"There's just too much money!" Nvidia has nowhere to invest. Instead of repurchasing, it would be better to choose "let AI close the loop".

Release Time：2025-09-24

Nvidia, which holds a huge amount of cash, is attempting a brand-new way of capital utilization, investing tens of billions of dollars in key customers, partners, and even competitors to create a self-sustaining "AI closed-loop ecosystem", thereby locking in the long-term demand for future chips.

The most notable move is that NVIDIA recently announced plans to invest up to 100 billion US dollars in OpenAI to support its large-scale expansion of data centers. This move is not only a strong support for OpenAI, but also reflects the core idea of NVIDIA CEO Jensen Huang - to consolidate the entire AI industry chain with a powerful balance sheet and market confidence.

As soon as the news came out, the market immediately responded enthusiastically: Nvidia's market value soared by nearly 160 billion US dollars on the same day. This investment not only alleviated the outside world's concerns about OpenAI's financial situation, but also allowed people to see the so-called "circularity" business model: Nvidia invests money in start-ups, and these companies use the money back to buy NVIDIA's chips.

This model is not limited to OpenAI. From cloud service provider CoreWeave to rival Intel and then to Musk's xAI, NVIDIA's investment portfolio has extended to every corner of the AI field.

The trouble of having too much money

For most companies, having abundant cash is a good thing, but for NVIDIA, this has become a special problem. FactSet data shows that in the past four quarters, Nvidia has generated $72 billion in free cash flow and is expected to approach $100 billion in the current fiscal year. Apart from Apple, no other tech giant can match it.

The question is, how to spend it?

Mergers and acquisitions are almost impossible. As early as when Nvidia acquired Mellanox in 2020, it encountered regulatory resistance. In the current geopolitical environment, large-scale chip mergers and acquisitions are almost impossible to pass easily. Moreover, Huang Rengxun himself prefers a flat organizational structure and is more inclined towards small-scale technological integration.

The repurchase and research and development cannot be digested. Nvidia spent nearly 50 billion US dollars on share buybacks in the past four quarters and added an additional 60 billion US dollars in plans, but still couldn't stop the pace of cash inflows. Although R&D investment has doubled in two years, its proportion of revenue has dropped to 9%, far lower than the previous 22%.

Strategic investment: customers, partners, and even rivals

Under the constraints of traditional paths, NVIDIA has chosen to use strategic investment to build an AI ecosystem with an internal circulation.

The commitment of a $100 billion investment in OpenAI is at the core of it. NewStreet Research analyzed that for every $10 billion Nvidia invests in OpenAI, the latter will spend $35 billion to buy back NVIDIA chips. Although such transactions may lower NVIDIA's per-chip profit margin, they can ensure continuous demand and also provide a "lifeline" for cash-strapped AI companies.

Other investment cases include:

CoreWeave: Nvidia holds a 7% stake in it. Recently, the two sides signed a $6.3 billion agreement, under which NVIDIA committed to repurchasing its unused cloud capacity and further binding key customers.

Intel: NVIDIA unexpectedly invested 5 billion US dollars in its old rival to jointly develop new products, enabling Gpus to integrate more smoothly with Intel cpus and taking the opportunity to expand in the PC market.

xAI: Musk's xAI has long listed NVIDIA as a strategic investor. In March this year, NVIDIA, along with partners such as xAI, joined a multi-billion-dollar AI data center and energy construction plan.

Win-win chess game: Stabilize demand and lower financing costs

This investment strategy is essentially a win-win game for NVIDIA - it not only stabilizes its own demand but also helps its partners reduce financing costs.

For NVIDIA, the most direct benefit is to secure a large and stable chip order in the coming years and avoid fluctuations in demand. For a partner like OpenAI, NVIDIA's endorsement is a huge credit guarantee.

It should be noted that OpenAI is expected to have accumulated losses of up to 44 billion US dollars by 2029, and it will also have to bear huge expenses such as chip procurement and data center leasing. Before Nvidia's involvement, the interest rate for their financing was as high as 15%, reflecting the market's distrust of their business model. For projects backed by Microsoft, the interest rate is only 6% to 9%. Even Moody's downgraded its outlook from "stable" to "negative" because Oracle is overly dependent on OpenAI.

Nvidia's entry has changed everything. With the "almost unbreakable confidence" of the capital market in it, NVIDIA can directly support its AI infrastructure projects with its balance sheet or even additional share issuance, at a cost far lower than external financing. Industry insiders predict that this will significantly reduce OpenAI's credit risk, enabling it to obtain loans at lower interest rates.

Risk Warning and Disclaimer

The market involves risks. Please invest with caution. This article does not constitute personal investment advice and has not taken into account the individual user's specific investment objectives, financial situation or needs. Users should consider whether any opinions, views or conclusions in this article are suitable for their specific circumstances. Any investment made based on this is at your own risk.