The strongest AI companies are not choosing between subscriptions and usage-based pricing. They are building pricing systems that expand as customer value expands.
AI has created a strange pricing problem.
The underlying technology is expensive to operate, customer usage is unpredictable, and the value users receive can vary dramatically from one task to another.
One person might use an AI product twice a week. Another might use it to generate hundreds of assets, write production code, run workflows, or automate an entire business process.
Charging both customers the same flat subscription eventually stops making sense.
But pure usage-based pricing creates its own problems. Customers dislike unpredictable bills. They struggle to understand what a token is worth. And many buyers still want the simplicity of knowing what they will pay each month.
The emerging answer is not subscription or usage-based pricing.
It is a combination of both.
In my recent conversation with Andrew Garvin, Co-Founder of Metronome, we discussed how monetization is becoming part of the product infrastructure itself. Pricing, billing, metering, entitlements, packaging, and product experience can no longer be treated as separate systems.
The pricing page is where that infrastructure becomes visible to the customer.
Lovable, Notion, and Gemini are three strong examples of companies approaching the problem from different directions:
- Lovable starts with credits and product usage.
- Notion starts with seats and layers usage-based AI products on top.
- Gemini bundles access, capabilities, storage, tools, and usage limits into a broad consumer pricing ladder.
Their models are not identical. But the underlying principles are remarkably similar.
1. Lovable makes the value metric visible
Lovable’s pricing page immediately establishes a familiar SaaS structure:
Free, Pro, Business, and Enterprise.
That part is conventional. What makes the model interesting is what sits inside the paid plans: credits.
The subscription establishes a predictable monthly commitment. Credits determine how much product capacity the customer receives.
That makes Lovable a clear example of hybrid monetization.
Customers are not simply paying to unlock a static collection of features. They are buying access to the product and a defined amount of productive capacity.
The free plan creates a low-friction entry point
Lovable allows customers to experience the product before making a financial commitment.
This is particularly important for AI products because users often need to see the output before they understand the value.
A landing page can explain what an AI builder does. A demo can show it. But the moment a user generates their own application is what creates conviction.
The free plan is therefore not merely a cheaper package. It is part of the acquisition engine.
Credits create a visible expansion path
The Pro and Business plans make monthly credits visible and selectable directly on the pricing page.
That matters.
Usage-based pricing becomes difficult when the value metric is hidden, abstract, or introduced only after the customer has already purchased.
Lovable tells customers upfront that credits are part of the model. The customer can see that higher usage will require more capacity.
This creates a natural progression:
- Start free.
- Upgrade to receive more credits and capabilities.
- Purchase additional credits when usage increases.
- Move into a larger plan as the team and operation become more sophisticated.
The pricing architecture mirrors customer growth.
Rollovers and top-ups reduce the weaknesses of credits
Credit-based models can create anxiety.
Customers worry about unused balances. They also worry about suddenly running out of capacity during an important project.
Lovable addresses both sides of the problem through credit rollovers and on-demand top-ups.
Rollovers reduce the feeling that unused capacity is wasted.
Top-ups prevent customers from being forced into a larger recurring commitment simply because of a temporary spike in usage.
This creates flexibility without abandoning the predictability of a subscription.
Unlimited users encourage adoption
Lovable also avoids relying entirely on per-seat monetization.
Unlimited users make it easier for a customer to invite more people into the product. That reduces internal friction and can help Lovable spread across a team.
Instead of charging for every person who enters the workspace, Lovable can capture more value as the team collectively consumes more product capacity.
That is a critical distinction.
Seat-based pricing monetizes access.
Usage-based pricing monetizes activity.
For collaborative AI products, charging primarily around activity can make expansion easier because customers do not have to debate the cost of every additional collaborator.
2. Notion is building a monetization stack
Notion’s pricing architecture starts from a different place.
Its core value metric is still familiar: price per member per month.
That works because Notion is fundamentally collaborative software. As more people join a workspace, the product becomes more useful and the account becomes more valuable.
But Notion is no longer monetizing only collaboration and software access.
It is increasingly monetizing intelligence, automation, agents, search, and completed work.
This is where the pricing page becomes particularly interesting.
The packaging follows the customer journey
Notion divides its pricing structure into two clear sections.
The left side serves individuals and smaller teams through Free and Plus.
The right side serves growing businesses and enterprises through Business and Enterprise.
This creates a clean progression:
- Personal organization
- Small-team collaboration
- Business-wide workflows
- Enterprise control and governance
The structure is easy to understand because the packages represent increasingly sophisticated customer needs.
The user does not need to compare every feature individually before understanding which general plan is intended for them.
