The largest cloud computing transaction in history, Stargate, originated from an inconspicuous Linkedin private message, which was a gamble made by Oracle co-founder Larry Ellison, but now the enthusiasm on Wall Street is cooling down.

In the spring of 2024, a senior executive from OpenAI sent a private message to the sales manager of Oracle on LinkedIn. At that time, OpenAI was frantically searching for computing power, while Oracle had just vacated a large data center project in Texas due to the suspension of Musk's xAI project.

In January of this year, both sides and SoftBank jointly announced the construction of an AI data center project called "Stargate" with US President Trump at the White House. This transaction caused Oracle's market value to soar by $250 billion, and Ellison briefly became the world's richest man.

But the frenzy is fading, and Oracle, which a few months ago symbolized investors' confidence in AI, is now becoming a symbol of panic.

From its peak in September to the release of its financial report in December, Oracle's stock price has fallen by over one-third. To fulfill the contract, the company has experienced negative free cash flow for the first time since 1992. Wall Street traders began to buy Oracle's credit default swap, which was used to short the real estate foam during the subprime crisis.


(Oracle's stock price drops sharply from its peak)

The birth of the largest cloud computing contract in history

This unprecedented transaction originated from a LinkedIn private message sent by OpenAI Infrastructure Director Peter Hoeschele to the Oracle sales team. According to Hoeschele's public statement in October, this news facilitated dialogue between the two sides.

On December 12th, Bloomberg Businessweek reported that Oracle was drawing blueprints for a large data center in western Texas, originally planned to collaborate with Musk's xAI, and the building would even be designed in an X-shape to match the company name.

Musk has a close relationship with Ellison, who has served on Tesla's board for many years. But the fickle Musk believed he could build it faster and gave up the project.

At this moment, OpenAI's call for help just arrived.

This AI company has been desperately searching for computing power since ChatGPT emerged in 2022. It has signed numerous agreements to lease data centers, purchase specialized AI chips GPU, and even attempted to develop its own chips, but none of these can meet its huge demand for processing power.

OpenAI has committed to spending approximately $300 billion to lease servers from Oracle, and the two parties have also discussed the possibility of further expanding their cooperation. To provide this computing power, Oracle needs to build about five data center complexes, each of which will rank among the top in the world, requiring millions of chips and 4.5 gigawatts of electricity in total, which exceeds the household electricity consumption of the entire city of Chicago.

But according to Wall Street reports, in early trading on Friday, the media reported that Oracle had postponed the completion of some data centers developed for OpenAI from 2027 to 2028, mainly due to labor and material shortages.

Oracle subsequently denied the news, and an Oracle spokesperson stated in a statement that the company is confident in fulfilling its obligations and future expansion plans. The statement states:

The site selection and delivery schedule are closely coordinated and determined with OpenAI. All stations required by the contract have not experienced any delays, and all progress is progressing according to plan.

Financial Pressure and foam Warning

Oracle's aggressive strategy is bringing enormous financial pressure.

To build these data centers, the company requires billions of dollars in upfront investment. Against the backdrop of soaring material and labor costs, Microsoft spent nearly $40 million to purchase concrete for just one data center in Wisconsin within six months of 2024.

Energy is also a major issue. Many utility companies are facing tight production capacity, and data centers need to wait for a long time to connect to the power grid. According to reports, Oracle has adopted novel but expensive methods to provide faster electricity, such as using gas generators to power the entire site. In a data center park in rural Shackleton County, Texas, this will cost over $1 billion annually.

Through off balance sheet financing, Oracle has excluded many AI related development costs from its own books.

Real estate developer Vantage Data Centers is borrowing $38 billion for two data center campuses and leasing them to Oracle. Other facilities also adopt similar arrangements, with Oracle typically signing a 15 year lease. According to analysts compiled by Bloomberg, Oracle's large-scale infrastructure investment is expected to yield returns, but only if it can withstand a free cash flow loss of approximately $70 billion over the next four to five years.

