Is Nvidia financing its own growth?

Why is Nvidia investing billions in companies that buy its chips?

Nvidia is now helping finance the very companies buying its AI chips, and the amount of money involved has grown dramatically.

According to Nvidia's latest quarterly SEC filing, the company purchased $42.4 billion in equity securities during the first half of fiscal 2027, compared with $1.25 billion a year earlier. That is roughly a 34-fold increase. The figure includes investments beyond direct customer financing, but it shows how aggressively Nvidia has expanded its financial activities. A chip maker backing its own customers raises the question of who is funding whom, and we dig into it in our report on AI chips.

The company invested $2 billion in CoreWeave in January 2026, announced another $2 billion investment in Nebius in March, and joined OpenAI's $110 billion financing round alongside Amazon and SoftBank. Nvidia also signed a major agreement to support the financing of OpenAI infrastructure developed by SB Energy, with potential obligations capped at $105 billion.

These transactions connect Nvidia to several stages of the AI economy. It supplies GPUs, invests in cloud operators, helps AI laboratories raise money and supports the financing of the data centers where its equipment will operate.

The harder question is how much of its future growth depends on that financial support.

What does it actually mean for Nvidia to finance its own growth?

Nvidia finances its own growth when its money or financial guarantees help customers purchase and operate Nvidia-powered computing infrastructure.

The simplest example is an equity investment. Nvidia provides cash to an AI company, which can then use its broader financing resources to purchase computing capacity. OpenAI's February 2026 funding announcement illustrates this relationship: Nvidia was among the investors in a $110 billion round while also expanding its computing partnership with OpenAI.

There are more complicated arrangements. Nvidia can agree to purchase unused cloud capacity, helping infrastructure operators secure financing for GPU installations. It can also guarantee certain financial obligations if a customer cannot fulfill a contract. A supplier that also buys back unused capacity is an unusual setup; we compare which business models are winning in our AI chips report.

A $2 billion equity investment involves an identifiable transfer of capital. A $105 billion conditional guarantee could create a much larger liability, but only if the specified conditions occur. An infrastructure-financing partnership involving outside investors may require relatively little Nvidia capital.

Adding all those headline values together would give a pretty misleading picture of how much Nvidia is actually spending.

How much money is Nvidia actually investing in AI companies?

Nvidia's investment activity has exploded, with net equity-security purchases reaching approximately $35.2 billion during the first half of fiscal 2027.

Its SEC filing for the period ended July 26, 2026, shows $42.4 billion in equity-security purchases and $7.24 billion in proceeds from sales. A year earlier, net purchases were approximately $1.18 billion.

That represents an almost 30-fold increase in net equity-security purchases in twelve months.

The comparison with Nvidia's cash generation is particularly revealing. Equity-security purchases were equivalent to 57% of operating cash flow during the latest six-month period, compared with just 2.9% a year earlier.

Nvidia's balance sheet also showed $42.8 billion in marketable equity securities and $51.2 billion in non-marketable securities. Together, those holdings approached $94 billion, although the amounts include changes in investment valuations and cannot be treated as cumulative customer financing.

Nvidia financial measure	First half FY2026	First half FY2027
Operating cash flow	$42.8B	$74.4B
Equity-security purchases	$1.25B	$42.4B
Equity-security sales	$0.07B	$7.24B
Net equity-security purchases	$1.18B	$35.16B
Equity purchases / operating cash flow	2.9%	57.0%
Property and intangible-asset spending	$3.12B	$4.43B
Net cash used in investing activities	$12.34B	$35.12B

Is Nvidia really investing $100 billion in OpenAI?

Nvidia's original $100 billion OpenAI investment proposal has effectively been overtaken by a different financing arrangement, although the relationship has become financially significant in another way.

In September 2025, Nvidia and OpenAI announced a letter of intent under which Nvidia could invest up to $100 billion as OpenAI deployed at least 10 gigawatts of Nvidia systems.

