Three years after ChatGPT shook the world, the generative AI industry is entering a more uncomfortable phase: the technology keeps getting more impressive, but the economics are getting harder to ignore.
New models can reason, generate realistic video, and automate everyday work — yet the infrastructure required to run them costs a fortune.
And according to one analyst, OpenAI may be approaching a financial cliff.
A company burning cash faster than it can earn it
The core argument is simple: building and operating cutting-edge AI is brutally expensive. Sebastian Mallaby, a senior fellow at the Council on Foreign Relations, argues that OpenAI could face a major cash crunch in the next 18 months if it can’t keep raising capital at the same pace as its spending.
In his view, the AI race has become an infrastructure war — massive data centers, specialized chips, and relentless compute bills — and investors may not be willing to fund that indefinitely, no matter how large the potential payoff of AI becomes.
The “scaling” paradox: smarter models, exponentially higher costs
One of the biggest traps in modern AI is that progress often becomes more expensive over time. As models grow more complex, they demand more compute, more training runs, and more energy — and the costs can rise faster than revenue.
Analysts point to a familiar paradox: the more powerful the product becomes, the harder it is to make the business sustainable at consumer-friendly prices.
Reports and estimates circulating in the industry suggest 2025 was a painful year financially, with losses widely believed to be in the multi-billion-dollar range. Even after major fundraising, the “war chest” can evaporate quickly when each new generation of models raises the operating baseline.
The real problem: monetization in a world full of free alternatives
OpenAI’s challenge is not just cost — it’s pricing power. Unlike Big Tech rivals that can fund AI with cash from massive legacy businesses (ads, search, app ecosystems), OpenAI relies heavily on subscriptions and external capital.
But converting users into paying customers is harder when strong alternatives are increasingly available at low cost or even free.
There’s also a switching problem — but in the opposite direction. Right now, most people can jump from one model to another with little friction. Until AI products become deeply integrated into workflows and personal context, users can churn easily, limiting predictable revenue.
“WeWork on steroids”? A harsh comparison starts circulating
That combination — huge spending, unclear profitability, heavy reliance on funding — has led some observers to use brutal comparisons. In the broader debate, OpenAI has sometimes been framed as a company with historic impact but an extremely difficult path to long-term independence.
Whether that analogy is fair or exaggerated, it reflects a growing sentiment: the AI revolution can be real while the financial structure of a specific AI company can still be fragile.
If OpenAI stumbles, ChatGPT likely survives — but independence might not
Even critics who worry about OpenAI’s runway don’t necessarily predict a sudden “death” of ChatGPT. The more likely scenario, many analysts argue, would be a shift in ownership or control — a deeper absorption into a larger platform company, or an acquisition-style outcome where the product continues but the independent structure fades.
In that case, today’s users might barely notice the transition day-to-day. The real change would happen behind the scenes: governance, strategy, priorities, and who ultimately benefits from the next decade of AI.
The bigger picture: AI isn’t going away, even if one company loses the race
The debate ultimately separates two ideas. One is the future of AI itself — which appears increasingly durable as businesses find real productivity gains. The other is whether OpenAI, specifically, can remain independent while sustaining the cost curve of frontier models.
If the next wave of “agents” and automation tools doesn’t translate into massive, dependable revenue, the most plausible future may be an AI landscape that continues to expand — but with OpenAI operating as a division inside a much larger corporate empire.
