The Rising Cost of Artificial Intelligence

Artificial Intelligence (AI) has transformed the technology sector at a remarkable speed over the past few years. What once appeared to be an experimental innovation, has quickly evolved into a mainstream tool used across almost every part of society.

From students and freelancers to professional services firms, AI has become embedded in everyday workflows, often faster than the broader public debate about its long-term consequences has been able to keep pace.

The appeal of AI is obvious. With tools like ChatGPT, Claude, Gemini, GitHub Copilot and many more such tools, tasks that would once have required hours of human effort can now be completed in seconds and minutes. AI systems are remarkably efficient at drafting documents, creating presentations, analysing data and automating tedious administrative tasks. Although this has driven lower operational costs and higher productivity for organisations, this quick uptake of AI has also raised serious concerns. More and more people are finding themselves without employment.[i] Why? Because organisations may view the use of AI as cheaper than paying the full salary of an individual, or so it seemed at first.

Over the past few months, AI businesses have been providing consumers with access to increasingly powerful AI tools at remarkably low costs. Advanced services have been made available to users for as little as $20 a month.[ii] These premium tools now do far more than generate basic text responses. With the help of AI agents that can perform complicated and time-consuming tasks, the advanced versions can now create apps from prompts, edit and design websites, automate workflows and even generate high-resolution images.

However, the period of cheap, flat-rate AI subscriptions may not be sustainable according to the recent developments.[iii] As reported by PCWorld, Microsoft-owned GitHub appears to be one of the first major AI providers to publicly abandon flat-rate subscription models. In response to growing concerns about the true cost of advanced AI services, the company is reportedly moving its Copilot AI offerings toward more costly usage-based pricing structures.[iv] The main issue is that large amounts of processing power are required for more powerful and sophisticated AI systems. Traditional AI chat features are relatively modest in comparison with the capabilities now being offered by newer tools. These systems are no longer limited to producing basic text outputs or responding to prompts; they are increasingly capable of generating software, carrying out multi-steps tasks, providing accurate advice and editing documents with little to no human supervision.[v] These functions have drastically raised operating costs and put significantly greater strain on the computing infrastructure. As a result, there is a gap between user expectations and economic reality when such services are offered at low fixed rates.

While AI service providers continue to bear the enormous infrastructure costs necessary to maintain those systems[vi], consumers have grown accustomed to having access to remarkably powerful AI capabilities at comparatively low prices. Anthropic, the company behind Claude AI, has mentioned that the purpose of its tools was primarily conversational and that certain newer functions were never intended to be supported by its current subscription models. It is reportedly experimenting with removing Claude Code from its Pro plan and adjusting Pro and Max users’ use limiting in an effort to discover a combination that makes those plans financially viable.

With these ongoing rises in subscription prices, it is hard to imagine that ChatGPT Plus and Pro will not eventually follow suit, even though OpenAI has so far signalled confidence in maintaining its flat-rate plans.[vii]

Ultimately, the market may be approaching a pricing correction. The true cost of advanced AI is likely to become clearer as providers seek business models that reflect the computational and infrastructure demands of these tools. What users currently access for $20 or $100 per month may, in time, cost significantly more particularly for business and enterprise users. Rather than signalling the end of AI’s appeal, this shift may simply mark the end of artificial cheap access to increasingly sophisticated technology. The magic of AI maybe fading…….

© Lawrence Power 2026


[i] Trade Arabia (2026) ‘AI “proving more expensive than human workers”, 10 May 2026. Available at https://www.tradearabia.com/News/462306/AI-‘proving-more-expensive-than-human-workers’-

[ii] Patterson B. (2026) ‘The $20 AI subscription era has become untenable’,  PCWorld 01 May 2026. Available at https://www.pcworld.com/article/3129267/the-20-ai-subscription-era-has-become-untenable.html

[iii] Patterson B (2026) ‘GitHub Copilot’s price shakeup could signal the end of cheap AI coding’, PCWorld 27 April 2026. Available at https://www.pcworld.com/article/3125621/github-copilots-price-shakeup-could-signal-the-end-of-cheap-ai-coding.html

[iv] Patterson B. (2026) ‘The $20 AI subscription era has become untenable’,  PCWorld 01 May 2026. Available at https://www.pcworld.com/article/3129267/the-20-ai-subscription-era-has-become-untenable.html

[v] The Economic Times (2026) ‘Replacing humans with AI is turning out to be costlier than expected, Uber, Nvidia execs say’, 08 May 2026. Available at https://economictimes.indiatimes.com/nri/work/replacing-humans-with-ai-is-turning-out-to-be-costlier-than-expected-uber-nvidia-execs-say/articleshow/130952035.cms?from=mdr

[vi] The Guardian (2026) ‘AI costs are coming to consumers’, 05 May 2026. Available at https://www.theguardian.com/global/2026/may/04/ai-costs-are-coming-to-consumers

[vii] Patterson B. (2026) ‘The $20 AI subscription era has become untenable’,  PCWorld 01 May 2026. Available at https://www.pcworld.com/article/3129267/the-20-ai-subscription-era-has-become-untenable.html

Leave a Reply

Your email address will not be published. Required fields are marked *