The proposal attracted attention because the same company supplying the computing hardware would potentially provide much of the money needed to buy it.

By early 2026, however, the proposed structure had changed. OpenAI announced a $110 billion financing round on February 27, including $30 billion from Nvidia, $30 billion from SoftBank and $50 billion from Amazon. Jensen Huang subsequently said the original $100 billion Nvidia investment was unlikely to proceed as initially envisioned.

Then Nvidia took on a different and potentially larger financial exposure.

In August 2026, an SEC filing disclosed that Nvidia had entered into residual-value guarantees connected to approximately 4.25 gigawatts of infrastructure at SB Energy's Portsmouth site, where OpenAI would be the tenant. Nvidia's aggregate payment obligation was capped at $105 billion under the initial agreements.

That $105 billion is a ceiling on contingent obligations, rather than money Nvidia has already paid. The guarantees generally become effective when the relevant leases commence, with service expected to begin in 2028. Payments would depend on specified defaults and the value recovered through replacement leasing or asset sales.

Could Nvidia really lose $105 billion if OpenAI fails?

Nvidia has taken on a potentially huge OpenAI-related obligation, but a $105 billion guarantee does not mean the company would automatically lose that amount if OpenAI encountered financial problems.

The August 17, 2026 SEC filing describes residual-value guarantees involving leases at SB Energy's Portsmouth project.

If OpenAI defaults because of insolvency or fails to make required lease payments, Nvidia could have to cover the difference between a guaranteed minimum value and the amount recovered through a replacement lease or sale.

That recovery mechanism matters. A completed data center containing valuable computing infrastructure may still attract another tenant. The financial loss would depend on the remaining asset value, the contractual guarantees and conditions at the time of default.

Nevertheless, the maximum commitment is extraordinary. At $105 billion, the guarantee ceiling exceeded Nvidia's $74.4 billion in operating cash flow during the first half of fiscal 2027 and was almost twice its $56.6 billion of cash, cash equivalents and marketable debt securities at the end of that period.

Those comparisons describe the size of the exposure, rather than expected losses. The agreements also cover future infrastructure deployments, so the maximum amount would not ordinarily become payable immediately.

Is Nvidia's CoreWeave deal a clear example of circular financing?

CoreWeave is one of the clearest examples of Nvidia acting as an equipment supplier, investor and computing customer within the same business relationship.

In September 2025, CoreWeave disclosed a $6.3 billion agreement under which Nvidia would purchase certain computing capacity that CoreWeave could not sell to other customers, subject to contractual conditions. The arrangement runs through April 2032.

In January 2026, Nvidia bought another $2 billion of CoreWeave shares at $87.20 each. The companies also announced plans to accelerate deployment of more than five gigawatts of AI infrastructure by 2030. Prices like $87.20 a share raise the question of what these companies are really worth, which we look at in our report covering AI chips.

The two transactions complement each other. Nvidia supplies capital through the equity investment and helps support CoreWeave's future revenue through the capacity agreement.

CoreWeave still has a substantial business beyond Nvidia. Its annual revenue increased from $229 million in 2023 to $1.9 billion in 2024 and $5.1 billion in 2025. It also secured approximately $18 billion in debt and equity financing during 2025.

Yet the financial pressure has become difficult to ignore. In October 2026, CoreWeave's 9.625% notes due in 2032 traded at yields reaching approximately 15% to 16%, according to market reporting. Those yields reflected investor concern about its debt burden and future financing needs.

Nvidia CEO Jensen Huang said CoreWeave did not need an additional financial backstop. CoreWeave also pointed to a revenue backlog that had grown to approximately $104 billion.

The circular financing concern is legitimate. Nvidia is helping create demand for its own products by financially supporting a company that buys them. But CoreWeave's expanding outside revenue means the relationship cannot simply be dismissed as artificial sales.

Is Nvidia financing an entire generation of AI cloud companies?

Nvidia is now supporting several major AI cloud providers, effectively helping build a wider customer base for its chips.

