In brief

‘VC subsidy’ is an informal term for the benefit users may experience when investors finance a service’s growth and losses. VC stands for venture capital. It is not a fixed financial category, nor is it proof that every low-cost AI service operates at a loss. It refers to a possible funding dynamic, not a universal explanation for API prices.

How can financing support a low price?

A company may raise capital to fund development, infrastructure and customer acquisition before revenue covers all those expenses. A supplier may also provide temporary credits or promotional credits. Investment funds and supplier credits are different forms of support. In both cases, part of the bill may temporarily fall outside the direct customer price.

Why does a low token price prove very little?

A provider can operate efficiently, utilise existing capacity or offer a smaller model. Furthermore, operating loss is not the same as a loss on every API request: research costs and other expenditure also count towards this. Without suitable information on costs and revenue, you cannot determine the extent to which a specific service is subsidised. You should therefore not present assumptions as facts in this regard.

A fictional example

Suppose a team receives temporary cloud credits for a trial. During that trial, a workflow appears to cost almost nothing. Once the credit has been used up, the standard usage charges apply. This example illustrates the difference between a temporary bill and ongoing costs. It says nothing about the profitability or future pricing of a particular AI provider.

How should you handle this?

Carry out a calculation without taking temporary discounts into account and include all steps in the workflow. Also investigate how a different usage volume or a different model would affect the costs. You do not need to predict future price rises to do this. It is helpful simply to make the assumptions transparent and avoid locking in a model choice unnecessarily.

What does this mean for AITJE?

Compare API usage and in-house infrastructure in terms of total costs and usable results. An in-house server requires purchase, power and management; an external service offers different economies of scale and dependencies. Let a pilot demonstrate what works and then assess the structural setup. A temporary low price can help you get started, but is not a complete business case.

From concept to practice

How AITJE applies this

When comparing costs, AITJE also factors in usage without temporary credits or trial discounts. We highlight which rates and volumes determine the calculation. This helps to translate a pilot into long-term use. In doing so, we make no assumptions about hidden costs from a provider: the choice is based on your own costs, data policy and options for customising the environment.

View advice and analysis

From explanation to application

Would you like to know what this means for your work?