Pricing Infrastructure: A Framework

Most infrastructure companies get pricing wrong. Not because they don't understand their costs, but because they don't understand how their customers derive value.

Pricing infrastructure is fundamentally different from pricing SaaS. Your customers are developers. They think in API calls, compute hours, and data volumes. They'll model your costs before you do. They'll find the edge cases in your pricing that destroy your margins.

The Three Pricing Models

Every infrastructure pricing model is a variation of three approaches: usage-based, seat-based, or hybrid. Each has trade-offs that matter more than most founders realize.

Usage-Based Pricing

Twilio charges per SMS. Stripe charges per transaction. AWS charges per compute hour. Usage-based pricing ties revenue directly to consumption.

The advantage: low barrier to entry. Customers pay nothing until they use something. This drives adoption. Developers can experiment without procurement approval.

The risk: revenue volatility. When customers optimize their usage (and they will), your revenue drops. When they churn, it happens gradually as usage declines, making it harder to detect.

Usage-based works when your costs scale linearly with usage and your customers' value scales with usage. Twilio's costs per SMS are relatively fixed. Their customers' value (reaching users) scales directly with messages sent. The alignment is clean.

Seat-Based Pricing

GitHub charges per developer. Datadog charges per host. Seat-based pricing ties revenue to the size of the customer's team or infrastructure.

The advantage: predictable revenue. You know what each customer will pay next month. Expansion is tied to their growth, which is visible and forecastable.

The risk: misalignment with value. A host running a critical production database and a host running a dev environment cost you the same to monitor, but deliver vastly different value to the customer. This creates pressure to discount or tier.

Seat-based works when your costs are relatively fixed per unit and your customers' value correlates with the number of units. Datadog's monitoring costs don't vary much per host. Their customers' value (visibility into infrastructure) scales with hosts monitored.

Hybrid Pricing

Stripe charges 2.9% + $0.30 per transaction. MongoDB charges for compute plus storage plus data transfer. Hybrid pricing combines elements of both models.

The advantage: flexibility. You can capture value at multiple dimensions. You can have a base fee for predictability and usage fees for growth.

The risk: complexity. Customers struggle to predict costs. Sales cycles lengthen as procurement tries to model scenarios. Support burden increases as customers ask "why did my bill spike?"

Hybrid works when your costs have both fixed and variable components, and your customers' value has multiple dimensions. Stripe's costs include fixed processing overhead plus variable interchange fees. Their customers' value includes both transaction count and transaction value.

The Margin Math

Pricing isn't just about revenue. It's about margin. The best infrastructure companies maintain 70-80% gross margins at scale. Getting there requires understanding your cost structure.

Most founders underestimate their costs. They count infrastructure and payment processing, but miss:

A payment API might have 10% direct costs (infrastructure, card network fees) but 25% fully-loaded costs when you include fraud prevention, compliance, and support. Pricing at 15% margin on direct costs means losing money on every transaction.

Case Study: MongoDB's Pricing Evolution

MongoDB's pricing journey illustrates how infrastructure companies adapt their models as they scale.

Early MongoDB: open source, free to use. Revenue came from support contracts and enterprise features. This drove adoption but created a conversion problem. Most users never paid.

MongoDB Atlas (2016): usage-based cloud pricing. Compute hours, storage, data transfer. This aligned revenue with customer growth and removed the conversion barrier. Customers could start small and scale.

The result: Atlas now represents over 60% of MongoDB's revenue. ARR grew from $100M to over $1.5B. The pricing model change unlocked growth that the original model couldn't capture.

When to Change Pricing

Pricing changes are high-risk, high-reward. Get them right and you unlock growth. Get them wrong and you lose customers and trust.

Signs you need to change pricing:

Signs you should not change pricing:

The Value Conversation

The best pricing comes from understanding how customers measure value. This requires actual conversations, not assumptions.

Questions to ask customers:

The answers reveal how customers think about value. A customer who says "you save us 3 engineers" values you differently than one who says "you reduce our latency by 50ms." Price accordingly.

Enterprise vs. Self-Serve

Infrastructure companies often need two pricing models: self-serve for developers and enterprise for large organizations.

Self-serve pricing should be:

Enterprise pricing should be:

The transition from self-serve to enterprise is where most infrastructure companies leave money on the table. A customer paying $5,000/month on self-serve might pay $100,000/year on an enterprise contract with the right packaging.

The Pricing Checklist

Before launching or changing pricing, answer these questions:

If you can't answer these questions, you're not ready to price. Do the work first.

The Bottom Line

Infrastructure pricing is a strategic decision, not a financial one. The right model drives adoption, aligns incentives, and creates durable margins. The wrong model creates friction, misalignment, and margin compression.

Most infrastructure companies underprice. They're afraid of losing customers, so they leave money on the table. But underpricing has costs too: it signals low value, attracts price-sensitive customers, and limits your ability to invest in the product.

If you're building infrastructure and wrestling with pricing, we'd like to hear from you. We've seen dozens of pricing models across infrastructure companies. We know what works.