Nowadays, Oracle is just one of many companies building data centers for OpenAI. KeyBanc Capital Markets analyst Jackson Ader said:

OpenAI's commitment is inversely correlated with Oracle's stock price. As OpenAI spending increases, the value of each commitment is decreasing.

On December 10th, Oracle's financial report showed that spending growth far exceeded analysts' expectations and will continue, causing Oracle's stock price to continue to decline. Another signal of AI nervousness is the disappointing order data from chip manufacturer Broadcom AI, with its stock price plummeting by 11% on Friday.


Microsoft's cautious comparison

Microsoft has been a partner of OpenAI since 2016, and the model that ChatGPT relies on is built on the Microsoft Azure cloud computing platform. But Microsoft is gradually concerned about taking up too much data center capacity on OpenAI, and even has to limit the processing power of some of its largest customers to meet OpenAI's needs.

Microsoft CEO Satya Nadella summarized these concerns in a podcast interview in October. He said:

Sometimes Sam says, 'Hey, can you build a dedicated Dogawa data center for me to use for training?'? From OpenAI's perspective, it makes sense, but it doesn't make sense for Microsoft Azure.

In 2024, Microsoft allows OpenAI to search for other cloud providers as long as Microsoft reserves the right of first refusal. In October, Microsoft even gave up this right.

According to the analysis, despite the huge investment of hundreds of billions of dollars in data centers, the economic impact of AI is still unclear, and warnings of foam are emerging one after another.

Oracle Corporation may be particularly vulnerable. No tech giant has invested as heavily in artificial intelligence as it has. To achieve profitability in the Stargate project, Oracle's partner OpenAI must maintain the astonishing growth momentum of the past few years.

That's also why Microsoft, Oracle's old rival, hesitated when OpenAI proposed a similar deal after careful consideration.

The transaction structure of the Stargate project gives OpenAI the opportunity to rededicate or exit in about five years, which is about ten years before many of Oracle's leases expire.

If OpenAI's revenue stops growing rapidly, the startup may neither need so much computing power nor have the ability to pay. Oracle may ultimately only have some extremely expensive and illiquid assets left.

81 year old Ellison leads the tech boom

Silicon Valley culture typically values youthfulness and boldness, but Ellison, who is 81 years old, still plays a central role in another tech boom.

According to documents reviewed by Bloomberg in November, he actually increased the proportion of supervised employees in this year's restructuring. The most crucial thing is that he controls the company's expenses, and now the top financial executive reports directly to him, bypassing the two CEOs.

Although Ellison has made relatively few public appearances, he continues to be willing to openly discuss topics that are far from Oracle's core business. In his keynote speech at AI World in October, he spent 10 minutes elaborating on his views on agriculture. But most of his 90 minute speech was focused on AI.

He pointed out that two years ago he publicly asked if AI would become the "most important technology in human history". He no longer doubts now. He said:

This is the largest and fastest-growing business in human history, bigger than railways and bigger than the Industrial Revolution.

Oracle's bet only needs to be right to yield returns. But at a time when AI expectations are falling, this veteran company is facing a question: is it over committed to a future that has not yet been proven profitable? The market's answer is changing.

D. Analyst Gil Luria from Davidson&Co. stated:

I see three possible scenarios. Either Oracle will lower its forecast next year but still be able to salvage some of its business, or OpenAI will collapse and its contract will fall through, or OpenAI will achieve super intelligence at a cost of $1.4 trillion. We don't have to work anymore, and Oracle is fine.

Analysis suggests that the current hype cycle is partly based on some grand predictions, such as computers being able to cure various cancers, eliminate a large amount of human labor, and even introduce post life forms. But the date for achieving these goals is getting farther and farther away, perhaps returning to science fiction.

The lack of transformative progress makes the logic of investing in AI on a massive scale vulnerable. Even popular consumer products like OpenAI, which have 800 million active users, are not enough to justify the book numbers.

Once the hundreds of billions of dollars spent on the data center are put aside, foam warnings abound, and no technology giant has made such an all-out bet on AI as Oracle. Whether this company can win the last laugh will become a key indicator of whether the entire industry is prosperous or a foam.

