AlphaWire

newswire

600 billion US dollars → 750 billion US dollars Open AI is exposed to significantly increase its computing power expenditure forecast

2026-07-22·newswire-us-stock-183919
600 billion US dollars → 750 billion US dollars Open AI is exposed to significantly increase its computing power expenditure forecast.

The latest news on Wednesday night showed that artificial intelligence giant OpenAI’s computing power spending ambitions have expanded. OpenAI has raised its forecast for computing power spending by 2030 to $750 billion from about $600 billion earlier this year, according to media reports citing people familiar with the matter.

The increase reflects the company's new agreements with cloud computing providers and its efforts to strengthen the construction of its own data centers as the "Stargate" data center project is not progressing.

As the latest manifestation of this dynamic, OpenAI has just announced that it will invest $20 billion in Effingham County, Georgia, to launch a data center called "Project Camellia." Sachin Katti, vice president of computing strategy at OpenAI, revealed that this project is the first time OpenAI has led the design and development of a data center.

The company has completed the land acquisition and storage work for the proposed project and signed a contract with Georgia Power Company to obtain 3.2 gigawatts of power supply between 2028 and 2032.

Carty declined to say how much the company paid the utility to reserve power, but said it was a "substantial fee" that was enough to give the utility the confidence to significantly expand capacity.

As envisioned early last year, OpenAI attempted to form a joint venture with SoftBank, Oracle and Middle Eastern investors to jointly promote the US$500 billion "Stargate" data center plan. People familiar with the matter said that as the project has struggled so far, OpenAI is restarting internal plans.

The latest report states that OpenAI recently invited Brent Mayo, who once built a data center for Musk, as the person in charge of data center construction and delivery. Mayo was one of the core architects of the xAI data center construction plan and played a key role in the construction of the company's first Colossus data center.

Mayo's focus is to ensure that the data center built by OpenAI's cloud computing partners is completed on time, and will also participate in the just-announced Georgia data center project. In addition to self-built projects, the bulk of OpenAI's expenditure is still purchasing computing power from external cloud service providers.

OpenAI has previously signed a contract with Oracle to acquire a total of 6 gigawatts of data center capacity, most of which is still in the development stage.

At the end of February this year, OpenAI expanded its multi-year cloud computing agreement with Amazon AWS to 8 years and US$138 billion, including the use of 2 gigawatts of computing power from Amazon's self-developed artificial intelligence chip Trainium.

In addition, OpenAI committed an additional $250 billion in cloud computing spending to Microsoft's Azure cloud platform last year, but did not provide a specific timetable for the spending. It is worth mentioning that OpenAI’s computing power expenditure figures caused market concerns last year.

In a live broadcast last October, OpenAI CEO Sam Altman said that the company expected to invest US$1.4 trillion in AI-related infrastructure between 2025 and 2033. As if in response to market uneasiness, OpenAI CFO Fryer later changed his tone. She privately told investors that OpenAI plans to spend $600 billion on computing power by 2030.

However, the two figures cover different time frames, and the latter also limits expenditures to computing power. According to reports earlier this year, after tripling revenue to $13 billion in 2025, OpenAI predicts revenue will reach $30 billion in 2026 and looks forward to revenue in 2030 of more than $280 billion.

Revenue growth also means an increase in computing power/cash consumption. The company expects cash consumption to reach US$25 billion this year and US$57 billion next year, a total of approximately US$30 billion more than previously forecast.

#Stocks #Tesla #Microsoft #Amazon #AI

Full text

600 billion US dollars → 750 billion US dollars Open AI is exposed to significantly increase its computing power expenditure forecast

The latest news on Wednesday night showed that artificial intelligence giant OpenAI’s computing power spending ambitions have expanded. OpenAI has raised its forecast for computing power spending by 2030 to $750 billion from about $600 billion earlier this year, according to media reports citing people familiar with the matter. The increase reflects the company's new agreements with cloud computing providers and its efforts to strengthen the construction of its own data centers as the "Stargate" data center project is not progressing.

The latest news on Wednesday night showed that artificial intelligence giant OpenAI’s computing power spending ambitions have expanded. OpenAI has raised its forecast for computing power spending by 2030 to $750 billion from about $600 billion earlier this year, according to media reports citing people familiar with the matter. The increase reflects the company's new agreements with cloud computing providers and its efforts to strengthen the construction of its own data centers as the "Stargate" data center project is not progressing. As the latest manifestation of this dynamic, OpenAI has just announced that it will invest $20 billion in Effingham County, Georgia, to launch a data center called "Project Camellia." Sachin Katti, vice president of computing strategy at OpenAI, revealed that this project is the first time OpenAI has led the design and development of a data center. The company has completed the land acquisition and storage work for the proposed project and signed a contract with Georgia Power Company to obtain 3.2 gigawatts of power supply between 2028 and 2032. Carty declined to say how much the company paid the utility to reserve power, but said it was a "substantial fee" that was enough to give the utility the confidence to significantly expand capacity. As envisioned early last year, OpenAI attempted to form a joint venture with SoftBank, Oracle and Middle Eastern investors to jointly promote the US$500 billion "Stargate" data center plan. People familiar with the matter said that as the project has struggled so far, OpenAI is restarting internal plans. The latest report states that OpenAI recently invited Brent Mayo, who once built a data center for Musk, as the person in charge of data center construction and delivery. Mayo was one of the core architects of the xAI data center construction plan and played a key role in the construction of the company's first Colossus data center. Mayo's focus is to ensure that the data center built by OpenAI's cloud computing partners is completed on time, and will also participate in the just-announced Georgia data center project. In addition to self-built projects, the bulk of OpenAI's expenditure is still purchasing computing power from external cloud service providers. OpenAI has previously signed a contract with Oracle to acquire a total of 6 gigawatts of data center capacity, most of which is still in the development stage. At the end of February this year, OpenAI expanded its multi-year cloud computing agreement with Amazon AWS to 8 years and US$138 billion, including the use of 2 gigawatts of computing power from Amazon's self-developed artificial intelligence chip Trainium. In addition, OpenAI committed an additional $250 billion in cloud computing spending to Microsoft's Azure cloud platform last year, but did not provide a specific timetable for the spending. It is worth mentioning that OpenAI’s computing power expenditure figures caused market concerns last year. In a live broadcast last October, OpenAI CEO Sam Altman said that the company expected to invest US$1.4 trillion in AI-related infrastructure between 2025 and 2033. As if in response to market uneasiness, OpenAI CFO Fryer later changed his tone. She privately told investors that OpenAI plans to spend $600 billion on computing power by 2030. However, the two figures cover different time frames, and the latter also limits expenditures to computing power. According to reports earlier this year, after tripling revenue to $13 billion in 2025, OpenAI predicts revenue will reach $30 billion in 2026 and looks forward to revenue in 2030 of more than $280 billion. Revenue growth also means an increase in computing power/cash consumption. The company expects cash consumption to reach US$25 billion this year and US$57 billion next year, a total of approximately US$30 billion more than previously forecast.

← Back to archive