Importance of Usage-Based Pricing for AI

Salesforce just fired the starting gun on a seismic shift in how we pay for software. At Salesforce #Agentforce, they announced they’re moving away from the traditional per-seat SaaS model to a consumption-based pricing for their AI agents. This is huge. Why? Because it signals the end of paying just to have access to technology. Instead, we’re moving toward paying for outcomes—the actual value delivered. Think about it. In a world where AI agents can perform the job functions of entire departments, does it make sense to charge per seat? Probably not. Here’s what’s changing: - From access to outcomes: Companies will pay for what the AI actually accomplishes. - From subscriptions to value: Pricing adjusts based on usage and results. - From Software-as-a-Service to Agent-as-a-Service: Technology that collaborates with you as a partner This isn’t just a tweak in pricing—it’s a radical upending of commercial models for large SaaS companies. What does this mean for businesses? - Budgeting will evolve: Costs align directly with value received. - ROI becomes clearer: Easier to measure the direct impact of technology investments. - Greater flexibility: Scale usage up or down based on needs without worrying about seat counts. It’s an exciting time, but also a challenging one. Is every SaaS company ready to embrace a model where companies pay directly for the value they receive? At Uniti AI, we’ve been thinking along these lines. We price our AI agents based on the amount of work they do, not on how many seats a company has. I believe this is the future. What do you think? Is the per-seat model on its way out?

🔵 Stripe just paid $1 billion for something it could have built. That tells you everything about the complexity and urgency of usage-based billing in the AI era. The biggest shift in software monetization since SaaS is happening. Patrick Collison isn't mincing words: 𝐮𝐬𝐚𝐠𝐞-𝐛𝐚𝐬𝐞𝐝 𝐩𝐫𝐢𝐜𝐢𝐧𝐠 𝐢𝐬 "𝐭𝐡𝐞 𝐧𝐚𝐭𝐢𝐯𝐞 𝐛𝐮𝐬𝐢𝐧𝐞𝐬𝐬 𝐦𝐨𝐝𝐞𝐥 𝐟𝐨𝐫 𝐭𝐡𝐞 𝐀𝐈 𝐞𝐫𝐚," potentially as big as (or bigger than) the advent of SaaS itself. UBB - Usage Based Billing Payment processing is one layer; monetization logic is another. Stripe is focused now on both. 🔷 𝐓𝐡𝐞 𝐂𝐨𝐧𝐭𝐞𝐱𝐭 Metronome's valuation doubled in less than a year (from $470M in February to $1B now), with 8x growth in platform volume during 2024. Their client roster speaks volumes: OpenAI, Anthropic, Databricks, Nvidia—companies where consumption-based pricing isn't optional, it's essential. The shift makes sense. AI value correlates directly with consumption: API calls, compute time, tokens processed. Traditional seat-based subscriptions simply don't capture how customers actually derive value. 🔷 𝐁𝐞𝐲𝐨𝐧𝐝 𝐌𝐞𝐭𝐫𝐨𝐧𝐨𝐦𝐞: 𝐓𝐡𝐞 𝐔𝐬𝐚𝐠𝐞-𝐁𝐚𝐬𝐞𝐝 𝐁𝐢𝐥𝐥𝐢𝐧𝐠 𝐄𝐜𝐨𝐬𝐲𝐬𝐭𝐞𝐦 This isn't a one-company phenomenon. This market has exploded with specialized players, each carving out territory. Here are four infrastructure players comparable to Metronome: ‣ Orb – Usage-based billing and pricing infra with strong adoption among modern SaaS and AI companies; Metronome itself positions Orb as its primary direct comparator. ‣ m3ter – Purpose-built usage metering and rating engine for complex B2B SaaS and hybrid models, often grouped with Metronome and Amberflo as the core UBB infra cohort. ‣ Amberflo.ai – Developer-first consumption billing that focuses on metering at scale and “AWS-style” usage pricing; regularly listed alongside Metronome and m3ter as leading UBB startups. ‣ Lago – Open‑source usage-based billing and metering, explicitly branded as a Metronome alternative and highlighted as the strongest choice when teams want control and self-hosting. Stripe chose to acquire rather than build. That signals how complex and critical this capability has become. 🔷 𝐖𝐡𝐲 𝐓𝐡𝐢𝐬 𝐌𝐚𝐭𝐭𝐞𝐫𝐬 Usage-based pricing aligns revenue with value delivery in ways subscriptions never could. It's more transparent for customers, more scalable for providers, and infinitely more adaptable to hybrid models. For financial services and fintech, this is an infrastructure-level transformation. We're not just talking about billing—we're talking about how companies capture value in real-time, optimize pricing dynamically, and build monetization as a competitive advantage. The question isn't whether to consider usage-based models. It's how quickly you can implement them before your competitors do. #Fintech #AI #monetization