Business is positioned as the default growth plan
The Business plan carries a recommended label.
This appears minor, but pricing pages are decision-making environments. Customers are frequently looking for guidance, not unlimited choice.
A recommended plan reduces cognitive load.
It tells a growing team: this is probably where you belong.
The best pricing pages do not simply present options. They help customers make a decision.
Enterprise features are packaged around organizational risk
Notion’s Enterprise plan does not rely on consumer-style feature volume.
It focuses on capabilities that matter to larger organizations:
- User provisioning
- Audit logs
- Security controls
- Compliance connections
- Domain management
- Administrative governance
- Customer success
This is effective because enterprise buyers are not purchasing only more product usage.
They are purchasing control, reduced risk, implementation support, and organizational confidence.
Enterprise pricing is therefore paired with a Contact Sales motion rather than a simple self-serve checkout.
That allows Notion to price around account complexity, scale, support requirements, deployment conditions, and strategic value.
Agents create a second monetization layer
The most important section of Notion’s pricing page may sit beneath the traditional plan comparison.
Custom Agents are free to try and then priced through Notion credits. Workers also introduce a usage-based component.
This gives Notion two monetization systems:
- Seat-based subscriptions for the workspace.
- Credit-based monetization for AI work and automated activity.
That is the foundation of a more modular pricing model.
A team can pay for access based on the number of members while also paying for additional work completed by agents.
This matters because the economic value of an agent is not necessarily tied to the number of humans in the workspace.
One employee might run thousands of automated tasks. Another might barely use the AI features.
Charging only per seat would fail to capture that difference.
The credit layer gives Notion room to monetize the amount of AI activity generated inside the account without abandoning the predictable seat-based model customers already understand.
3. Gemini captures a wide range of willingness to pay
Gemini faces a broader packaging challenge.
It is serving casual consumers, students, professionals, creators, developers, and advanced AI users within the same product ecosystem.
A single subscription would either be too expensive for casual users or too limited for power users.
Google addresses this with a wide pricing ladder:
- Free
- Google AI Plus
- Google AI Pro
- Google AI Ultra
The price gap between the plans is significant. That is intentional.
Gemini is not simply offering small feature upgrades. It is attempting to capture several distinct levels of willingness to pay.
Free creates mass distribution
The free plan gives users access to a meaningful collection of AI capabilities.
This lowers the cost of experimentation and allows Gemini to become part of the user’s daily behaviour before asking for payment.
For a broad consumer product, free access is not only an acquisition strategy. It is also a distribution strategy.
Every free user becomes a potential future subscriber as their usage, reliance, or sophistication increases.
Usage limits make the upgrade tangible
AI products often struggle to explain why one plan should cost more than another.
Gemini uses usage multiples to communicate the difference.
The customer sees language such as two times higher access, four times higher access, or dramatically higher limits at the Ultra level.
That is more understandable than explaining the underlying cost of tokens, inference, model classes, or compute.
The user does not need to understand the technical input. They only need to understand that the paid plan allows them to do more.
The value metric is therefore translated into customer language: access and limits.
The Ultra plan acts as a premium anchor
Google AI Ultra sits far above the lower-priced plans.
The plan is designed for customers who want the highest limits, early access, advanced models, agent capabilities, creative tools, and developer-oriented products.
It also plays another pricing role: anchoring.
Once a $99.99 plan exists, the $19.99 Pro plan can feel comparatively accessible.
The Ultra plan does not need to become the most popular package to improve the pricing architecture. It gives high-intensity customers somewhere to go while establishing a premium reference point for the rest of the page.
Bundling increases perceived value
Gemini’s plans combine several forms of value:
- Model access
- Higher usage limits
- Creative tools
- Search capabilities
- Coding agents
- Notebook functionality
- Storage
- Google application integrations
- Early access to new features
Bundling makes the package feel larger than a subscription to a single chatbot.
Google is monetizing an ecosystem.
That creates stronger retention opportunities because the customer can use different parts of the bundle for different jobs.
It also creates cross-product expansion. A user may join for Gemini, begin using Flow, rely on Notebook, experiment with Jules, and eventually develop workflows that make the broader subscription difficult to replace.
The pricing ladder follows user sophistication
Gemini’s progression is not purely based on company size.
It is based on user intensity.
A casual user begins with Free.
A regular user may upgrade to Plus.
A professional or creator moves to Pro.
A power user, developer, or AI-heavy operator may justify Ultra.
This is land-and-expand applied to an individual customer rather than an organization.
The account expands as the user becomes more sophisticated and dependent on the product.
