Adobe: value trap, bond, or compounder? Running it through the Decay Lens

Two things are true about Adobe at the same time, and the gap between them is the entire investment case. The trailing numbers look like a compounder: FY2025 revenue $23.8B (+10.5% YoY), free cash flow $9.9B (up over 25%), non-GAAP operating margin near a record ~44.5%, Digital Media ARR still +11.5%. Meanwhile the market has already re-rated the stock as if it were structurally impaired — down roughly 50% from its 2025 high to $195/share, a forward P/E in the low teens (the cheapest the stock has been in about a decade), and at least one major sell-side Sell rating. That's not "AI is going to eat Adobe" as a mood, circulating as a soundbite — it's a specific, checkable disagreement: either the market is early, pricing decay that hasn't shown up in the numbers yet, or the numbers are about to catch down to the price.

Four drivers, not one verdict

For a seat-based, AI-exposed software business, FCF growth decomposes into four things happening at once: Seats (structural license demand), Price (monetization per seat), Usage (intensity per seat), Margin (operating leverage). Seats and Usage are the AI drag; Price and Margin are the offset. The bear/bull binary collapses the question of whether drag outruns offset into a single yes/no — the framework's job is to keep the two sides separate and arguable.

The base case, with some teeth

Net FCF growth = Seats + Price + Usage + Margin. Base case — Seats −5%, Price +1%, Usage −2%, Margin +2% — nets to about −4%/yr, walking the $9.9B FY2025 base down to roughly $8.1B by Year 5. That's managed decay, bond-like: shrinking, but predictably. At a 14x terminal multiple that path is worth ~$113B terminal value plus ~$44B of cash collected along the way — about +102% versus today's ~$77.6B market cap. (Bull nets to +2%/yr and ~+193%; bear nets to −11%/yr and ~+24% — the live tool's sliders let you test shapes in between, not just these three points.)

What's already priced in

The market cap implies roughly a 12.8% FCF yield — about 7.8x FCF — already cheaper than the 10–12x exit multiple this framework assumes for a structural-disruption outcome. Run the actual bear case (−11%/yr) with zero multiple recovery and the result lands close to flat from today's price, not materially down. That asymmetry — limited further de-rating if the bear case turns out right, real upside if it doesn't — is closer to the real setup than the headline growth number.

Risks worth tracking

• Legal / training-data exposure — a pending class action (Kleiner v. Adobe) alleges some smaller Adobe AI models trained on pirated material; an adverse outcome could force retraining or a margin-pressuring settlement.
• Compute-cost margin pressure — cost of subscription revenue is growing faster (~13%) than revenue (~11%) as GPU inferencing ramps; this is the swing factor for the Margin lever.
• Prosumer / SMB share loss — Canva and OpenAI's Sora target the low end of Adobe's base; Adobe Express is playing catch-up there.
• Freemium dilution — the recent freemium push grows reported ARR optics without growing paid seats at the same rate, which can mask seat compression rather than offset it.
• Pricing-model transition risk — regulatory scrutiny of subscription billing could open the door to smaller, month-to-month competitors.

Bottom line

Adobe isn't being bought as a 10% FCF-yield compounder right now — it's a ~12.8% yield, already discounting real decay. The thesis reduces to one question: does AI shrink seats and per-seat usage faster than Adobe can raise price and margin on what's left? The enterprise numbers — record $10M+ ARR cohort growth, custom-model adoption, still-expanding margin — currently say no. The price action and the freemium/usage-pricing pivot say the market isn't willing to bet on that holding yet. Watching the Seats/Price/Usage/Margin split each quarter, not the revenue headline, is how you find out who's right before the multiple moves again.

Built for seat-based, AI-exposed software specifically (creative, dev, support, data/research tools) — not a fit for usage-based or transaction-based SaaS. Illustrative scenario framework, not a price target or investment advice.