Observed Signal · Aug 27, 2026 · Opinion / Pricing Guidance · Source: a16z · Impact: 2/5 · Sentiment: Positive
Don’t Price AI Applications Per Token
The article explores how AI application companies can maintain premium pricing despite falling model costs by delivering layered value: domain expertise, integrations, reliable execution, and hard-to-access context. It recommends pricing at the highest measurable value layer—tokens for raw model access, credits for recognizable tasks, and outcome-based pricing where possible—while preserving flexibility through hybrid models. Case studies illustrate success: Harvey's deployment at Repsol (saving 4-6 hours per lawyer weekly) and Decagon's partnership with Notion (improving ticket resolution by 34%). Clay's pricing memo exemplifies separating Data Credits from Actions and selectively passing through volatile model costs. Founders should benchmark against customer alternatives and measure full delivery costs. The article notes that 94% of enterprise LLM decision-makers expect increased spending, with 37% expecting it to double, indicating confidence in AI investments.
Practical guidance on pricing models for AI applications affects SaaS and AI vendors' margins, procurement conversations, and product economics but is not a platform policy or technical release.
Track OpenAI Signals & Market Shifts in Real-Time
Polaris7 autonomous intelligence agents track regulatory filings, primary sources, executive changes, and deal flow 24/7. Create your Polaris7 workspace to monitor these entities.
Key Takeaways & Evidence Grounding
- Harvey deployed to Repsol's legal team after an 8-week pilot with 50 lawyers across 12 jurisdictions; lawyers save 4-6 hours per week with 96% adoption.
- Notion selected Decagon after evaluating internal build, citing up to 34% improvement in ticket resolution time.
- 94% of enterprise LLM decision-makers expect increased LLM spending, 37% expect it to double in the next year.
- Clay separates Data Credits from Actions, passing through token costs for variable reasoning models; their 2026 pricing memo noted mispricing in 2022 and consequent loss in Pro segment.
- In a survey of 50 technical AI buyers, 27 preferred credits tied to recognizable work while 14 preferred tokens.
Connected Companies & Entities
“When OpenAI launched its API in 2020, charging for the computation a model consumed was a sensible way to meter raw inference.”
“Clay offers a helpful example and has often been ahead of the curve in how they think about pricing. In a 2026 pricing memo, the company wro...”
“This newsletter is provided for informational purposes only... a16z has not independently verified nor makes any representations about the c...”
“Harvey is an AI application for legal work, deployed at Repsol, and its Playbook Review illustrates multi-agent system execution.”
“Microsoft Copilot was one of the alternatives compared against Harvey in Repsol's blind evaluation.”
“Notion considered building internally, then chose external solution Decagon after evaluating quality, integrations, security, and partnershi...”
“Decagon was selected by Notion as its external AI support solution, reporting up to a 34% improvement in ticket resolution time.”
Ontology Mapping & Concepts
Related Market Signals & Shifts
Recent verified developments and strategic activity across this market segment.
The Weird Economics of AI Tokens
The article analyzes how modern AI billing and infrastructure have made “tokens” the primary economic unit of intelligence. It explains that tokens (pieces of text processed by models) are a useful billing abstraction but differ widely in computational cost and economic value: input vs output tokens, short vs reasoning-heavy requests, and token usage vs usefulness. The piece describes data centers as “token factories,” highlights risks from subscription and context-window costs, and argues that model routing, token observability, and measuring cost per useful task (not cost per token) will shape AI economics going forward.
AI Economy Shifts as Token Costs Bite
A developer essay by Hicham Douch (published 2026-05-01) argues the era of 'AI is almost free' is ending as providers move to token-based pricing and advanced capabilities become more expensive. The piece cites Anthropic removing Claude Code from a cheaper tier and GitHub Copilot moving from action‑based to token pricing as examples. It reports companies (including a claim about Uber) burning through AI budgets, and warns product teams to impose token budgets, use cheaper models for high-volume scaffolding, and treat AI calls like metered cloud compute. The author dubs the new phase the “tokenogen era,” where every AI call has explicit cost and product roadmaps must account for token economics.
AI Startups Face Pricing Power Squeeze from Model Suppliers
An analysis by Trending Topics highlights a structural challenge for AI startups: they often act as token resellers with thin margins, akin to middlemen, rather than classic software businesses. Using a fictional sports app example, the piece illustrates how costs for app store fees, token consumption, and free-tier AI features can erode profits. Citing a market study, it notes inference costs average 23% of revenue for scaling AI firms, with gross margins around 52% versus 78-80% for traditional SaaS. The article discusses how providers like OpenAI and Anthropic hold pricing power, and some startups, like Cursor, invest heavily in own infrastructure to reduce dependence, though this is often not feasible for most. Neoclouds are seen as not solving the fundamental dependency issue. However, a counterview suggests that rapidly falling inference costs could improve margins, and AI-native startups have already captured significant market share in some segments. The piece concludes with strategic advice for startups to focus on proprietary data, workflow integration, and cost optimization.
Receive all future signals for OpenAI & competitors in your briefing
Never miss material M&A deals, leadership changes, account wins, or product rollouts. Polaris7 monitors 20,000+ primary sources 24/7, delivering verified telemetry directly into your executive briefing.
