Newsletter · · Ashutosh Agarwal

Is SaaS Broken? - Week of June 12, 2026: Adobe's Margins Crack as the Token Price War Opens

Enterprise software and SaaS newsletter for the week of June 12, 2026. Adobe beat on revenue and ARR yet fell to a seven-year low on deteriorating margins, while the labs prepped a token price war and Salesforce handed the re-rate camp its proof point.

Is SaaS Broken?

Week of June 12, 2026: Adobe's Margins Crack as the Token Price War Opens

Two weeks ago the question was whether per-seat software was dying. This week it showed up in the financials: Adobe printed a beat-and-raise and fell to a seven-year low, because the part of the P&L that matters got worse, while the labs lit the fuse on a token price war before their IPOs.

TL;DR

  • Adobe beat revenue and ARR and still fell ~7% to a seven-year low: the tape finally cares about margin direction, not headline growth.
  • OpenAI is reportedly prepping drastic token price cuts ahead of expected Anthropic cuts: good for SaaS COGS, bad for anyone betting pricing power lives at the model layer.
  • Salesforce gave the bulls real numbers: $3.4B AI ARR up 200%, FY2030 guide raised to $63B, the cleanest proof an incumbent can re-rate to consumption.

What's new

1. Adobe's print made margin compression visible. On Closing Bell Overtime (CNBC), 6/11, Brian Schwartz of Oppenheimer cut through the beat (revenue $6.62B vs $6.45B expected, ARR $27.1B vs $26.6B): "if you peel away the onions… you're seeing a deterioration in the margins, despite the revenue upside, on both the gross margin and the operating margin. So the profit structure of this business is deteriorating." His macro read is the whole thesis: "dollars are flowing to the equipment manufacturers… it's not happening in the application layer."

2. The labs may start a price war, a SaaS tailwind. On Tech Brew Ride Home, 6/11, the host relayed the WSJ report that OpenAI is "considering drastically lowering its prices… in anticipation of similar cuts the startup expects Anthropic to soon make." Sam Altman conceded costs are "a huge issue" and promised "ways we can help people get more value for less spend." Cheaper tokens relieve Adobe's COGS pressure, but confirm the model layer has no pricing power.

3. How much room to cut? A lot. On The AI Daily Brief, 6/12, NLW cited analyst Max Weinbach: margins on inference-intensive API tokens run ~70%, so OpenAI "could cut prices by like 60% and still be profitable." Median enterprise AI spend is just $11.38 per employee per month, vs ~$7,500 for the top 1%, usage headroom that dwarfs any revenue lost to efficiency.

4. Salesforce handed the re-rate camp its proof point. On The 7investing Podcast, 6/8, the analyst walked Q1 FY2027: AI ARR $3.4B up 200% YoY (Agentforce $1B, Data360 $2.2B), FY2030 target raised to $63B from $60B, and a $27B buyback that cut the share count ~10%. "Salesforce is no longer just a per-seat license… now you pay us for credits for the AI on a per-usage basis."

5. "Analytical SaaS is over." On All-In, 6/8, Nikesh Arora, CEO of Palo Alto Networks, drew the line: "If you're an analytical SaaS company, it's over", an LLM runs against your data without your modules. A host gave the anecdote of the week: a 20-seat product where "nobody was logging in," so "we created like three accounts… got rid of 17," connecting it to Slack and Claude.

The debate

Bear: per-seat SaaS is structurally broken. Per-seat demand is collapsing as agents replace logins; 20-to-3 is the template. On Unchained, 6/12, investor Tommy noted subscriptions are "subsidized by about 40x versus the API," so as enterprises shift to metered usage they "hit a wall really quick on spend", and route to open-source inference "for one to one-100th the cost… 80 to 90% as good."

Bull: incumbents re-rate to consumption and margins expand. On CAIS Building With Alts, 6/9, David Breach of Vista Equity argued "the rule of 40 is going to go to the rule of 60", 5-10 points of growth plus 15-20 points of EBITDA margin, because "software is going to become the worker." His proof: a customer paying $7M of new agentic ARR to retire 300 jobs. On Topline, 6/7, Yext CEO Michael Walrath called software at "1x revenue, 4x EBITDA" mispriced and ripping out 20-year workflows for vibe-coded apps "preposterous." Swing factor: whether consumption revenue scales faster than seats bleed.

Stocks in play

  • ADBE: Bull: freemium AI users doubled to 90M (The Rundown, 6/12); 9% grower adding $2.5B/yr at a 7-yr-low multiple. Bear: gross and operating margins deteriorating; ChatGPT/Gemini image-gen threat; CFO out 6/30, CEO retiring. Catalyst: new-CFO margin-path framing.
  • CRM: Bull: $3.4B AI ARR +200%, FY2030 raised to $63B, ~10% of shares retired, <$200. Bear: Commerce + Marketing Cloud "<2%, could be negative"; acquisitive, not organic (Watson Weekly, 6/5). Catalyst: next Agentforce consumption-ARR print.
  • DDOG: Bull: purest consumption model in scope; last week's hero (+~100% YTD). Bear: no fresh operator data; lapping the AI-native cohort. Catalyst: AI-native cohort update. (No coverage this week.)
  • TEAM: Bull: Rovo rides the agent wave on a sticky cloud base. Bear: still seat-priced into seat erosion. Catalyst: Rovo consumption metrics. (No coverage this week.)
  • HUBS: Bull: Breeze + SMB land-and-expand. Bear: most seat-exposed in the group; SMBs cut seats first. Catalyst: NRR and seat-add trend. (No coverage this week.)
  • ASAN: Bull: AI Studio agents as multiplayer teammates. Bear: own CMO admits org-level 10x is "not yet" and AI is "creating more chaos than less" (The Agile Brand, 6/8); smallest, most seat-dependent. Catalyst: AI Studio monetization.
  • MNDY: Bull: fast-growing Work OS with usage-based add-ons. Bear: SMB seat base in the erosion path. Catalyst: AI attach + NRR. (No coverage this week.)

Read-throughs

  • Seat-heavy SMB SaaS (HUBS, ASAN, MNDY) is most exposed to the 20-to-3 dynamic, they sell logins to the buyers most eager to consolidate. Watch NRR; it breaks before ARR.
  • Model/inference vendors told on themselves: a price war and 40x subsidies mean pricing power lives with the application, not the model. Net inference COGS likely falls: Bloomberg Intelligence's Mandeep Singh put serving costs on year-old models down "85 to 90%" (First Trust ROI, 6/9), a SaaS gross-margin tailwind.
  • Multiple de-rating risk is bifurcated: consumption names (DDOG, CRM's AI line) get the benefit of the doubt; pure seat-counters re-rate down until metered revenue outpaces seat bleed.

What changed vs last week

Last week Datadog was the hero (+~100% YTD) and the macro story was a "token shortage keeping prices up." Both flipped: Datadog went silent, and the new vector is a price war, the opposite of scarcity pricing. We also got hard numbers where we had vibes: Salesforce's $3.4B AI ARR and $63B guide, and a quantified inference subsidy ($200 plans costing labs ~$5,000; ~70% API margins). Adobe moved from a Firefly-partnerships story to a margin-compression one. Still quiet: an explicit AI-vs-legacy gross margin % and hard NRR prints, the two numbers that would settle this debate.