CoreWeave received $2 billion in January 2026. Nebius announced another $2 billion Nvidia investment in March, alongside plans to deploy more than five gigawatts of Nvidia systems by the end of 2030.

Nscale provides a third example. In September 2026, the company announced $3.36 billion in pre-IPO convertible financing involving investors such as Nvidia, Apollo and Third Point. Its regulatory filing also described an anticipated $1 billion Nvidia subscription, with closing expected later in the year and subject to the relevant transaction terms.

Nscale's ambitions are enormous relative to its existing business. Its prospectus reported only $33 million in revenue for 2025, while describing more than $103 billion in active and contracted total contract value as of August 2026.

CoreWeave, Nebius and Nscale have different business models and financing profiles. But together, they show that Nvidia is helping multiple companies build the capacity to purchase and operate its hardware.

Does Nvidia's $500 billion financing plan mean it will spend $500 billion?

Nvidia's $500 billion infrastructure-financing initiative relies on outside investors, so the headline figure substantially overstates the company's direct spending commitment.

On August 10, 2026, Nvidia announced partnerships with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to establish financing platforms intended to mobilize more than $500 billion of third-party capital over time.

The idea is to make Nvidia-powered infrastructure easier for institutional investors to finance.

Instead of requiring every cloud provider to purchase equipment entirely through its own balance sheet, financing vehicles can help fund computing facilities and recover their investments through long-term customer payments.

There are precedents in aircraft leasing, telecommunications infrastructure and other industries where expensive equipment generates revenue over several years.

Nvidia benefits because more available financing can support more infrastructure deployments. Its financial partners potentially benefit from a new category of income-producing assets.

The $500 billion remains a financing target. It is neither a completed investment total nor a guarantee that all the planned computing capacity will be built.

Is Nvidia's chip business growing fast enough to justify all this financing?

Nvidia's core business is growing at an extraordinary rate, giving the company much more room to support customers than most hardware suppliers would have.

Nvidia generated $215.9 billion in revenue during fiscal 2026, up 65% from $130.5 billion a year earlier. Revenue up 65% in a year makes the gap hard to ignore; we line up who leads and who is slipping in our report on who's ahead in AI chips.

Its second quarter of fiscal 2027 was even stronger. Revenue reached $96.2 billion, increasing 106% year over year. Data Center revenue reached $89.0 billion, up 117%, and the company reported a 75% gross margin.

The scale puts individual investments into perspective. Nvidia's combined $4 billion investments in CoreWeave and Nebius represented approximately 4.2% of just one quarter of Nvidia revenue.

That comparison does not capture OpenAI financing, infrastructure guarantees or indirect customer support. But it does show that Nvidia's current revenue base dwarfs individual investments in AI cloud providers.

The latest SEC filing also reported $74.4 billion in operating cash flow over six months, up approximately 74% year over year.

Nvidia can clearly afford to make investments that would overwhelm smaller companies. Whether all those investments will earn adequate returns is a separate question.

Are Nvidia's profits turning into cash or just rising investment values?

Nvidia's semiconductor business is producing enormous amounts of cash, although gains on its investments have also become a significant part of reported earnings.

During the first half of fiscal 2027, Nvidia recorded approximately $118 billion in net income and $74.4 billion in operating cash flow.

The cash-flow statement included a $23.7 billion adjustment for net gains on equity securities. Those gains increased accounting profits without generating an equivalent amount of operating cash.

Working capital also absorbed substantial cash. Accounts receivable contributed a $24.6 billion outflow, while inventories contributed another $10.2 billion.

These movements help explain why operating cash flow was considerably lower than net income.

Still, $74.4 billion in six-month operating cash flow is an exceptionally strong result, particularly considering that the equivalent figure was $42.8 billion a year earlier.

Investment gains should be separated from Nvidia's operating profits when evaluating the chip business. Rising valuations of Nvidia-backed AI companies can improve reported earnings, but the value of those holdings may fall if market expectations change.

