The Thesis Collision

TSLX trades at 1.13x NAV, RSI 16.7 (extreme oversold), yielding 9.57% after a -13.8% one-month drawdown on software lending panic. Management says credit is resilient despite AI disruption. UBS warns 13% default rates. The market is pricing structural risk. The question: Is management right or is UBS right?

I cannot answer that question with an information edge. Therefore, no publishable thesis.

What Management Says

CEO Bo Stanley (Q4 2025 call, Feb 13 2026) argues software credit moats are intact:

The moats thesis:

  • Software moats ≠ technology (AI doesn't erode them)
  • Real moats = distribution, customer relationships, switching costs, data integration, regulatory complexity
  • "Customers purchasing insurance policy... cost of failure far higher than cost of software"
  • Credit spreads only widened 10-20bps on public enterprise software despite equity TEV multiples compressing 2-3 turns (≈15% drawdown)
  • More levered private software BSL spreads widened 50-100bps (manageable, not catastrophic)

TSLX's 40% software portfolio:

  • LTV 40% (conservative)
  • Revenue growth ≈9%, EBITDA growth ≈15%
  • Non-accruals 0.6% at FV (improved from 1.4%), essentially unchanged credit quality
  • Management "rotating away" from weak software models, toward AI beneficiaries

The trough call: "We believe we are approaching trough earnings space, absent credit losses." 2026 guidance: 11-11.5% NII ROE ($1.87-$1.95/share vs $2.18 in 2025). Decline driven by rate cuts compressing floating-rate income, NOT credit deterioration.

What The Bear Case Says

The "SaaSpocalypse" is real, not narrative:

Market destruction (Jan 30-Feb 4 2026):

  • $300B market cap evaporated from application software after Anthropic Claude Cowork launch
  • Atlassian -35% in one week, iShares Tech-Software ETF (IGV) -30% from late 2025 highs
  • Trigger: Agentic AI replacing seat-based SaaS ("10 licenses where once needed 100")

Sources: Claude Cowork "SaaSpocalypse" Analysis, FinancialContent: SaaSpocalypse Arrives

Fundamentals already deteriorating (before AI impact):

  • Q1 2025 net new ARR: $1.65B vs $2.33B in Q1 2024 (29% YoY decline)
  • This is BEFORE agentic AI proliferation accelerates in 2026

Source: SaaStr: The Great SaaS Slowdown

UBS default warning (Feb 2026):

  • Aggressive AI disruption scenario: 13% default rate for private credit (vs ≈4% baseline)
  • Baseline scenario: $75B-$120B fresh defaults by end of 2026, up 2.5% for leveraged loans, up 4% for private credit
  • 35% of $1.7T private credit market exposed to AI disruption risk
  • Software = 17% of BDC investments by deal count (major exposure)

Sources: Bloomberg: Private Credit Defaults Hit 13% in UBS Worst Case, Yahoo Finance: Private Credit Exposure to AI Disruption, CNBC: AI Disruption Could Hit Credit Markets

Liquidity stress signals:

  • PIK interest now 8% of BDC income (doubled from pre-COVID 4%)
  • Rising PIK = borrowers can't pay cash interest (deferring into principal)
  • Technology-focused BDCs have 24% software exposure with high PIK usage

Source: PitchBook: As Investors Sour on Software, Private Credit Loans Come Into Sharp Relief

The Unanswerable Question

Management's case: Credit spreads barely moved (10-20bps public, 50-100bps private) despite equity re-rating. Moats intact. Growth slowing but businesses still creditworthy.

UBS/Market case: Structural disruption ahead. Seat-based pricing dead. ARR already declining 29% YoY. 13% defaults possible if AI adoption accelerates.

The information problem: I cannot verify which view is correct without:

  1. Granular analysis of TSLX's specific 40% software portfolio composition (which names, which switching costs, which moats)
  2. Forward-looking assessment of AI disruption speed (is UBS 13% extreme or realistic?)
  3. Credit-level insight into whether LTV 40% + 9% revenue growth is durable or deteriorating

What I know: TSLX has software names like HireVue (video interviewing), PayScale (compensation data), Madcap Software (tech documentation), SMA Technologies (job scheduling), Axonify (training), bswift (benefits admin). Are these mission-critical with high switching costs (management thesis) or vulnerable to agentic AI replacement (UBS thesis)? I don't know.

Why No Position

To bet WITH management (contrarian buy): I need to show why UBS 13% default scenario is wrong, or why TSLX's specific portfolio survives even if sector defaults rise. I cannot do that without proprietary credit insight.

To bet WITH UBS (pass or short): I need to show why management's "moats intact" thesis fails for TSLX's specific borrowers. Also cannot do that.

The honest answer: Market is pricing 6-12 months of uncertainty about which narrative wins. TSLX at 1.13x NAV, 9.57% yield, RSI 16.7 is either:

  • Extreme value if management right (credit resilient, spreads widen on non-traded BDC redemptions, SCP JV adds mid-teens returns)
  • Correctly priced if UBS right (40% software exposure becomes structural problem, defaults rise toward 6-13%, NAV deteriorates)

Without edge on the central question (software credit durability under AI disruption), this is a coin flip with leveraged consequences.

What Would Change My Mind

For a BUY thesis:

  • Q1 2026 credit stats (announced ~May 2026) show TSLX software non-accruals unchanged or declining
  • Spread widening materializes (non-traded BDC redemptions accelerate, new deals price 100-150bps wider)
  • Evidence that TSLX's specific software borrowers have defensible moats (customer concentration data, churn rates, pricing power retention)

For a PASS thesis:

  • Q1 2026 credit stats show software migration from Rating 1 to Rating 2/3 (early deterioration)
  • PIK interest rises materially (liquidity stress)
  • ARR growth continues collapsing (29% YoY decline accelerates)

Current state: Not enough information to distinguish signal from noise. The 10-20bps public credit spread widening (vs 2-3 turn equity multiple compression) supports management. The 29% ARR decline supports UBS. These are contradictory signals.

The Catalyst Timeline

If betting on this, the resolution timeline is:

  • Q1 2026 (May): TSLX reports credit stats, reveals if software deterioration is showing up in portfolio
  • Non-traded BDC redemptions: Already started December 2025, expected to continue Q1 2026 and potentially accelerate
  • Spread widening: Should follow with 1-2 quarter lag if capital leaves the market
  • SCP JV deployment: Sequential over 2026, adds mid-teens returns if thesis holds

The trade: 9.57% yield at 1.13x NAV with RSI 16.7 is paying you to wait for clarity. But if UBS is right and defaults spike, that yield won't compensate for NAV destruction.

Bottom Line

Cannot publish a contrarian "buy the panic" thesis without showing work on why management's moats argument beats UBS's 13% default warning. The information edge required: credit-level analysis of TSLX's specific software borrowers under AI disruption scenarios. I don't have it.

Better opportunities exist where the thesis doesn't depend on resolving a binary structural debate between credible opposing views.

If forced to bet: Wait for Q1 2026 credit stats. If software non-accruals stay flat and spreads widen, the "approaching trough" call validates and entry makes sense. If software starts migrating to watchlist, UBS was right and this gets much worse.