After analyzing OpenAI’s AI requirements, Barclays concluded that the AI capital expenditure cycle will continue, and a technological breakthrough may trigger a surge in computing power demand in 2027/2028.

After analyzing OpenAI’s AI requirements, Barclays concluded that the AI capital expenditure cycle will continue, and a technological breakthrough may trigger a surge in computing power demand in 2027/2028.

According to the latest Barclays research report, OpenAI's revenue performance has significantly exceeded its internal expectations, indicating that AI demand is rapidly increasing and the large-scale capital expenditure cycle will not end in the short term. As long as OpenAI can maintain its current strong growth momentum, the risk of a bubble bursting in the AI sector will remain at a relatively low level.

Analysis shows that OpenAI’s revenue in 2025 was about 15% higher than its mid-year internal forecast, while the expected revenue in 2027 increased by as much as 50% compared to previous estimates.

For the capital markets, this trend means that internet giants and hyperscale cloud service providers will continue to maintain intense capital investment, and semiconductor demand will remain robust. Barclays expects that by 2028, OpenAI's computing expenditure will reach a peak level of approximately $110 billion, and technological breakthroughs may trigger a new surge in computing power demand at that time.

This Barclays report attempts to answer a core question: How far are we from an AI investment slowdown? The answer: still a long way off.

Revenue dramatically surpasses expectations

Barclays’ report shows that OpenAI's revenue performance continues to exceed internal expectations, with actual revenue in 2025 about 15% higher than the mid-year forecast, and the expected revenue for 2027 raised by 50%.

Specific data shows that OpenAI’s total expected revenue for 2027 has been raised from $60 billion to $90 billion, inference computing costs increased from $21 billion to $30 billion, weekly active users (WAU) grew from 1.4 billion to 1.8 billion, and annual average revenue per paying user (ARPU) rose from $748 to $880.

The company’s CEO, Sam Altman, recently revealed in a public interview that OpenAI is expected to achieve the annual recurring revenue (ARR) target of $100 billion in 2027, a full year ahead of previous forecasts.

AI capital expenditure cycle will continue

Barclays’ latest research report points out that OpenAI’s continued outperformance confirms that the AI capital expenditure cycle will be sustained over the medium to long term.

First, all of OpenAI’s revenue translates into computing power, and business growth directly drives the company’s computing power investment. The four main revenue streams of ChatGPT—paid ChatGPT, free ChatGPT (ad-supported), agents, and API—each require computing resources differently, but all use the same basic computing architecture. OpenAI’s total computing operating expenditure budget between 2024 and 2030 has exceeded $450 billion, and is expected to peak at around $110 billion in 2028.

Second, continuous model iteration keeps raising demand for computing power, forcing computing partners to accelerate infrastructure deployment. OpenAI is steadily advancing research and development of next-generation models such as GPT-6 and Sora 3; each model upgrade brings a significant increase in training and inference costs, thereby continuously driving investment in underlying computing infrastructure.

Barclays’ report notes, OpenAI expects 2027–2028 to be the key implementation window for “recursive self-improvement,” further increasing demand for computing power. This technology, through “drop-in AI researcher,” allows AI to autonomously develop next-generation models (such as GPT-6, Sora, etc.), forming a closed-loop of “AI developing AI.” The company has reserved about $43 billion in additional “Monetizable Compute” specifically for 2028 to support this technological breakthrough.

Meanwhile, OpenAI has signed about $650 billion in computing power leasing contracts with multiple partners, spanning the next decade. Among them, Oracle OCI’s contract has a total value of $300 billion, lasting 5 years from 2027 at an annual average of $60 billion; Microsoft Azure’s total contract value is $250 billion, lasting 7 years from mid-2026 at an annual average of $36 billion. In addition, Google GCP’s contract is worth $40 billion over 7 years, Amazon AWS offers a $38 billion contract over 7 years, and CoreWeave provides a $22.4 billion contract over 5 years.

Third, increasing industry competition has triggered an “arms race.” To respond to OpenAI’s current 6–12 month technological lead, competitors such as Google and Meta have been forced to expand user base and accelerate model iteration simultaneously. From 2024 to 2030, the global AI data center’s total capacity is expected to double from 114.3 GW to 236 GW. OpenAI alone will require its partners (such as Oracle, Microsoft, etc.) to bear over $600 billion in capital expenditure for building compute clusters.

Lastly, the long-term strategic commitment of tech giants further solidifies high investment levels. Tech giant founders are placing greater emphasis on long-term AI competition; for example, Larry Page stated “Better bankrupt than lose,” and even in the face of market volatility, they are willing to continue investing to secure their spot, driving industry-wide capital expenditure to remain high.

Risk Warning and DisclaimerThe market has risks; investment should be made with caution. This article does not constitute personal investment advice, nor does it consider individual users’ specific investment goals, financial situation, or needs. Users should consider whether any opinions, perspectives, or conclusions in this article are suitable for their particular circumstances. Any investment made accordingly is at your own risk.