The operating cash flow is more reassuring. Nvidia is collecting substantial cash from its commercial activities even as it expands its investment portfolio.

Are Nvidia's customers taking longer to pay for their chips?

Nvidia is allowing some large customers more time to pay, and its receivables have increased sharply enough to deserve attention.

According to its latest quarterly SEC filing, Nvidia's accounts receivable rose from $38.5 billion in January to $63.1 billion six months later.

That is approximately 64% growth, compared with substantial but slower revenue growth over the same six-month period.

Nvidia attributed some of the pressure on operating cash flow to extended payment terms on large multi-quarter agreements with certain investment-grade customers.

These arrangements effectively provide trade credit. A customer receives equipment under agreed commercial terms while paying later than it otherwise would.

Large, financially strong technology companies commonly negotiate payment arrangements for complicated infrastructure deployments. Extended terms alone do not establish that customers are struggling financially.

But Nvidia's receivables are now enormous in absolute terms. The $24.6 billion cash-flow impact from receivables during the latest six-month period compares with $4.7 billion a year earlier.

The important question going forward is whether collections continue matching contractual expectations. Rising overdue balances, credit losses or repeated extensions would make the financing concern substantially more serious.

How much financial risk is Nvidia taking, and can it afford it?

Nvidia can currently support substantial investments, but its growing financial commitments expose it to risks well beyond ordinary chip sales.

The latest quarterly SEC filing reported $3.3 billion in equity-method investments in infrastructure financiers. For investments classified as variable interest entities, Nvidia disclosed maximum loss exposure of $4.7 billion, including carrying values and future committed amounts. The separate SB Energy arrangement added the much larger $105 billion contingent guarantee ceiling.

At July 26, 2026, Nvidia held $56.6 billion in cash, cash equivalents and marketable debt securities, alongside $42.8 billion in marketable equity securities. We measure how far the challengers trail a company with this much cash in our report on AI chips.

During the first six months of fiscal 2027, it generated $74.4 billion in operating cash flow and spent approximately $4.4 billion on property, equipment and intangible assets.

That leaves roughly $70 billion after those expenditures, before investments, financing activities and other cash uses.

Nvidia also purchased $42.4 billion in equity securities, repurchased approximately $39 billion of its own shares on a cash-flow basis and paid $6.3 billion in dividends. Its cash-flow statement reported approximately $24.9 billion in net debt issuance proceeds.

Nvidia capital allocation	First half FY2027
Operating cash flow	$74.4B
Property and intangible-asset spending	$4.4B
Cash flow after those expenditures	$70.0B
Equity-security purchases	$42.4B
Equity-security sale proceeds	$7.2B
Net equity-security purchases	$35.2B
Share repurchases, cash-flow basis	Approximately $39B
Dividends paid	$6.3B
Net debt issuance proceeds	$24.9B
Cash and marketable debt securities at period-end	$56.6B

Nvidia is financially strong enough to absorb substantial setbacks. But $105 billion in potential guarantees and tens of billions in equity investments should not be treated as ordinary business development expenses. The company's exposure to AI infrastructure financing is getting too large to ignore.

Are Nvidia's biggest AI customers all depending on the same money?

Nvidia's customer network is more concentrated than the growing number of cloud providers might suggest.

Its latest quarterly SEC filing disclosed that five direct customers accounted for 22%, 14%, 13%, 11% and 10% of accounts receivable. Combined, those five customers represented 70% of the balance.

Those percentages measure receivables rather than revenue, but they show that a relatively small group of customers accounts for much of Nvidia's outstanding payments.

There is another layer of concentration beneath the direct customers.

A cloud company might buy Nvidia GPUs and rent them to OpenAI, Anthropic or Microsoft. Several cloud companies may therefore appear as separate Nvidia customers while relying on overlapping end users.

