The Inside Story of the “Trillion-Dollar AI Loop”: The Strategies of Contemporary Tech Titans such as Altman, Jensen Huang, Satya Nadella, Masayoshi Son, and Others

The Inside Story of the “Trillion-Dollar AI Loop”: The Strategies of Contemporary Tech Titans such as Altman, Jensen Huang, Satya Nadella, Masayoshi Son, and Others

Sam Altman, CEO of OpenAI, is binding the world's largest tech giants tightly to his AI juggernaut through a series of carefully orchestrated deals. By skillfully leveraging both the ambitions and anxieties of these giants, Altman has not only secured seemingly limitless computing power for OpenAI, but has also built a colossal business ecosystem that is “too big to fail,” linking the fate of the entire tech industry to this still-unprofitable startup.

In recent months, Altman has led a storm of deals sweeping through Silicon Valley. In January this year, SoftBank’s Masayoshi Son and Altman announced a $500 billion “Stargate” project at the White House. Nvidia then proposed a similar project to OpenAI, which was finalized last month with a $100 billion partnership. After Oracle signed a $300 billion contract, its stock surged nearly 40%, briefly making chairman Larry Ellison the world’s richest man; AMD and Broadcom also announced deals worth tens of billions of dollars.

These partnership announcements caused dramatic reactions in the capital market. On just the first trading day after the news broke, the related companies saw a combined market value increase of $630 billion, repeatedly pushing the US tech indices to new record highs.

The core of this high-stakes bet on computing power is a “finite game” initiated by Altman. By exploiting his competitors’ “fear of missing out” (FOMO), he has managed to entice one company after another to join the game. However, this financial one-upmanship has also heightened market concerns: Is the excitement over AI morphing into a giant bubble reliant on just one company—or even one person's vision?

As OpenAI racks up computing bills worth hundreds of billions of dollars, while annual revenue stands at just $13 billion, this massive financial leverage is testing the nerves of all participants. Some deals even show “circular” characteristics, with partners funding OpenAI’s purchase of their own products. Underlying this is Altman’s faith in the exponential logic that computing power is the future; he has told employees the company's long-term goal is to build 250 GW of computing power by 2033, a gargantuan plan comparable to the electricity usage of the entire nation of Germany.

Masayoshi Son’s Ambition: Lighting the FOMO Fuse

Before igniting this computing arms race, Altman faced setbacks. Last year, he proposed to Microsoft CEO Satya Nadella that Microsoft invest at least $100 billion to build a new “Stargate” data center for OpenAI, only to be rejected. Similarly, his $7 trillion global chip foundry plan was not supported by TSMC, whose CEO C.C. Wei bluntly stated during the 2024 shareholders meeting that Altman was “too aggressive, unbelievable.”

After repeated setbacks, Altman turned to SoftBank’s CEO Masayoshi Son. Known for his risk-taking investment style and eager for a major AI bet after losses like WeWork, Son was receptive. At a Tokyo event, Son recalled that when Altman emphasized “without more computing power, none of the grand AI vision is possible,” he began to think, “If more is better, then we should do even more.”

To push the deal forward, Son invited Altman to his mansion on the outskirts of Tokyo for several days last November. In this estate, dubbed “Versailles” by colleagues, the two reached an agreement. Shortly after, they jointly announced at the White House a $500 billion “Stargate” investment project, led by SoftBank to finance and build data centers for OpenAI.

Although media reports say the joint venture hit snags over site selection, it sparked market frenzy. SoftBank's stock rose 11% after the announcement. More importantly, it triggered the industry’s FOMO, paving the way for Altman’s subsequent maneuvers.

Microsoft’s “Forbearance”: From Bystander Back to the Table

As OpenAI’s exclusive cloud partner for six years, Microsoft was once its staunchest supporter. Yet, faced with Altman’s unending hunger for computing power, doubts began to surface within Microsoft. Asked about the $500 billion Stargate project at Davos in January, Nadella laughed, “I only know that my $80 billion is fine,” referring to Microsoft's annual data center budget.

