AI makes 10-15yr business durability unknowable → terminal-value multiples de-rate + long-dated software credit re-prices → credit curves steepen

The chain

1. AI makes terminal value unknowable — vlad-barbalat in 2026-06-23-podcast-invest-like-the-best-vlad-barbalat-investing-120-billion-in-permanent: "should multiples be actually lower across the board?... the future is so unpredictable" — the durability of a business 10-15 years out can no longer be assumed

2. Long-dated credit on mature software re-prices — vlad-barbalat in 2026-06-23-podcast-invest-like-the-best-vlad-barbalat-investing-120-billion-in-permanent: "Would I be worried about holding 30 year credit on Salesforce or Oracle... a much riskier proposition... that should drive steepness in credit curves"

3. → Credit curves steepen as the long end of IG software credit demands more spread for unknowable terminal durability (this leg is Barbalat's explicit conclusion; the magnitude is the gap)

4. → Equity multiple compression on "AI-crossfire" incumbents (CRM, and even HD / DE) as the same terminal-value uncertainty bleeds into equity DCFs (⚠ unverified — the across-the-board equity-de-rate leg is the hypothesis; Barbalat frames it as a question, not a confirmed effect)

Why it matters

This is a rates/credit-regime expression of the AI-disruption thesis, and it's breadth-positive (a thin vertical for the wiki). Instead of betting on which AI winner emerges, it bets that uncertainty itself re-prices the long end: short/avoid 30yr IG paper on mature software, and position for a steeper credit curve. The equity leg (multiple compression on crossfire incumbents) is the higher-beta, less-certain extension.

• Tradeable: short/avoid long-dated (30yr) IG credit on CRM, ORCL; position for credit-curve steepening in long-dated tech IG; watch for multiple compression on CRM / HD / DE.
• The asymmetry: the long end of software IG credit is priced as "money" (near-certain repayment). If AI makes 10-15yr durability genuinely unknowable, that pricing is wrong and the curve has to steepen.

Why it may not work

• Uncertainty resolves: if AI winners become legible and durable, the de-rate reverses — long-dated software credit really is money and there's no steepening.
• CRM / ORCL prove durable: if the named incumbents demonstrate they can ride AI rather than be disrupted, both the credit and equity legs fail.
• Rates regime dominates: curve steepening can be driven by macro (fiscal, inflation) forces that swamp the idiosyncratic software-durability signal, making the trade un-attributable.
• Equity leg is the weakest link: Barbalat poses the across-the-board equity de-rate as a question, not a finding — it may simply not materialize even if the credit leg does.

What to watch

To graduate this from hypothesis to active thesis:

1. Actual steepening in long-dated tech IG credit spreads (30yr vs 5yr) on the named issuers
2. Spread-widening evidence specifically on CRM / ORCL long-dated paper
3. Multiple-compression evidence on the named AI-crossfire incumbents (CRM, HD, DE)
4. Any move toward removal of quarterly reporting (would raise vol and amplify terminal-value uncertainty)

Sources

• 2026-06-23-podcast-invest-like-the-best-vlad-barbalat-investing-120-billion-in-permanent — primary (Barbalat / Invest Like the Best; terminal-value, 30yr-credit, curve-steepening claims)

Related

• p-and-c-insurer-float-income-rate-regime — adjacent (rate-regime / credit-curve expression)
• agentic-ai-seat-erosion-to-saas-rerate — adjacent (the SaaS-disruption mechanism behind the incumbent de-rate)
• seat-based-saas-ai-disruption — adjacent (parent concept for AI-crossfire on software incumbents)
• vlad-barbalat — source entity