ChatGTP, Claude, CoPilot new pricing models mean big changes
A OPINION by Mike Stone, Andover NY
The AI bubble may not be bursting, but it is fundamentally changing, and quickly.
For the past several years, AI has altered our digital landscape and it’s done so in a way that was never sustainable. Companies have been going “All In” on AI solutions, to the point of reducing labor forces, but that may prove to bite them in the posterior.
You see, the computing power to run AI solutions like OpenAI’s ChatGPT or Anthropic’s Claude is immense and expensive. The hardware, physical footprint, power cost, cooling cost, and labor cost adds up to an absolute beast of an infrastructure. Yet most of its valuation was based on hype. With billions of users hammering away at these products, oftentimes paying little-to-nothing, it wasn’t the profitable product; it was the hot product. But at some point, valuation needs to meet reality and companies can’t sustain themselves while bleeding cash.
OpenAI and Anthropic are private companies (for now) and not subject to annual financial reports, however, many internal documents and financial leaks have provided us with a clear enough view of some startling numbers. From 2023-2025, it is estimated that OpenAI has net losses in the tens of billions of dollars with internal projections suggesting they may not see a profit until 2029 or 2030, and by that point their cumulative losses are estimated to be close to $100 billion dollars. Anthropic’s landscape isn’t quite as bleak but it’s certainly not rosy, with Claude not investing so heavily in marketing, but its focus on code development has it using more resources.
AI for me is a double-edged sword. I acknowledge its many failings and issues while still utilizing it within my own moral guardrails. In all my usage of AI as a value-added resource in my life, one thing I’ve never lost sight of is the soulless nature of it. Although it has been a godsend in programming, log parsing, and casual inquiries, I detest generative AI in pictures, music, and videos for that very reason: a soul is a prerequisite for art, in my opinion.
More than once, I’ve made the argument that AI is very much like driving a car. There is an ecological, societal, infrastructure, and human cost to its operation… but while I know it is problematic, it still enriches my life and I acknowledge at least some level of hypocrisy on my own part. Perhaps it’s a small consolation that I do so with eyes wide open. Perhaps.
So that’s the general outlook – but what is changing *now* that is causing me to write this editorial? Since I am a part of the landscape that utilizes AI daily on a personal and professional level, I also have my ear to the ground with the stirrings of the industry. Within the last week or so, many internal changes have been announced. GitHub’s Copilot implementation is changing their billing structure so instead of a monthly subscription with a finite number of premium requests, it is changing to a much more restrictive “token-based” model. The more complex the task, the faster your monthly allotment will run out. “Today, a quick chat question and a multi-hour autonomous coding session can cost the user the same amount. GitHub has absorbed much of the escalating inference cost behind that usage, but the current premium request model is no longer sustainable.”, the company reported on April 27th.
Anthropic is also fundamentally changing its pricing infrastructure, with The Information reporting a deviation from monthly flat fees to a usage-based model that will wind up costing users considerably more for the same workload.
If this shift in pricing model trends continues, it has the potential to funnel down to every site, app, resource, and software tool using AI – from Agentic usage in software IDEs to friendly chatbots answering questions about making dandelion wine.
As impressive and problematic as AI is, the business model was fundamentally flawed from the start. These price increases are simply the market correcting itself and business models shifting to ones that are more sustainable and financially responsible. Is this too little, too late? Is this still a bubble that could burst, collapsing the biggest market traders and index funds? Maybe. Maybe not. Time will tell, but this indicates an important shift from the wild west we’ve been experiencing the last few years.
The ones that will be hurting the most with these changes are perhaps the ones that deserve to: those companies that have reduced their workforce and invested the bulk of their productive workflows toward AI. Over the past month or so, many have been discontinuing their AI subscriptions and begun hiring actual programmers and developers again. But doing so takes time and with these new shifts, will also take a good deal of money as people with these skills will be in high demand once again.
In the grand scope of mankind’s future, the human race may or may not be replaced by machines. But not quite yet… if for no other reason than at the moment, it’s just too darned expensive.
[This op-ed was written by an AI model] (just kidding!)
