The wave of artificial intelligence is sweeping the globe, and at the center of the storm, OpenAI is shaking the technology and financial sectors with a series of massive procurement orders like never before.
The latest update is that OpenAI has signed a major agreement to collaborate with Broadcom to deploy 10 gigawatts of customized chips and network equipment over the next four years.
The Financial Times, citing informed sources, reported that this deal could mean that OpenAI may need to spend an additional $350 billion to $500 billion on top of the approximately $1 trillion chip and data center procurement agreements signed in recent months to meet its enormous demand for building AI infrastructure.
This series of actions has once again focused the market's attention on a core question: Where does OpenAI's money come from? How does this AI startup, which is expected to lose about $10 billion this year, support its seemingly endless capital expenditures?
On October 14, former Goldman Sachs banker and Bloomberg columnist Matt Levine vividly pointed out that OpenAI's financing strategy is &#34;world-class financial engineering,&#34; with its essence being &#34;financial time travel.&#34;
Levine explained that traditional financial models are based on clear business plans for factories and products to secure loans, but tech startups represent an extreme version—they rely on a &#34;crazy ambitious vision&#34; and the &#34;ability to make people believe in that vision&#34; to attract investment. Levine likened it to entrepreneurs &#34;putting their hand on the investor's shoulder, pointing into the distance, and softly saying: 'Omniscient (AI) robot',&#34; prompting investors to &#34;have an epiphany.&#34;
For example, after last night's news, Broadcom's stock surged 11%, with investors betting that this alliance will bring the chip manufacturer hundreds of billions of dollars in new revenue. However, the details of how OpenAI will pay for this equipment remain unclear.
<blockquote>
If you owe the bank $100, that's your problem. If you owe Broadcom $500 billion, that's Broadcom's problem. If you owe every major tech company hundreds of billions, that's their problem. They will definitely find a way to solve it! Or you will. The money problem will solve itself.
</blockquote>
Levine's sharp implication is that OpenAI is deeply binding the fate of its suppliers to itself through massive orders, prompting them to seek solutions together. &#34;The money problem will solve itself.&#34;
Additionally, Levine cited Altman's recent disclosure to employees: OpenAI aims to build 250 gigawatts of new computing power by 2033, which would cost over $10 trillion at current standards. Altman has stated that new financing tools need to be created to fund this massive construction, to which Levine commented: &#34;No! Wrong! He has already shared a lot of details about his new financing tools! He shares more every day!&#34;
DA Davidson analyst Gil Luria is skeptical about OpenAI's financial capabilities, stating bluntly that &#34;OpenAI has no ability to make any of these commitments,&#34; and predicting that the company may lose about $10 billion this year. He believes this reflects Silicon Valley's &#34;fake it till you make it&#34; mentality, aimed at getting &#34;many people to invest a lot of energy into OpenAI.&#34;
<h2>The &#34;Financial Alchemy&#34; Behind the Trillion-Dollar Orders</h2>
OpenAI's computing power procurement is not a simple cash transaction, but a carefully designed &#34;financial alchemy.&#34; According to calculations from the &#34;Little Bear Runs Fast&#34; public account, by 2029, OpenAI's chip procurement expenditures with the three major chip manufacturers—Nvidia, AMD, and Broadcom—will total as much as $285 billion.
Among them, the collaboration with Nvidia is valued at about $100 billion, the agreement with AMD involves 6GW of computing power, estimated at about $75 billion (approximately 3 million MI450 chips), while the 10GW AI server collaboration with Broadcom corresponds to an investment of about $110 billion (approximately 5.5 million custom TPU chips).
Wall Street Insight reported that regarding these massive procurements, questions about where the money comes from have been analyzed by institutions like Goldman Sachs, which pointed out that OpenAI mainly employs two innovative models.
One is the &#34;Equity for Procurement&#34; model, exemplified by the agreement with AMD. On the surface, OpenAI plans to procure $90 billion worth of AMD Instinct series GPUs, but the core of the agreement is that AMD issues warrants to OpenAI. If AMD's stock price rises due to OpenAI's large-scale adoption in the future, the potential equity value held by OpenAI will be close to the hardware procurement cost, allowing it to almost &#34;obtain&#34; this batch of computing power for free. This essentially transforms hardware sales into equity allocation, directly linking AMD's long-term valuation with OpenAI's infrastructure growth, forming a symbiotic relationship.
The other is the &#34;Recurring Revenue&#34; model, represented by the collaboration with Nvidia. Nvidia plans to invest up to $100 billion in OpenAI, which in turn can be used by OpenAI to purchase Nvidia's chips. Goldman Sachs defines this as &#34;recurring revenue,&#34; meaning that the supplier's capital injection ultimately flows back in the form of revenue.
<h2>Huge Funding Gap and Future Uncertainty</h2>
Despite the clever design of these financial instruments, OpenAI's real funding gap remains enormous.
Goldman Sachs analyzed OpenAI's financial situation from two perspectives. From the operating cash flow standpoint, OpenAI's annual operating infrastructure cost in 2026 is about $35 billion, with contributions from its own revenue, supplier financing, and external equity/debt financing being relatively balanced.
However, once future significant capital commitments, such as building data centers in collaboration with Nvidia and investments in the &#34;Stargate&#34; project, are taken into account, OpenAI's total funding requirement for the year 2026 will soar to approximately $114 billion
In this case, its capital structure will be completely unbalanced, with reliance on external equity and debt financing skyrocketing to an overwhelming 75%, while the contribution from its own revenue is diluted to only 17%.
As an OpenAI executive said, &#34;Most people would rather have a smaller piece of a larger cake.&#34; OpenAI's gamble is attempting to leverage its disruptive AI vision to attract the entire tech ecosystem's investment. Over the next four years, its chip spending of up to $285 billion far exceeds its current annual revenue of about $13 billion.
This indicates that unless OpenAI can find new business avenues for aggressive monetization, or major chip manufacturers invest through equity stakes, its ongoing, large-scale financing from external capital will be key to the success of its ambitious &#34;Stargate&#34; plan

OpenAI

AVGO

If you owe the bank $100, that's your problem. If you owe Broadcom $500 billion, that's Broadcom's problem. If you owe every major tech company hundreds of billions, that's their problem. They will definitely find a way to solve it! Or you will. The money problem will resolve itself

- OpenAI is making significant waves in the tech and finance sectors with massive procurement orders, including a recent agreement with Broadcom for customized chips and network equipment.  
- Analysts suggest that OpenAI's financing strategy involves innovative models like "equity for procurement" and "recurring revenue," but concerns remain about its ability to meet its projected $1 trillion in chip and data center expenses.  
- The company's financial structure may become heavily reliant on external financing, raising questions about its long-term sustainability and the feasibility of its ambitious AI infrastructure plans.

Wallstreetcn

Broadcom

1 trillion orders plus 350-500 billion USD! OpenAI "can't stop," "when you owe every company hundreds of billions, the money problem will solve itself"?