Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Repair Windows errors before they cause bigger problemsFix Now →iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
An AI product has a sustainable business model when customers keep receiving meaningful value, revenue captures enough of that value, and the cost of delivering it does not overwhelm the revenue as use grows. Evaluate those links together: adoption alone, impressive revenue growth, or a high company-wide gross margin cannot prove durability by itself.
Start with repeat customer value
Ask whether the product solves a recurring problem important enough for customers to keep using and paying for it. Look beyond sign-ups and initial pilots to repeat use, renewals, account expansion, and evidence that the product is embedded in a consequential workflow. Battery’s 2025 State of AI report recommends considering product usage, customer value, and gross retention alongside revenue growth and efficiency; it does not establish a universal pass mark for any of these measures.
Usage is meaningful only in context. A high task count could reflect valuable repeat work, but it could also reflect experimentation or low-value activity. Connect usage to outcomes customers care about, such as work completed, time saved, or a workflow they continue to rely on. Then test whether that value persists across customer cohorts and supports renewal or expansion.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Clear out junk files and repair common Windows errors3Fix the driver behind crashes, sound loss and screen glitchesCheck whether pricing captures the value and usage
Identify how the company earns revenue: fixed subscription, consumption-based charges, services, or a combination. Then compare what customers pay with the amount and type of use they generate. A flat subscription can work when the price covers typical delivery costs and the customer sees enough value to renew. It can become a subsidy if a small group of heavy users consumes far more costly resources without generating proportionate revenue.
#1 Best Overall
Consumption pricing can align revenue more closely with usage, but it does not automatically produce healthy economics. Consider whether customers can predict their bills, whether usage translates into customer value, and whether revenue is realized consistently. Services may help customers adopt a product, but labor-heavy delivery can also make revenue less profitable or less repeatable.
Calculate the cost of delivering the service
Include the costs that rise with usage or customer complexity, not just model inference. Depending on the product, relevant costs can include external model services, cloud and infrastructure, and support or implementation labor. BigBear.ai’s 2025 SEC filing describes a hosted SaaS product using third-party large language models, illustrating that delivery costs may include services supplied by another model provider.
Review gross margin and, when the data allow it, contribution economics by customer, workflow, or usage tier. A blended company-wide margin can hide an expensive use case or a contract that requires unusually intensive support. Consistent definitions matter: compare like periods and include the relevant inference, infrastructure, and service costs rather than treating an accounting label as a complete explanation.
- Cost per successful task or output: Track a comparable unit of work, and count successful outcomes rather than raw requests when those differ materially.
- Margin by cohort or use case: Check whether the economics vary with customer type, workflow, usage intensity, or deployment stage.
- Cost mix: Separate inference and infrastructure costs from labor or support where possible, so a change in one does not obscure another.
The available company disclosures show that inference efficiency, infrastructure utilization, and labor mix can matter. They do not establish a universal accounting template or a required gross-margin level for AI products.
Test what happens as usage scales
More users and tasks do not necessarily make a product more sustainable. Ask whether revenue grows with the additional use, whether the cost of serving each comparable task falls or rises, and whether the product still delivers the quality customers value. Improvements in unit economics are useful only if they do not come from degrading the service or shifting costs out of view.
Company-reported examples demonstrate why context matters, not what every AI business should target:
| Company-reported measure | Reported figures | Context |
|---|---|---|
| AI+SaaS segment gross margin | 86.3% in 2024; 87.7% in 2023 | An issuer’s 2025 HKEX listing document attributed the decrease partly to labor costs associated with expansion into niche enterprise markets. Company listing document. |
| AI-native product gross margin | Negative 8.1% in 2024; negative 380.2% in 2023 | An issuer’s 2025 company listing document reported that margins improved alongside cost of sales declining from 124.7% to 87.8% of revenue, and attributed the shift partly to lower inference costs and better infrastructure utilization. Company listing document. |
These are different companies’ reported results, with differing business mixes and accounting contexts; they are historical examples, not industry benchmarks or forecasts. For a product under review, compare its own unit costs, customer value, and margins over time using consistent definitions. Investigate whether a change came from greater delivery efficiency, a different customer mix, pricing, or a change in product quality.
Separate realized revenue from future commitments
Do not treat contracted obligations, capacity purchases, or usage forecasts as if they were revenue already earned. In its FY2025 annual report, C3.ai cautions that remaining performance obligations can be less predictive under consumption pricing. Renewal timing, conversion of production deployments, capacity purchases, contract term length, and seasonality can all affect when those obligations become revenue.
For a durability assessment, distinguish recognized revenue from contracted amounts and forecasts, and ask what assumptions must hold for future commitments to convert into sales. Backlog-style disclosures can add context, but are not a standalone revenue forecast.
Rank #4
Compare products on the same basis
When assessing multiple AI products or business models, use the same definitions and time periods. Otherwise, apparent differences may reflect measurement choices rather than stronger businesses.
- Customer value and repeat adoption, including retention and expansion.
- Revenue composition: recurring subscription, consumption, services, and contract commitments.
- Gross margin and contribution economics after relevant model inference, cloud, infrastructure, and service labor.
- Sensitivity to usage intensity, customer mix, and external model or infrastructure costs.
- Whether unit economics improve without damaging product quality or customer value.
Make a product-specific judgment
There is no universal gross-margin, retention, or growth threshold established by the cited evidence. The answer depends on the product’s customers, pricing, delivery costs, and ability to improve efficiency while maintaining value. A general framework can identify what to test, but it cannot establish a particular company’s customer-level margins, retention, cash needs, or capacity to fund growth without evidence about that company.
Free tools Windows power users keep installed
One-click scans. No signup required.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