The largest cloud computing transaction in history, Stargate, originated from an inconspicuous Linkedin private message, which was a gamble made by Oracle co-founder Larry Ellison, but now the enthusiasm on Wall Street is cooling down.

In the spring of 2024, a senior executive from OpenAI sent a private message to the sales manager of Oracle on LinkedIn. At that time, OpenAI was frantically searching for computing power, while Oracle had just vacated a large data center project in Texas due to the suspension of Musk's xAI project.

In January of this year, both sides and SoftBank jointly announced the construction of an AI data center project called "Stargate" with US President Trump at the White House. This transaction caused Oracle's market value to soar by $250 billion, and Ellison briefly became the world's richest man.

But the frenzy is fading, and Oracle, which a few months ago symbolized investors' confidence in AI, is now becoming a symbol of panic.

From its peak in September to the release of its financial report in December, Oracle's stock price has fallen by over one-third. To fulfill the contract, the company has experienced negative free cash flow for the first time since 1992. Wall Street traders began to buy Oracle's credit default swap, which was used to short the real estate foam during the subprime crisis.


(Oracle's stock price drops sharply from its peak)

The birth of the largest cloud computing contract in history

This unprecedented transaction originated from a LinkedIn private message sent by OpenAI Infrastructure Director Peter Hoeschele to the Oracle sales team. According to Hoeschele's public statement in October, this news facilitated dialogue between the two sides.

On December 12th, Bloomberg Businessweek reported that Oracle was drawing blueprints for a large data center in western Texas, originally planned to collaborate with Musk's xAI, and the building would even be designed in an X-shape to match the company name.

Musk has a close relationship with Ellison, who has served on Tesla's board for many years. But the fickle Musk believed he could build it faster and gave up the project.

At this moment, OpenAI's call for help just arrived.

This AI company has been desperately searching for computing power since ChatGPT emerged in 2022. It has signed numerous agreements to lease data centers, purchase specialized AI chips GPU, and even attempted to develop its own chips, but none of these can meet its huge demand for processing power.

OpenAI has committed to spending approximately $300 billion to lease servers from Oracle, and the two parties have also discussed the possibility of further expanding their cooperation. To provide this computing power, Oracle needs to build about five data center complexes, each of which will rank among the top in the world, requiring millions of chips and 4.5 gigawatts of electricity in total, which exceeds the household electricity consumption of the entire city of Chicago.

But according to Wall Street reports, in early trading on Friday, the media reported that Oracle had postponed the completion of some data centers developed for OpenAI from 2027 to 2028, mainly due to labor and material shortages.

Oracle subsequently denied the news, and an Oracle spokesperson stated in a statement that the company is confident in fulfilling its obligations and future expansion plans. The statement states:

The site selection and delivery schedule are closely coordinated and determined with OpenAI. All stations required by the contract have not experienced any delays, and all progress is progressing according to plan.

Financial Pressure and foam Warning

Oracle's aggressive strategy is bringing enormous financial pressure.

To build these data centers, the company requires billions of dollars in upfront investment. Against the backdrop of soaring material and labor costs, Microsoft spent nearly $40 million to purchase concrete for just one data center in Wisconsin within six months of 2024.

Energy is also a major issue. Many utility companies are facing tight production capacity, and data centers need to wait for a long time to connect to the power grid. According to reports, Oracle has adopted novel but expensive methods to provide faster electricity, such as using gas generators to power the entire site. In a data center park in rural Shackleton County, Texas, this will cost over $1 billion annually.

Through off balance sheet financing, Oracle has excluded many AI related development costs from its own books.

Real estate developer Vantage Data Centers is borrowing $38 billion for two data center campuses and leasing them to Oracle. Other facilities also adopt similar arrangements, with Oracle typically signing a 15 year lease. According to analysts compiled by Bloomberg, Oracle's large-scale infrastructure investment is expected to yield returns, but only if it can withstand a free cash flow loss of approximately $70 billion over the next four to five years.

