.png)
Frameworks, core principles and top case studies for SaaS pricing, learnt and refined over 28+ years of SaaS-monetization experience.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
This article expands on a discussion originally shared by robbylit on Reddit — enhanced with additional analysis and frameworks.
SaaS companies face a genuine paradox when implementing AI features: prove your AI is being used (showing margin impact) while simultaneously maintaining the high gross margins investors expect. This tension is creating difficult strategic decisions for product and pricing leaders across the industry.
The AI pricing dilemma is straightforward: LLM costs are real and significant. Notion's CEO has admitted that 10% of their profits go directly to LLMs. Meanwhile, investor Gavin Baker notes it's "definitionally impossible to succeed in AI without gross margin pressure." So how should SaaS companies navigate this landscape?
The AI Pricing Catch-22 for SaaS Companies
Every SaaS company today wants an AI story. Investors demand it, the market expects it, and competitors are racing to deliver it. But implementing AI creates an unavoidable tension:
- To prove AI features are being used (and not just "AI washing"), companies must show some margin impact
- Yet showing margin deterioration risks punishment from Wall Street and investors who expect SaaS companies to maintain 75-85% gross margins
This is creating divergent strategies across the industry. Companies like Figma are bundling AI capabilities into their existing pricing while absorbing the costs. Others like Freshworks have implemented usage-based pricing for AI features, recently increasing their AI agent pricing by 5x after reaching $20M ARR.
Two Strategic Approaches to AI Pricing
Based on analysis of pricing models across the industry, two distinct approaches are emerging:
1. The Bundling Approach: Protect Market Position
For established companies with strong market positions like Figma, bundling AI features into existing pricing structures offers several advantages:
- Defends against new AI-native competitors
- Avoids creating a separate margin line item that might alarm investors
- Positions the core product as having greater value
- Keeps pricing simple for customers
When it works: This approach is most effective when AI features supplement but don't define the core product value. Companies can absorb the costs behind the scenes by slightly raising overall pricing or accepting temporarily lower margins.
The risk: Power users can destroy margins. If a small percentage of customers use AI features extensively, you're effectively subsidizing them with revenue from light users—creating an unsustainable economic model.
2. Usage-Based AI Pricing: Align Value and Cost
For companies where AI is central to the product's value proposition, usage-based pricing aligns economics with reality:
- Creates transparent connection between value delivered and costs incurred
- Allows for margin preservation as usage scales
- Enables premium pricing for power users who extract more value
- Provides clearer unit economics for investors
When it works: This model excels when AI is a primary value driver rather than a nice-to-have feature. It's particularly effective when AI usage directly correlates with customer value received.
The risk: Usage-based pricing can create unpredictability for customers and potentially limit adoption if users fear unexpected costs.
The Decision Framework: What Determines Your AI Pricing Strategy?
The optimal approach depends on several key factors:
1. Core Product Relationship
Enhancement vs. Foundation: Is AI enhancing an existing product, or is it foundational to your value proposition?
- Enhancement: Bundling makes sense (Adobe adding generative features)
- Foundation: Usage-based pricing aligns better (ChatGPT, Claude)
2. Market Position and Competition
Established vs. Emerging: Your competitive landscape influences pricing flexibility.
- Market leaders: Can absorb costs to defend position (Microsoft bundling Copilot)
- Challengers: Need transparent economics to prove viability (newer AI-first tools)
3. User Consumption Patterns
Predictable vs. Variable: How consistent is AI feature usage across your customer base?
- Uniform usage: Flat pricing works (tools with defined AI workflows)
- Wide variation: Usage-based prevents margin destruction (open-ended generative tools)
Implementation Approaches That Work
Companies finding success with AI pricing are taking pragmatic approaches:
Hybrid Models
Many companies are implementing tiered usage models:
- Basic tier: Limited AI capabilities included in base pricing
- Professional/Enterprise tiers: Higher usage limits or more powerful AI features
- Overage fees: For usage beyond tier limits
This creates predictability while protecting margins from extreme outliers.
Transitional Strategies
Legacy SaaS companies with strong cash flows can adopt what Gavin Baker suggests: run AI features at breakeven and use existing profitable products to fund them. This approach:
- Protects overall company margins
- Gives time to refine AI features and understand usage patterns
- Allows for future pricing optimization
Value-Based Metrics
The most sophisticated companies are avoiding direct LLM token pricing, instead charging based on business outcomes:
- AI-generated leads converted
- Time saved through automation
- Revenue influenced by AI insights
These metrics create stronger connection to business value than technical consumption metrics.
Real-World Examples: Contrasting Approaches
The industry is seeing a variety of approaches:
- Figma: Bundling AI design features into existing product tiers
- Freshworks: 5x increase in AI agent pricing after reaching $20M ARR
- Notion: Absorbing 10% profit impact from LLM costs
- HubSpot: Tiered AI capabilities across pricing plans
- OpenAI: Pure usage-based pricing for API consumers
The Future of AI Pricing in SaaS
As the market matures, several trends are becoming apparent:
Initial bundling, later unbundling: Many companies start with bundled AI to drive adoption, then shift to usage-based as features prove their value
Efficiency innovations: The cost curve of AI implementation will improve through:
- Fine-tuned smaller models replacing general-purpose LLMs
- Caching and retrieval augmentation reducing API costs
- Companies building proprietary models for high-volume use cases
- Margin recovery: The initial margin hit will recover as:
- AI capabilities command premium pricing
- Implementation efficiencies reduce costs
- Customer value metrics become more sophisticated
Making Your Decision: Key Questions to Ask
When determining your AI pricing strategy, consider:
- How central is AI to your product's value proposition?
- What is the variability in usage patterns across your customer base?
- How price-sensitive are your customers to AI capabilities?
- What is your competitive position and threat level from AI-native alternatives?
- Can your financial structure support temporary margin pressure?
The AI pricing paradox doesn't have one universal solution. The right approach depends on your specific market position, product strategy, and financial requirements.
The "badge of honor" that a16z's David George mentions—showing margin impact from AI adoption—may be unavoidable for companies seriously implementing AI. But by thoughtfully designing your pricing strategy around your unique circumstances, you can navigate this transition while building a sustainable AI-enhanced business.