AI Agents Don’t Buy Seats—Why Your Pricing Should Follow Suit In the past 12 months, a clear pattern has emerged: as AI systems replace manual effort with automated intelligence, pricing structures tied to “seats” no longer reflect the value customers receive. Pricing models have surfaced as a hot topic with every portfolio company at Mosaic Ventures and is top-of-mind for nearly every founder building applied-AI products. When one person and an AI agent can outperform an entire legacy team, charging per user starts to feel arbitrary; what matters is how much business impact the product delivers. Founders are experimenting with three broad approaches: 1. Usage-metered plans that bill against tokens, API calls, or minutes of inference time. These create a direct bridge between consumption and margin and nudge teams to track cost from day one. 2. Outcome-based pricing that charges per lead booked, ticket resolved, or document drafted—tying revenue to measurable results. It’s the software analogue of value-based care. 3. Hybrid “starter bundle plus runway” tiers: a predictable monthly fee with a healthy allowance of AI credits, then pay-as-you-go beyond that. This balances budget certainty for customers with upside capture for the vendor. Across our portfolio, a few design principles keep showing up: 1. Anchor on a metric the customer already tracks. If your product shortens sales cycles, price per opportunity accelerated—not per login. 2. Bundle enough volume to eliminate credit anxiety. No one wants to ration prompts. 3. Expose real-time usage. Transparent dashboards prevent bill shock and build trust. 4. Instrument cost early. Metering and billing belong in the product backlog, not the finance queue. 5. Plan for non-linear jumps. When a model upgrade multiplies compute, re-grade tiers before your gross margin does it for you. AI’s promise is to shift human effort from repetitive execution to higher-order creativity. If our pricing still counts bodies instead of business results, we undermine that promise. The companies that map price to outcomes—while keeping the buying experience refreshingly simple—will capture the most upside. I’d love to hear how others are managing the move from seats to usage and outcomes. What’s working, what still feels messy, and where do you see the biggest opportunities to innovate on pricing? #appliedAI #pricing #startups

From Airline Seats to AI Tokens: The Revenue Management & Dynamic Pricing principles remain timeless. The “capacity” may have shifted from aircraft seats to GPU cycles, but the core challenge is unchanged: price intelligently, or risk leaving value on the table. Pricing is never just about revenue extraction. It’s about aligning value delivered, costs incurred, and demand patterns — whether for an airline seat, a hotel room, or an AI token. In streaming, flat subscriptions work because marginal costs per user are almost zero. In AI, however, serving each request consumes expensive GPU cycles and electricity. Here, token-based pricing aligns usage with cost: ~$5 per million input tokens, ~$15 per million output tokens. Why this matters: Fairness: Heavy users pay proportionally more. Cost recovery: Output-heavy tasks (costlier to run) are priced higher than simple inputs. Efficiency: Firms are incentivized to optimize prompts, reduce waste, and choose model tiers wisely. Looking ahead, hybrid models will likely dominate — much like telecom plans: a base subscription plus overage fee. This balances customer predictability with provider sustainability. Indian Institute of Management, Lucknow