Nowadays, Oracle is just one of many companies building data centers for OpenAI. KeyBanc Capital Markets analyst Jackson Ader said:

OpenAI's commitment is inversely correlated with Oracle's stock price. As OpenAI spending increases, the value of each commitment is decreasing.

On December 10th, Oracle's financial report showed that spending growth far exceeded analysts' expectations and will continue, causing Oracle's stock price to continue to decline. Another signal of AI nervousness is the disappointing order data from chip manufacturer Broadcom AI, with its stock price plummeting by 11% on Friday.


Microsoft's cautious comparison

Microsoft has been a partner of OpenAI since 2016, and the model that ChatGPT relies on is built on the Microsoft Azure cloud computing platform. But Microsoft is gradually concerned about taking up too much data center capacity on OpenAI, and even has to limit the processing power of some of its largest customers to meet OpenAI's needs.

Microsoft CEO Satya Nadella summarized these concerns in a podcast interview in October. He said:

Sometimes Sam says, 'Hey, can you build a dedicated Dogawa data center for me to use for training?'? From OpenAI's perspective, it makes sense, but it doesn't make sense for Microsoft Azure.

In 2024, Microsoft allows OpenAI to search for other cloud providers as long as Microsoft reserves the right of first refusal. In October, Microsoft even gave up this right.

According to the analysis, despite the huge investment of hundreds of billions of dollars in data centers, the economic impact of AI is still unclear, and warnings of foam are emerging one after another.

Oracle Corporation may be particularly vulnerable. No tech giant has invested as heavily in artificial intelligence as it has. To achieve profitability in the Stargate project, Oracle's partner OpenAI must maintain the astonishing growth momentum of the past few years.

That's also why Microsoft, Oracle's old rival, hesitated when OpenAI proposed a similar deal after careful consideration.

The transaction structure of the Stargate project gives OpenAI the opportunity to rededicate or exit in about five years, which is about ten years before many of Oracle's leases expire.

If OpenAI's revenue stops growing rapidly, the startup may neither need so much computing power nor have the ability to pay. Oracle may ultimately only have some extremely expensive and illiquid assets left.

81 year old Ellison leads the tech boom

Silicon Valley culture typically values youthfulness and boldness, but Ellison, who is 81 years old, still plays a central role in another tech boom.

According to documents reviewed by Bloomberg in November, he actually increased the proportion of supervised employees in this year's restructuring. The most crucial thing is that he controls the company's expenses, and now the top financial executive reports directly to him, bypassing the two CEOs.

Although Ellison has made relatively few public appearances, he continues to be willing to openly discuss topics that are far from Oracle's core business. In his keynote speech at AI World in October, he spent 10 minutes elaborating on his views on agriculture. But most of his 90 minute speech was focused on AI.

He pointed out that two years ago he publicly asked if AI would become the "most important technology in human history". He no longer doubts now. He said:

This is the largest and fastest-growing business in human history, bigger than railways and bigger than the Industrial Revolution.

Oracle's bet only needs to be right to yield returns. But at a time when AI expectations are falling, this veteran company is facing a question: is it over committed to a future that has not yet been proven profitable? The market's answer is changing.

D. Analyst Gil Luria from Davidson&Co. stated:

I see three possible scenarios. Either Oracle will lower its forecast next year but still be able to salvage some of its business, or OpenAI will collapse and its contract will fall through, or OpenAI will achieve super intelligence at a cost of $1.4 trillion. We don't have to work anymore, and Oracle is fine.

Analysis suggests that the current hype cycle is partly based on some grand predictions, such as computers being able to cure various cancers, eliminate a large amount of human labor, and even introduce post life forms. But the date for achieving these goals is getting farther and farther away, perhaps returning to science fiction.

The lack of transformative progress makes the logic of investing in AI on a massive scale vulnerable. Even popular consumer products like OpenAI, which have 800 million active users, are not enough to justify the book numbers.

Once the hundreds of billions of dollars spent on the data center are put aside, foam warnings abound, and no technology giant has made such an all-out bet on AI as Oracle. Whether this company can win the last laugh will become a key indicator of whether the entire industry is prosperous or a foam.