OpenAI's billion-dollar gamble: Cash consumption surges to $115 billion over the next four years, aiming for $200 billion in revenue by 2030

Global artificial intelligence leader OpenAI is embarking on the most capital-intensive gamble in the history of technology.

According to the latest financial forecasts disclosed to shareholders, OpenAI is facing an unprecedented capital consumption battle. From this year until 2029, OpenAI's cumulative cash consumption is expected to reach as high as $115 billion, a figure that has surged by about $80 billion compared to the company's expectations six months ago.

Adding to the approximately $2 billion consumed over the past two years, its enormous funding needs confirm the judgment of its CEO Sam Altman: OpenAI may be "the most capital-intensive startup in history."

On the other hand, driven by the strong growth of ChatGPT, the company is also more optimistic about its business prospects, raising its revenue forecast for 2030 by about 15%, aiming for the $200 billion mark.

Rapidly Expanding Costs: Where Will the Money Go?

OpenAI's latest cash flow forecast reveals an extremely steep cost curve. The company expects to consume over $8 billion in cash this year, about $1.5 billion higher than the forecast at the beginning of the year; next year, this figure will double to over $17 billion, an increase of a full $10 billion from the original forecast.

Looking ahead, the expected consumption for 2027 and 2028 will reach approximately $35 billion and $45 billion, respectively—where the new forecast for 2028 is more than four times the previous estimate of $11 billion.

The destination of this massive funding can be categorized into four main areas:

1. Building Infrastructure: To reduce reliance on cloud service providers and control costs in the long term, OpenAI plans to invest nearly $100 billion in the latter part of this decade to build its own controlled servers and related facilities. Its CFO Sarah Friar has hinted that in the future, OpenAI may follow in the footsteps of Amazon AWS and offer server leasing services to other AI developers.

2. AI Model Training Costs (Training Compute): This is a key driver of costs that far exceed expectations. The company expects to spend over $9 billion on this item this year, about $2 billion higher than the original forecast; next year, it will reach approximately $19 billion, also exceeding the forecast by more than $2 billion. Projections indicate that training costs will continue to grow steadily until 2030, and may even rise beyond that. As demonstrated by the challenges the company faced in developing GPT-5, AI training is fraught with uncertainty, and failures and retries will directly drive up costs.

3. AI Model Inference Costs (Inference Compute): This refers to the costs of running AI models. From 2025 to 2030, the company expects to spend over $150 billion on this item. Although the cost per run is close to expectations, the cumulative effect of small differences will lead to over $11 billion in additional expenses by 2030.

4. Talent Competition Costs: In the fierce talent competition, OpenAI expects to incur approximately $20 billion in additional stock compensation expenses by 2030 to attract and retain core talent

ChatGPT-Driven Optimistic Revenue Forecast

Despite significant cost pressures, OpenAI's commercialization pace is accelerating, and its revenue outlook is improving accordingly. The company expects total revenue to reach $13 billion this year, which is 3.5 times that of last year. More critically, the revenue forecast for 2030 has been raised to approximately $200 billion.

The core engine driving this optimistic expectation is ChatGPT:

• Strong Performance of Paid Subscription Business: The latest forecast indicates that over the next six years, the additional revenue generated by ChatGPT will exceed the original plan by nearly $70 billion. Specifically, revenue from millions of individual and thousands of enterprise paying users is expected to approach $10 billion this year (which is $2 billion higher than the previous forecast), and by 2030, this annual revenue is expected to be close to $90 billion, a significant increase of about 40% from previous estimates.

• Tapping into the Huge Potential of Free Users: OpenAI has also raised its expectations for monetizing free users, planning to generate approximately $110 billion in revenue from this large user base between 2026 and 2030. Although the monetization model is not yet clear, it may include affiliate fees related to shopping or some form of advertising. The company has previously informed investors that the gross margin for such products is expected to reach 80% to 85%, comparable to that of Meta (Facebook).

Of course, not all business expectations are on the rise. The company has lowered its revenue forecast for API (Application Programming Interface) business over the next five years by $5 billion, while also reducing the revenue forecast for "Agents" (AI products capable of handling multi-step tasks) by about $26 billion.

One possible explanation is that the relevant technology will be more integrated into ChatGPT rather than offered as standalone products. Regardless, the strong growth in ChatGPT's revenue is sufficient to offset these downward adjustments in business expectations.

Capital Frenzy: $500 Billion Valuation and Path to IPO

The astonishing "burn rate" has not deterred capital.

Dozens of large investment institutions, including SoftBank, Thrive Capital, and Dragoneer, are still actively buying OpenAI's stock, pushing its latest valuation to $500 billion—almost double that of six months ago and equivalent to nearly one-fifth of Google's market value, which had a net profit of $100 billion last year. Many investors view OpenAI as a barometer for the proliferation of AI technology, firmly believing it can ultimately achieve significant commercial returns.

To support its massive data center construction plans, going public may be an inevitable choice for OpenAI. Becoming a publicly traded company would allow it to more easily raise funds through debt issuance and even emulate Amazon AWS by renting out servers to other AI developers.

However, this path is fraught with challenges. The company's unique structure of "for-profit operations controlled by a non-profit parent" and the complex legal and contractual relationships with Elon Musk and its largest external shareholder Microsoft (which, under an agreement, is entitled to 20% of OpenAI's revenue) may add uncertainties to its future IPO journey