SaaS Trained Us to Ignore Marginal Cost. AI Won’t. For two decades, the zero-marginal-cost assumption made us lazy. It convinced us that if we just built the software, the margins would engineer themselves. This dynamic produced 80%+ gross margins and justified why, at the peak, some companies traded at 20x revenue. But AI breaks that assumption. SaaS is a gym membership: pay once, visit forever. AI is a factory: every unit of output requires raw materials, energy, and labor. You don't scale a factory by just adding "users"; you scale it by managing yield. Every additional unit of usage consumes something real: tokens, inference, latency, and watts. Consider an AI product priced at $100 per user/month: A light user generates $5/month in inference cost. A power user generates $65/month. At small scale, the averages hide the problem. But when the top 10% of customers drive 80% of your compute costs, your unit economics collapse, even though your pricing looks "SaaS-like". This creates hard truths for the AI founder: Usage-based pricing isn’t optional: Flat per-seat pricing acts as a subsidy for your most expensive users. Wrappers are Resellers: If your differentiation is just a UI on top of Anthropic, you aren't a software company. You are a low-margin services firm paying a "compute tax" to a vendor who is also your biggest competitor. Gross margins matter again: Fast-growing AI apps are often operating at 25–40% gross margins, not 80%. The Valuation Trap: SaaS multiples assume software economics. Do not raise at a $100M+ valuation if you have manufacturing-style COGS. When you raise at SaaS multiples without SaaS margins, you are forced to "grow into" a valuation that physics won't allow. The result isn't just a missed target; it’s a consequential liquidation preference stack that can erase the common stock even in a "successful" exit. The Bottom Line: In the SaaS era, margins were a gift of software physics. In the AI era, margins are a discipline you have to engineer.

AI is getting cheaper. Over the past 12–18 months, more operators and investors have begun saying the same thing: AI is breaking SaaS economics. Marginal cost is back. Seat pricing is under pressure. Inference is not free. All of that is directionally correct. But it is incomplete. The more important question is not whether AI introduces marginal cost. The question is who captures the efficiency gains as inference gets cheaper. Some products expand margin. Some expand revenue. Some quietly subsidize heavy users. Others compress under competition. The difference is not model quality; rather, it is cost exposure and pricing architecture. I have been developing a framework to think about this, which is what I call the Jevons Capture 2×2. It maps products (not companies) across two structural dimensions: who bears inference cost and volatility, and who retains efficiency gains as AI becomes more efficient. From that lens, four economic states emerge: Structural Winners, Defensive Beneficiaries, Margin Tension, and Consumer Surplus Trap. More importantly, products migrate between them. Bundled seats can move to credits or metering. Bounded optimization surfaces can expand into open-ended agentic execution. Structural Winners can compress if competition erodes pricing power. Much of the public discourse stops at “move to usage-based pricing.” This piece connects Jevons effects, unit economics and product-level P&L mechanics...AND introduces a migration framework to think about how these states evolve. The theme may now be common. The framework is not.

The economics of large language models are about to collide with enterprise reality. For the first wave of adoption, AI came wrapped in subsidized – often per-seat pricing. Tokens felt free. That made experimentation easy for those now addicted to their daily agentic fix. In some organizations, usage itself has become the goal. The rise of the inane “tokenmaxxing” trend celebrates volume over value, as if inputs could be equated to outcomes. But now that the LLMs have people hooked on agents, the real bill is coming due. Hundreds of billions of dollars are being poured into AI infrastructure, and someone has to pay for it. Recent moves by major model providers signal a shift toward token-based pricing becoming the norm. What started as a buffet is turning into a meter. The bigger surprise for many executives will not be visible output tokens. It will be the quiet accumulation of “reasoning” tokens, consumed in the background every time a model makes a decision from scratch. And then rethinks it. And rethinks it again. These are the most expensive tokens of all, though no one explicitly asked for them. Reasoning AI is powerful. It is also costly, unpredictable, and often unnecessary. Re-reasoning every operational decision might feel sophisticated, but it is the opposite of being disciplined. Once logic, policy, and workflows are well designed, endlessly reinventing them at runtime adds unnecessary cost and risk without adding value. Reasoning at design time, where creativity, exploration, and insight matter most, is exactly where generative AI is at its best. But runtime execution is a different animal. There, predictability, consistency, and cost control matter more than creativity. Using a lighter-weight semantic AI capability at runtime allows agents to follow the established workflows at less cost and higher efficiency. This is how scaling AI works best. Sustainable AI is not magic. It is smart trade-offs. The next phase of enterprise AI will be defined less by flashy demos and more by discipline. Predictable outcomes. Predictable costs. Clear intent about when reasoning is worth paying for, and when it is not. Organizations that start addressing agentic addiction early, by designing for token efficiency and architectural clarity, will not just save money. They will build systems that are easier to trust, easier to govern, and easier to scale, especially as AI stops feeling free and starts feeling real.