The AI Pricing Trap: Corporate Productivity Fuelled by a  92% Venture Capital Illusion, warns Prof Ratnatunga

As organizations globally rush to integrate generative artificial intelligence into their core operations, a stark economic warning has been issued to the business community: current AI productivity gains are being heavily subsidized by venture capitalists, setting a financial “trap” for companies that become dependent on the technology.

In a newly published analysis, The AI Pricing Trap: The Economics of AI Addiction, Professor Janek Ratnatunga, CEO of the Institute of Certified Management Accountants Australia & New Zealand (ICMA ANZ), reveals the unsustainable unit economics powering the current AI boom. He warns that businesses are being systematically hooked on artificially low pricing structures that cover only a fraction of true computing costs.

Prof Ratnatunga says, “Walking into the AI market today is like walking into a car dealership, picking out a  vehicle, and being told you only owe  because an invisible investor is covering the rest. It is a brilliant strategy to breed immediate infrastructure dependency, but the money is running out. When the real bill finally lands, whole industries will face severe financial shocks.”

Using the example of heavy users utilizing tools like Anthropic’s Claude Code, Prof. Ratnatunga exposes a massive disparity between market pricing and operational reality. The Real Cost an advanced developer or power user can easily burn through  tokens (the fundamental units of AI computation) a year. Through standard, un-subsidized developer APIs, this level of compute costs approximately  annually. The Illusion is  that under current flat-rate premium subscriptions (priced at roughly  per month), users pay just  a year—enjoying a massive, subsidized  discount.

The report notes that with OpenAI on track to lose a projected  in 2026, a standard consumer subscription covers less than  of what an active power user actually costs to serve.

Furthermore, the analysis dismantles the tech industry defence that computing is naturally becoming cheaper. Prof. Ratnatunga introduces the concept of the “Token Tax”—explaining that modern, autonomous “agentic” workflows burn through  to  times more tokens than simple chat sessions by quietly spawning subtasks, checking their own work, and running background tests. This exponential explosion in token volume completely obliterates any incremental hardware efficiencies gained by chip manufacturers.

The strategy mirrors the classic penetration pricing playbooks previously executed by tech giants like Uber, Spotify, and Adobe. Uber famously spent billions of investor cash to artificially suppress ride fares and eliminate taxi competition, only to aggressively raise its “take rate” from  to  once market dominance was secured.

“We are seeing the exact same memo deployed in AI,” Prof. Ratnatunga notes. “Law firms running document reviews at five cents on the dollar, marketing agencies turning out massive campaigns at impossible prices, and hospitals trialling diagnostic tools are all operating on a economic mirage. Industry analysts already expect consumer subscription tiers to double in price over the next two years.”

Prof Ratnatunga warns corporate decision-makers against treating AI platforms as mere application software. Because these tools are fundamentally shifting how employees execute cognitive work, they are rapidly becoming corporate infrastructure.

If a company’s operational efficiency relies on a tool priced at a fraction of its cost, a sudden market correction or a  increase in pricing will decimate corporate margins. The report strongly urges CFOs and technology strategists to model their long-term financial forecasts around the raw, un-subsidized cost of compute rather than today’s promotional subscription rates.

 
Released by: The Institute of Certified Management Accountants (Australia and New Zealand)
Date: 9 July 2026
Location: Melbourne, Australia