Amid the Jensen Huang extramarital affair scandal, Nvidia and Silicon Valley's AI capital network
In an open letter to employees, Jensen Huang admitted to an extramarital affair; Nvidia's board of directors launched an independent investigation, and the company's stock price in overnight trading at one point fell to $177.90, down nearly 5% from last Friday's close. The article uses this to sort through the funding and compute loops that Nvidia has formed in recent years with OpenAI, Oracle, CoreWeave, xAI and others through investments, chip procurement and financing guarantees, and lists follow-up signals to watch such as financial reports, GPU deliveries, financing terms and data center approvals.
Why read this
AI score 73
The article sorts through the investment and procurement chains linking Nvidia with OpenAI, Oracle, CoreWeave and other companies, presenting the concrete structure of the AI compute funding loop.
Topics
Body · AI translation
[[**Jensen Huang admits to an extramarital affair and faces a board investigation; Nvidia shares fell nearly 5% at one point; through investments, procurement, and guarantees, he is deeply tied to OpenAI and others, and AI circular investment faces pressure from regulation and valuation reassessment.**
**Key points**
**1. The personal scandal has triggered an investigation and market pressure**
In an open letter to employees, Jensen Huang admitted to an extramarital affair; the board launched an independent investigation and is examining reports that he may have been a target of foreign intelligence operations; Nvidia fell to $177.90 at one point in overnight trading, and Polymarket predicts a 14.5% probability that he will resign before the end of the year.
**2. Nvidia has transformed from a chip supplier into an AI capital and computing power hub**
Since 2023, it has invested heavily in AI startups that use Nvidia GPUs. In 2024, it invested about $100 million in OpenAI, and later planned to invest up to $100 billion in tranches, while OpenAI used the funds to purchase Nvidia chips, sparking controversy over "circular investment."
**3. OpenAI, Oracle, and CoreWeave, among others, form a closed loop of funding and computing power**
OpenAI signed a $300 billion cloud computing power contract with Oracle, and Oracle plans to spend about $40 billion to purchase Nvidia chips to build data centers; Nvidia took a stake in CoreWeave and signed computing power orders, and CoreWeave in turn leases computing power to OpenAI.
**4. The AI industry faces multiple constraints from regulation, site selection, and financing**
After the Democrats won both chambers, AI regulation has tightened, and the American public opposes data centers; if signals such as Nvidia's financial reports, GPU deliveries and leasing revenue, financing conditions, data center approvals, and electricity prices deteriorate, weakening demand and a break in the funding chain could trigger an industry shakeout. Original 2026-11-09 20:19 Jensen Huang's love-struck mind: which Silicon Valley big shots did it hurt? Huang Tianyuan Quick overview
In the 33 years he has run Nvidia, Jensen Huang has always been one of Silicon Valley's most personally charismatic tech leaders.
This leather-jacket man loves his wife, loves his work, and loves fried bean sauce noodles. Jensen Huang's personal aura has given Nvidia a huge trust premium.
But the more perfect the image, the harder the fall when it collapses from the altar.
In the early hours of this morning, Jensen Huang admitted to an extramarital affair in an open letter to employees. Nvidia's board immediately announced the launch of an independent investigation to examine reports that Jensen Huang may have been a target of foreign intelligence operations. The prediction market Polymarket has already opened a market; as of Huxiu's publication time, the probability that Jensen Huang will resign before the end of the year is 14.5%. Nvidia's stock price also fell to $177.90 at one point in U.S. overnight trading, down nearly 5% at one point from Friday's Nasdaq close.
But the personal scandal cannot explain all of Nvidia's pressure; the AI industry is going through a sensitive moment. In early November, the U.S. Democratic Party, which has consistently emphasized strengthening AI regulation, won control of both the Senate and the House. At the same time, people in various U.S. states have erupted in strong opposition to the construction of data centers.
Jensen Huang is not only selling shovels; he has also begun putting up money to help customers buy shovels. It invests broadly in AI giants including OpenAI and Anthropic, and also sells chips to them; it invests in cloud service providers such as CoreWeave, and also buys computing power from them; it has even begun providing financing guarantees for large data center projects to help customers raise the funds needed to purchase computing power.
This personal scandal has once again focused outside attention on Jensen Huang. But more worthy of study than his personal fate is the vast relationship network he has woven for Nvidia over the past few years.
Nvidia has woven Silicon Valley into a net
At the end of 2022, ChatGPT went viral. The following year, Nvidia accelerated investment while selling GPUs. In 2023 alone, it participated in 38 rounds of financing.
The companies in this wave of investment have two characteristics: one is that they are basically startups using Nvidia GPUs, and the other is that they are AI startups. Runway, which does video generation, Databricks, which does data platforms, and CoreWeave, which does GPU cloud, all received Nvidia's money that year. It was also from this time that many companies began to rely on Nvidia's GPUs to develop and run models, and Nvidia's influence began to matter.
The core of the net: Nvidia and OpenAI
By October 2024, Nvidia had also appeared on OpenAI's list of investors. According to The Wall Street Journal, Nvidia put in about $100 million in that round.
A year later, the two companies began deeper cooperation. OpenAI plans to deploy at least 10 gigawatts of Nvidia systems, and Nvidia plans to invest in tranches as computing power comes online, up to $100 billion. But OpenAI immediately used the money to buy Nvidia chips.
This deal immediately caused an uproar, and Wall Street began discussing "circular investment" among several major companies.
