On April 20, Adobe’s president Anil Chakravarthy stood on stage at Summit 2026 and said four words that land differently when a $247 stock — down 42% from its 52-week high of $423 — says them: “Tokens don’t equate to value.”
He was introducing Adobe CX Enterprise, a rebrand and rearchitecture of Experience Cloud around AI agents. The platform includes over ten production agents — for site optimization, audience creation, journey orchestration, experimentation — plus a higher tier called “Coworkers”: persistent, self-learning agents with enterprise memory that run continuously and orchestrate other agents toward business goals.
But the structural news wasn’t the agents. It was the pricing. For CX Enterprise, Adobe is moving to outcome-based pricing — fees tied to completed ad campaigns, conversion lift, measurable business results. Not subscriptions. Not tokens. Not credits. Outcomes.
This matters because Adobe isn’t a startup experimenting with a new model. It’s a $200B company with 20,000+ enterprise customers rewriting the commercial contract for its flagship platform. And it’s happening the same week that Figma — the company Adobe tried to buy for $20 billion in 2022 — is enforcing the exact opposite model.
The Fork
Since March 18, Figma enforces AI credit limits across all paid seats. Professional users get 3,000 credits per month. An AI generation costs 28-72+ credits depending on model complexity. When you run out, AI features stop working until credits reset or you pay $0.03 per additional credit.
Figma’s model meters the input — how much compute you consume. Adobe’s model meters the output — what business result the AI delivered. Same industry. Same orbit. Opposite answers to the same question: what is the unit of value in AI-powered software?
The distinction isn’t philosophical. It’s structural. Credit-based pricing preserves the vendor’s margin by capping inference costs per user. Outcome-based pricing inverts the risk: Adobe has to deliver measurable results or eat the cost of failed agent runs. Chakravarthy’s interview with The Information made this explicit — the pricing is tied “in part” to the value produced, like the number of ad campaigns an AI agent completes.
Adobe will still use subscriptions and usage-based pricing for other products. CX Enterprise is the wedge, not a wholesale conversion. But the signal is clear: the company that defined the creative software subscription model is now saying subscriptions aren’t the right frame for AI.
What the Market Sees
Adobe’s stock at $247 sits near the bottom of its 52-week range ($224-$423). Tikr’s analysis notes a 46% decline from peak, driven by the same multiple compression hitting the entire SaaS sector — public app software now trades at 3.3x EV/NTM revenue versus a 7.1x five-year average.
The outcome-pricing move reads as either desperation or prescience. Adobe’s creative tools business (Photoshop, Illustrator, Premiere) faces AI-native competition from Midjourney, Runway, and a growing cohort of generative tools that didn’t exist three years ago. Its marketing cloud faces Salesforce’s Agentforce. Both flanks are under pressure.
Outcome-based pricing is a bet that Adobe can deliver enough measurable value through its agent stack to justify prices that pure-subscription competitors can’t match — and that the 20,000 enterprise customers already on Experience Cloud will pay more for guaranteed outcomes than they currently pay for software access.
The Hypothesis Test
This data complicates the hypothesis in a specific way. If software is losing its value because AI commoditizes the product, then outcome-based pricing shouldn’t work — the outcomes would also be commoditized, and vendors would compete prices to zero.
But Adobe is betting on the opposite: that whoever can reliably orchestrate agents across the full customer lifecycle captures more value than a seat-based tool ever did. Chakravarthy’s framing — that CX Enterprise targets “measurable business outcomes” rather than “isolated AI use cases” — maps to the Sequoia autopilot thesis from earlier this year: the real money is in the $50-200B of outsourced business services, not the $600B software market.
If it works, the hypothesis needs another qualifier: software doesn’t lose its value when it sells outcomes instead of access, because the reference price shifts from software budgets to labor and services budgets. Intercom’s $0.99/resolution was the proof of concept. Adobe’s CX Enterprise is the first attempt at enterprise scale.
If it doesn’t work — if agents can’t reliably deliver measured outcomes, or if the orchestration layer itself gets commoditized by open-source MCP tooling — Adobe absorbs the loss. That’s the structural difference between credit pricing and outcome pricing: credits are a cost floor; outcomes are a performance guarantee.
The Counter-Evidence
Three things to hold against the on-thesis reading.
First, Adobe explicitly said outcome pricing won’t apply to all products. Subscriptions and usage-based pricing continue for the creative tools and other offerings. This is a partial migration, not an existential pivot.
Second, Martech.org reports that enterprise customers are pushing back on autonomous agents. “Unpredictability and governance are emerging as top concerns,” with Adobe implementing two tiers of human oversight (human-in-the-loop for campaign planning, human-on-the-loop for consumer-facing agents). Outcome-based pricing requires autonomous execution. Governance requirements throttle autonomy. There’s tension.
Third, 1,770 customers are “entitled” to use the agent platform through a credit-based model — meaning Adobe launched credits first and is layering outcomes on top. The credit infrastructure is the fallback. This is less of a clean break than Chakravarthy’s rhetoric suggests.
What to Watch
The real test arrives when Adobe reports Q2 FY2026 earnings in June. Two numbers will matter: CX Enterprise adoption (how many of the 20,000 Experience Cloud customers migrate) and whether outcome-based contracts show higher or lower average revenue per customer than the subscriptions they replace. If outcomes expand ARPC, it validates the thesis that AI software can capture more value, not less. If they compress it, the hypothesis holds as stated.
Meanwhile, Figma’s credit model produces its own test. If credit limits drive churn — users hitting walls and switching to free AI tools — then metering the input fails too. The pricing fork isn’t settled. It’s just visible now.
Sources: Adobe press release (Apr 20), PYMNTS/The Information (Apr 21), Martech.org (Apr 21), Figma Help Center (Mar 18)