Escalating infrastructure costs force OpenAI to consider going public Is tech giant on path towards bankruptcy?

OpenAI is preparing for a potential initial public offering (IPO) later this year, a move that critics say could mark a turning point for a project originally founded to serve the “common good,” shifting it further toward market-driven priorities.

Some argue that the company’s early vision — promoting open research and broad access to artificial intelligence — has collided with a fundamental constraint: the immense and growing cost of developing generative AI systems, as highlighted in an analysis published by The Conversation.

Founded in 2015 by Sam Altman and Elon Musk, OpenAI began as a non-profit amid rising concerns about the risks of AI. It pledged to build technology “beneficial to humanity” and prevent its concentration in the hands of a few dominant firms.

That ambition set it apart from tech giants such as Google, Microsoft, Meta and Amazon, whose business models rely on proprietary systems and monetisation at scale.

Mounting costs of generative AI

Unlike traditional software — where marginal costs tend toward zero over time — generative AI requires vast, ongoing infrastructure. Every interaction consumes computing power, energy and specialised hardware.

A single ChatGPT query typically costs between $0.01 and $0.10, while generating a high-definition image can cost up to $0.20. At the scale of billions of daily queries in 2026, those costs quickly become enormous.

Much of this expense stems from the need for advanced graphics processing units (GPUs), largely supplied by Nvidia. These chips can cost tens of thousands of dollars each, or several dollars per hour via cloud services.

OpenAI relies on tens of thousands of such processors running continuously in large data centres, supported heavily by Microsoft, which has invested more than $13 billion and provides cloud infrastructure through Azure. Analysts estimate that total investment requirements across the sector could reach hundreds of billions of dollars by the end of the decade.

By the late 2010s, it had become clear that a purely non-profit structure could not sustain this level of capital intensity. In 2019, OpenAI adopted a hybrid model, allowing it to raise funds while maintaining oversight through a non-profit foundation.

That compromise was intended to balance financial realities with its founding mission. But the launch of ChatGPT in late 2022 dramatically accelerated the company’s growth — reaching 100 million users in just two months and surpassing 900 million weekly users by early 2026.

The surge in demand brought a shift toward a diversified business model, including subscription services, enterprise licensing, and partnerships.

Every generated dollar accompanied by losses

OpenAI now generates income from multiple streams: paid subscriptions, enterprise integrations, and strategic partnerships — particularly with Microsoft, which embeds its technology into products under the Copilot brand.

These combined revenues reportedly reached around $1 billion per month in 2025. Yet despite this rapid growth, the company remains structurally unprofitable.

In the first half of 2025, OpenAI generated approximately $4.3 billion in revenue but recorded losses estimated between $7 billion and $13 billion — more than $2 billion per month. Cumulative losses could exceed $140 billion between 2024 and 2029.

The gap reflects the nature of its business, with the article noting that every user interaction incurs a cost, while research and development spending remains exceptionally high. OpenAI is believed to have invested nearly $16 billion in R&D in 2025 alone.

Labour costs are also significant, with top AI specialists earning base salaries between $250,000 and $700,000 — and total compensation often exceeding $1 million.

The company also faces intense competition from emerging AI models such as Gemini, DeepSeek and Mistral AI, adding further pressure to maintain technological leadership.

According to the outlet's analysis, OpenAI’s hybrid structure — in which a non-profit foundation controls a for-profit “public benefit corporation” — has created internal tensions.

"The hybridisation of a public interest mission with private financing mechanisms resulted in a complex structure. A non-profit foundation controls a for-profit “public benefit corporation”, which is funded by investors and tasked with raising capital and developing activities – all while theoretically remaining subordinate to the foundation’s public interest mission. This construction, designed to avoid purely financial logic, quickly fuelled tensions between different stakeholders," the article argues.

IPO lifeline comes with fundamental trade-offs

An IPO is being positioned as a way to address these financial pressures, with the potential to raise between $50 billion and $100 billion by selling a minority stake.

However, the article points out that going public would subject OpenAI to stricter profitability expectations and transparency requirements — constraints that may conflict with the experimental and long-term nature of AI development.

The company’s continued reliance on partners such as Microsoft and Nvidia also raises questions about its strategic independence.

Crucially, analysts caution that an IPO alone may not resolve the underlying issues. Without significant changes to its business model, it could merely delay deeper financial challenges.

As The Conversation concludes, the economics of generative AI remain fundamentally unstable as they stand right now.

By Nazrin Sadigova