But this was only the beginning. In September 2025, according to The Wall Street Journal, OpenAI signed a cloud computing power contract worth about $300 billion with Oracle. A few months earlier, Oracle planned to spend about $40 billion to purchase Nvidia chips for a Texas data center that would provide computing power to OpenAI. OpenAI is also looking for other companies to buy chips from. In October of the same year, it announced a 6-gigawatt deployment agreement with AMD and received warrants to subscribe to up to 160 million AMD shares.
By 2026, Nvidia had invested $30 billion in OpenAI, and the two sides simultaneously expanded their cooperation on training and inference computing power. By August, Nvidia had also guaranteed the Ohio data center lease for an OpenAI affiliate. The campus is being developed by SB Energy, a SoftBank subsidiary, and under the agreement will deploy Nvidia systems. The guarantee takes effect in stages as the project is delivered, with a cap of $105 billion. If the tenant triggers the agreed default conditions and subletting or sale still cannot cover the shortfall, Nvidia may need to make up the difference to the landlord, and OpenAI has committed to repaying the amounts Nvidia actually pays.
OpenAI orders computing power from Oracle; to build data centers for OpenAI, Oracle in turn plans to buy chips from Nvidia. And Nvidia itself is also an investor in OpenAI.
This kind of money loop has long exceeded a trillion dollars, and their entanglement is growing deeper.
CoreWeave later also became a particularly critical node in it
CoreWeave is a cloud service provider specializing in GPU computing power. It installs Nvidia's GPUs in data centers and then rents the computing power to customers such as OpenAI and Meta for training and running AI models.
In March 2025, Nvidia took a 7% stake during CoreWeave's IPO process, helping to push it public. In January of this year, it spent another $2 billion to increase its stake, at which point it held about 11.5% of the Class A shares. The two also signed a computing power order initially valued at $6.3 billion. After the agreed computing power is delivered, if CoreWeave fails to rent it all out, Nvidia will take over the remaining portion.
On the other side, OpenAI is both a customer and a shareholder of CoreWeave. Several computing power contracts signed last year totaled up to about $22.4 billion. When CoreWeave's listing was completed, it also issued OpenAI shares worth $350 million at the offering price. This year, Meta also approached CoreWeave to rent computing power, signing a contract worth about $21 billion.
Nvidia's support for Musk
Nvidia and Musk also have deep ties.
In January of this year, Nvidia contributed to xAI's $20 billion financing. According to CNBC, Nvidia put in $10 billion of it. In February, xAI was merged into SpaceX. By the end of June, Nvidia disclosed that it held about 123 million SpaceX Class A shares.
By October, SpaceX had obtained about $40 billion in equipment finance lease commitments. Once the closing conditions are met, a financing entity managed by Apollo will buy Nvidia's next-generation GPUs and then lease them to SpaceX. Musk can use these chips to run Grok and can also rent the computing power to others. In May of this year, Anthropic, the maker of Claude, rented all the computing power of one of SpaceX's data centers. SpaceX's third-quarter report said that a single computing power customer contributed 34.5% of the company's total revenue for the quarter.
Who else is in this web?
Several other threads also intersect.
As of the end of September this year, Microsoft held about 24% of OpenAI's equity on a converted basis. It and Nvidia also separately committed to invest up to $5 billion and $10 billion in Anthropic, with some of the funds already entering Anthropic's Series G. Anthropic also buys computing power from Microsoft; the latest prospectus discloses that the revised minimum spending commitment under the Azure contract is $44.9 billion. Google is also a shareholder in and computing power supplier to Anthropic, and its parent company Alphabet also holds SpaceX stock.
As for Meta, in October this year Nvidia also announced it would contribute $3 billion to participate in its equity offering. Meta has also purchased chips from Nvidia, but it is also using Google's TPU chips.
Where things go from here
How will Nvidia's next quarterly earnings perform? It mainly depends on the growth of the data center business in the report, revenue guidance for the next quarter, and management's comments on customer purchasing and delivery pace.
How much of the newly purchased GPUs can turn into revenue? Whether high-end GPUs can be delivered on schedule, and how many customers are willing to rent the installed cards and at what price.
Will the next round of money still be easy to raise? Watch how much model companies and computing power service providers can still raise, and whether financing conditions have tightened.
The data centers announced for construction—can they start on schedule? Watch whether approval, electricity pricing, and tax policies change at the U.S. federal level and in key states such as Texas. Even after permits are obtained, when electricity can be connected and how much it will cost after connection will also affect whether projects can proceed as originally planned.
What everyone used to worry about may simply have been whether sustained high capital expenditure could translate into stable cash flow. But now there are more and more variables: Can new data centers be approved on schedule? Will electricity bills and tax burdens raise operating costs? Will Democrats impose more and more restrictions on AI? Will chip exports and antitrust reviews become increasingly strict?
One project hitting a snag can still be sidestepped by investors. But when risk appears simultaneously on the demand, price, and financing fronts—once downstream demand substantially weakens, computing power rental prices keep falling, projects already under construction are long delayed in delivery, or the funding chain breaks—at that point, industry consolidation, or rather the bursting of the bubble, may truly begin.
I remember very clearly that I once talked with an embodied intelligence investor about the "left hand transferring to the right hand" problem in the industry. He gave a scoffing laugh and said this also happens in Silicon Valley, citing Nvidia as an example. He asked me back: "Would you consider this a bubble? Do you think they will burst?"
At the time I had no answer; now it seems a little clearer.
Article title: Whose Silicon Valley big shots did Jensen Huang's lovestruck mind hurt?
AI Industry Signal Channel: Frontier Technology. This content is published with the author's authorization. The views represent only the author and do not represent Huxiu's position. If you have objections or complaints about this manuscript, please contact tougao@huxiu.com. Everyone who is changing and wants to change the world is on the Huxiu APP[ ]]]