CoreWeave's customer mix illustrates the issue. Its largest customer's share of backlog declined from approximately 85% at the beginning of 2025 to below 35% by year-end. That was genuine diversification, but large AI customers still play a major role in its business.

Nscale's September 2026 prospectus reported that its largest customer represented 52% of revenue during the first half of 2026. Its future contracted revenue was also heavily linked to major agreements with Microsoft and Anthropic.

We cannot derive the exact share of Nvidia revenue ultimately supported by each AI laboratory from these disclosures.

Still, counting every cloud provider as an independent source of demand exaggerates how diversified the market has become. A financing problem affecting one major AI laboratory could spread across several seemingly separate Nvidia customers.

Are OpenAI and Anthropic making enough money to support all this AI infrastructure?

OpenAI and Anthropic have created substantial commercial demand, but their future computing commitments are growing so quickly that infrastructure financing remains a serious question.

OpenAI's February 2026 financing announcement included $110 billion in new investment from Nvidia, Amazon and SoftBank. The company also described plans involving three gigawatts of dedicated inference capacity and two gigawatts of training capacity on Nvidia's Vera Rubin systems.

Anthropic has made similarly ambitious commitments. In August 2026, it signed agreements with Nscale covering up to approximately $44.6 billion in computing services.

Nscale's September prospectus disclosed that it had not yet obtained binding financing commitments for the infrastructure needed to fulfill those Anthropic agreements.

More recent reporting has also highlighted the gap between revenue from operating infrastructure and the capital required to construct future capacity. Nscale reportedly generated approximately $141 million in revenue during the first half of 2026 while recording a net loss of around $1 billion.

The problem is partly one of timing. AI laboratories may agree to purchase computing capacity years before the facilities are operational. Infrastructure providers then need to raise money for construction and equipment well before collecting most of their contracted revenue.

OpenAI and Anthropic have genuine customers and rapidly expanding commercial businesses. Their computing requirements may continue increasing for years.

But the long-term economics will depend on how much revenue those AI applications produce relative to the cost of delivering them. Right now, infrastructure commitments are racing ahead of the cash flows needed to support them.

Could Nvidia's financing make AI cloud companies look healthier than they are?

Nvidia's financing can help AI cloud companies expand faster than their operating cash flow would normally allow, which makes their underlying profitability harder to judge.

CoreWeave provides a concrete example.

As noted above, its revenue increased from $229 million in 2023 to $1.9 billion in 2024 and $5.1 billion in 2025. Those figures establish genuine commercial growth.

Yet CoreWeave also reported net losses of approximately $594 million, $863 million and $1.2 billion over those same three years.

The company has relied heavily on external financing to construct and equip its infrastructure. Nvidia's equity investment and capacity purchase arrangement form part of that wider funding environment.

Nscale shows an even larger difference between current revenue and future ambitions, with $33 million in 2025 revenue and more than $103 billion in active and contracted total contract value reported in its September 2026 prospectus.

The contracts may support future revenue, but successful delivery requires financing, construction and reliable operation over several years.

Nvidia's financial involvement can make those projects more attractive to lenders and investors. But being able to raise money says very little about whether a data center will earn an attractive return after equipment, energy and borrowing costs.

That's the danger with this model. The companies can look increasingly successful as their contracted capacity grows, even while their financial dependence deepens.

Is Nvidia using investments to stop customers from switching to other AI chips?

Nvidia's investments help strengthen customer relationships at a time when Amazon, Google and Microsoft are developing alternatives to its GPUs.

Amazon has Trainium, Google operates TPUs, and Microsoft has invested in Maia. These companies can use their internal processors for selected workloads when the economics are attractive. When the big clouds run their own chips for some workloads, we ask which technologies are really gaining traction in our AI chips report.

Nvidia is responding through a broader technology strategy that includes GPUs, CPUs, networking, storage systems and software.

Its January 2026 CoreWeave partnership, for example, included plans for CoreWeave to adopt Nvidia CPU and storage platforms alongside future GPU generations. The Nebius partnership also involves collaboration on infrastructure architecture and software.