Nadella’s cautious approach was not unique. In a podcast interview, he said:

“At some point, supply and demand must match. If you numb yourself with excitement on the supply side instead of truly understanding how to translate that into customer value, you could go totally off track.”

Subsequently, TD Cowen reported that Microsoft had canceled certain US data center leases, interpreted as a reduction in support for OpenAI’s single workload.

Microsoft granted OpenAI permission to seek other cloud providers while it shifted to attract more diverse clients. Right then, Oracle seized the opportunity, signing a $300 billion contract with OpenAI. The news sent Oracle’s stock up nearly 40%, pushing its market cap toward $1 trillion.

This win led some Microsoft executives to privately criticize Oracle’s ability to execute such a massive project. Nonetheless, the market’s feverish response couldn't be ignored. A week after Oracle's surge, Nadella announced a “world’s most powerful AI data center” in Wisconsin, specifying that part would be for OpenAI model training. Sources say Microsoft is now discussing providing OpenAI with additional computing power.

According to previous coverage, OpenAI’s almost reckless ambition convinced Microsoft it was better to let competitors take on astronomical financial risk than to do so itself. And even after relaxing its exclusivity, Microsoft remains the only cloud service provider offering OpenAI supercomputers for model training—the most critical element in AI development. Microsoft also continues to take a 20% cut of OpenAI’s soaring revenue.

Jensen Huang’s “Aggression”: From Supplier to Deep Integration

For Nvidia CEO Jensen Huang, seeing Altman and Son announce a historic AI infrastructure project at the White House was surely provoking. According to media sources, as the leading chip supplier powering OpenAI for the past decade, Huang wanted to be the one standing beside Altman.

Soon after, Nvidia privately pitched a similar project to OpenAI, aiming to replace SoftBank and help fund new data center construction. The two sides eventually agreed last month to a $100 billion deal which Huang called “the largest computing project in history.”

Yet, negotiations were not smooth. Reportedly, talks stalled in summer; the turning point came on June 27, when tech outlet The Information reported OpenAI had begun renting Google’s TPU chips for ChatGPT. This alarmed Nvidia internally. Sources say Huang immediately called Altman to confirm, declaring willingness to resume negotiations.

The final deal went far beyond initial plans. Nvidia agreed to lease OpenAI up to 5 million chips—costing $350 billion at current prices—and may invest as much as $100 billion to help OpenAI pay.

More notably, media cite sources saying an unreported arrangement involves Nvidia considering guarantees for part of OpenAI’s own data center construction loans. This means if OpenAI cannot repay, the chip giant could face debts of tens of billions, deeply binding the two companies’ futures as never before.

Competitors Joining In: AMD and Broadcom’s Big Bets

Altman’s “FOMO” strategy is infectious. AMD CEO Lisa Su also wanted a “shining AI moment” for her firm. At AMD’s “Advancing AI” conference in June, Altman appeared as the closing guest, and Su posed a pointed, symbolic question:

“May I call you an AI idol?”

AMD was working hard to catch up to Nvidia, making OpenAI’s endorsement crucial. The two firms closely collaborated on AMD’s latest MI450 chip and began talks on a deal worth tens of billions. To win this opportunity, Su was willing to offer up to 10% of AMD’s future stock as a bonus—essentially a huge subsidy for OpenAI. After the deal’s announcement on October 6, AMD shares jumped 24% in a single day, one of its biggest daily gains ever.

The race is still adding participants. Shortly after Nvidia announced their deal, Broadcom revealed an agreement with OpenAI to co-develop new chips and computing systems providing 10 GW of computing power—matching the scale of Nvidia’s deal.

Every time a new giant joins, others are forced to accelerate, stacking even more bets on OpenAI’s future and further reinforcing the “trillion-dollar closed loop” Altman is constructing.

Risk Warning and Disclaimer ClauseThe market involves risk; investment should be cautious. This article does not constitute personal investment advice, nor does it take into account the specific investment goals, financial situation, or needs of any individual user. Users should consider whether any opinions, views, or conclusions contained herein are suitable to their particular circumstances. Investments are made at your own risk.