The shared model: predictable access plus variable consumption
Lovable, Notion, and Gemini all approach monetization differently.
But each pricing system is moving toward the same basic structure:
A predictable recurring payment establishes access. Variable consumption determines expansion.
For Lovable, the recurring payment includes monthly credits.
For Notion, seat-based subscriptions provide access to the workspace while credits monetize agents and AI activity.
For Gemini, each subscription tier bundles higher usage limits, premium capabilities, and access to a wider set of tools.
None of these companies are relying on pure usage-based pricing.
That would create too much uncertainty for many customers.
They are also not relying entirely on flat subscriptions.
That would leave money on the table when high-intensity users consume significantly more infrastructure and receive substantially more value.
The hybrid model gives both sides something they need.
The customer receives predictability.
The company receives an expansion mechanism.
What these pricing pages teach us
1. The free plan needs to produce a real outcome
A free plan should not be a product tour disguised as a package.
The user should be able to complete something meaningful.
Lovable lets users begin creating.
Notion allows individuals to organize real work.
Gemini gives users access to a broad set of AI capabilities.
A useful free plan creates the moment when the customer understands the product’s value.
The paid plan should then remove a constraint: capacity, collaboration, usage, control, sophistication, or scale.
2. Packaging should follow customer maturity
Customers should be able to recognize themselves in the plan architecture.
A strong pricing ladder answers three questions:
- Where should I start?
- What will cause me to upgrade?
- Where can I go when my needs become more complex?
Lovable follows the journey from builder to team to organization.
Notion follows the journey from individual to small team to business to enterprise.
Gemini follows the journey from casual user to professional to AI power user.
The names, descriptions, features, and prices should reinforce the same progression.
3. The value metric must be understandable
A technically precise value metric is useless if the customer cannot understand it.
Tokens may map closely to AI infrastructure costs, but most customers do not know how many tokens they need.
Credits can simplify this, but only when customers understand what the credits allow them to accomplish.
Usage limits can work when the difference between plans is visible.
Seats work when value increases with collaboration.
The best value metric is not simply the metric that matches cost. It is the metric that customers can connect to value.
4. Usage should create expansion, not fear
Variable pricing becomes dangerous when customers feel that normal product use could create an unexpected bill.
That is why hybrid models are so powerful.
Subscriptions create a predictable base.
Included usage allows customers to develop habits.
Limits, top-ups, credits, or higher-capacity plans create expansion after the customer already understands the value.
The goal is not to charge for every possible action.
The goal is to create a fair relationship between increasing customer value and increasing revenue.
5. Enterprise pricing solves a different problem
Enterprise is not simply a larger self-serve plan.
Larger organizations require security, compliance, provisioning, governance, support, procurement, and legal flexibility.
Lovable and Notion both separate the enterprise motion from the standard product checkout.
That allows pricing to reflect the actual complexity of serving the account.
Trying to compress enterprise value into one public monthly price can significantly underprice the product.
6. The pricing page is part of the product
A pricing page is not merely a chart created by the finance team.
It communicates:
- Who the product is for
- What the company believes creates value
- How customers are expected to grow
- Which behaviours will be monetized
- Where the product is going next
The plan structure should be connected to the entitlement system.
The credit model should be connected to product usage.
The upgrade experience should be connected to the moments when customers hit meaningful constraints.
The billing system should support experimentation rather than locking the company into one permanent model.
Pricing is product infrastructure made visible.
The bigger shift in AI monetization
The cost of AI will continue to change.
Models will become cheaper. Open-source alternatives will improve. Capabilities that feel premium today will become widely available.
That means companies cannot build durable monetization around model access alone.
The stronger opportunity is to monetize the value pockets surrounding the model:
- Workflows completed
- Applications created
- Research performed
- Assets generated
- Processes automated
- Team collaboration enabled
- Risks controlled
- Time saved
- Revenue produced
This is why action-based and outcome-oriented pricing will become increasingly important.
But the market is unlikely to move directly from subscriptions to pure outcome pricing.
Measurement remains difficult. Outcomes are disputed. Customer systems are fragmented. And buyers still prefer predictable budgets.
For the foreseeable future, the most practical model will often be a blend:
Subscription for predictability. Usage for expansion. Packaging for segmentation. Enterprise pricing for complexity.
Lovable, Notion, and Gemini are not the final versions of this model.
Their plans will evolve as customer behaviour, AI economics, and product capabilities change.
That is exactly the point.
The winners in AI pricing will not be the companies that discover one perfect price.
They will be the companies that build a monetization system capable of evolving with the value their product creates.
Leave a Reply