OpenAI lowers computing power expenditure targets, causing a stir in the AI community. Is this bad news or a misunderstanding?

Hello everyone, here is a major update from OpenAI: they are lowering their total computing power expenditure target to $600 billion.

OpenAI Significantly Lowers Total Computing Power Spending Goal

On February 21, OpenAI informed its investors that the current goal is to reach approximately $600 billion in total computing power expenditure by 2030. Just a few months ago, CEO Sam Altman was still promoting a $1.4 trillion infrastructure commitment.

According to insiders, the AI company now provides a lower figure along with a clearer timeline for planned spending. Previously, there was growing concern that their expansion ambitions were too large and might not match potential revenue.

Sources say OpenAI expects its total revenue to exceed $280 billion by 2030, with contributions from consumer and enterprise businesses nearly split evenly. They indicate that the company’s proposed spending plan is aimed at more directly aligning with its expected revenue growth.

Last year in the second half, OpenAI announced a series of infrastructure deals worth billions of dollars and established partnerships with leading chip manufacturers and cloud service providers.

An insider revealed that OpenAI is finalizing a massive funding round potentially exceeding $100 billion, with about 90% coming from strategic investors. Nvidia is in talks to invest up to $30 billion as part of this round, which could bring the company’s pre-money valuation to $730 billion.

In addition to Nvidia, strategic investors in this round include SoftBank and Amazon.

It is reported that OpenAI achieved $13.1 billion in revenue in 2025, surpassing the $10 billion target. Meanwhile, the company’s capital expenditure was $8 billion, below the expected $9 billion.

Sources say that ChatGPT’s current weekly active users have exceeded 900 million, up from 800 million in October.

Facing competition from rivals like Google and Anthropic, OpenAI issued a “code red” in December last year to focus on improving its chatbot. Insiders note that ChatGPT’s growth slowed slightly in the fall but has now rebounded to record highs in both weekly and daily active users.

Additionally, insiders mention that the company’s coding product Codex has surpassed 1.5 million weekly active users. Codex’s direct competitor is Anthropic’s Claude Code, which has seen widespread adoption over the past year.

How to interpret this?

OpenAI’s significant reduction in its total computing power expenditure target is generally viewed by the market over the weekend as a negative signal for hardware-related investments, given that OpenAI is a leading global AI company and a sharp cut in capital expenditure on computing resources clearly signals something.

However, there is another perspective suggesting that these figures are comparing two different metrics over different timeframes.

Analysts explain that the $1.4 trillion figure Altman previously mentioned refers to the committed infrastructure investments over the next eight years, covering a broader scope of infrastructure-related spending, not the “cumulative capital expenditure by 2030.”

The $600 billion figure, on the other hand, refers specifically to computing power expenditure by 2030. Compared to the $1.4 trillion over eight years, this is a shorter timeframe and a narrower scope (only covering computing power, not all infrastructure commitments).

The market is comparing “a future commitment of $1.4 trillion over 8 years” with “computing power expenditure of $600 billion by 2030,” and concludes that “$800 billion has been cut.” This logic is flawed; it is a mismatch of metrics.

(Article source: China Fund Report)