looks' due to misclassification — healthcare software counted as healthcare exposure, etc. Software companies account for 22% of BDC loans as of Q3 2025.

🟢 [E3778] Major private credit lenders Blue Owl, Ares, and Blackstone saw stock prices fall 8-13% as they are primary lenders to at-risk software businesses. ~5,000 PE-owned software firms globally were underwritten at 18x-20x EV/EBITDA but now clearing at 9x-10x, effectively making lenders the equity holders.

🟢 [E3779] New-money lenders are now underwriting software at 4x leverage, making existing 9x debt stacks impossible to refinance when the private credit maturity wall arrives in 2027-2028. Global speculative-grade debt maturities expected to peak at ~$942B in 2028, largely driven by COVID-era vintage loans with 5-7 year durations.

🟢 [E3787] Failure to attract $300B additional capital for corporate bonds could push high-grade spreads wider by 20-30 bps. In high yield and private credit, spreads could widen 200-300 bps as technology and other sectors reprice. Software sector weighted average bid has plunged to lowest level in three years.

🟢 [E3784] BSL market has been 'quoted' lower but lack of volume suggests marks-to-myth rather than clearing prices. Volume acceleration has not occurred, suggesting clearing prices could be lower. Most Direct Lending and BDC loans remain marked near par despite estimated 'real' liquidity gap of 10-15 points.

💬 [E3786] In ironic twist, credit funds are funding AI data centre builds at expense of their own software portfolios. RBC Global Asset Management expects high-grade net issuance to reach unprecedented $1 trillion. Private credit lending of up to $800B required over next 4 years for AI infrastructure buildout.