Nvidia's investments can accelerate adoption of this wider technology stack. A cloud provider building around multiple Nvidia products may find switching architectures more complicated than replacing an individual accelerator.

Financial support is therefore doing more than funding additional GPU purchases. It also helps Nvidia establish its wider technology ecosystem inside some of the fastest-growing AI infrastructure operators.

Still, financial support does not guarantee exclusive purchasing. Large customers have strong reasons to maintain alternatives, particularly if competing chips become cheaper for specific workloads.

What happens to Nvidia if AI cloud companies struggle to borrow money?

Nvidia would likely face slower orders from highly indebted AI cloud providers before suffering comparable pressure from established technology giants.

The recent bond-market reaction to CoreWeave offers an early example. Yields on some of its debt reached the mid-teens in October 2026, reflecting concern over leverage, financing costs and future spending requirements.

Other AI infrastructure projects have also encountered financing difficulties, construction delays and investor skepticism.

Financial Times reporting in October identified at least 47 major US data-center projects that had been stalled or canceled during the year. That does not establish a collapse in Nvidia GPU demand, but it shows that planned infrastructure spending is becoming harder to turn into completed facilities.

CoreWeave's situation remains contested. Some analysts have argued that its backlog and secured financing provide substantial protection, while others are concerned about future capital expenditures and debt refinancing.

Microsoft, Amazon, Google and Meta have large established businesses and more ways to finance investment. Younger cloud operators often need frequent access to debt and equity markets to keep expanding.

If capital becomes substantially more expensive, the second group could delay equipment purchases even when it has strong demand from customers.

Nvidia's infrastructure-financing partnerships may ease this pressure, but they cannot make economically unattractive projects profitable. A broader slowdown would become more serious if large hyperscalers also started cutting their AI spending.

What would prove Nvidia is financing too much of its own growth?

The strongest evidence would be Nvidia-backed customers repeatedly needing fresh funding to keep buying equipment, while their underlying computing businesses fail to produce enough cash to support those purchases.

Several existing figures give us useful starting points.

Nvidia's equity-security purchases have grown almost 34-fold year over year. Its accounts receivable increased approximately 64% in six months. Its guarantees now include a potential $105 billion obligation linked to OpenAI infrastructure.

Meanwhile, companies such as CoreWeave face substantial borrowing costs, and Nscale has disclosed the difficulty of securing financing for some of its largest computing commitments.

The most decisive future evidence would come from customer cash flows and financing needs.

If AI cloud providers increasingly pay for new equipment using cash generated by existing facilities, the industry will have demonstrated a healthier growth model. If instead they require repeated equity injections, expanding guarantees and ever-larger debt packages simply to fulfill existing contracts, Nvidia's role in financing demand will have become more problematic.

We still lack a complete public accounting of how much Nvidia revenue ultimately depends on its own investments, guarantees or customer payment arrangements. Any precise percentage would go beyond the available disclosures.

For now, the warning signs are visible, but they do not prove that Nvidia is propping up an otherwise unsustainable market.

Is Nvidia financing its own growth?

Yes, Nvidia is now financing a meaningful part of the AI infrastructure market that buys its products. The evidence, however, does not support the stronger accusation that its entire growth story depends on recycled money.

The scale of Nvidia's involvement has changed dramatically. It purchased $42.4 billion in equity securities over six months, helped finance several major cloud providers and participated in OpenAI's financing. Its SB Energy agreement also created a potential $105 billion financial obligation linked to future infrastructure used by OpenAI.

That last arrangement is particularly revealing. Nvidia's financial role now extends well beyond supplying GPUs or investing in startups. It is willing to support the financing risk associated with enormous facilities designed to operate its technology.

At the same time, Nvidia reported $96.2 billion in quarterly revenue, a 75% gross margin and $74.4 billion in operating cash flow over six months. Those are the results of a highly profitable business with substantial existing demand.

The clearest risk lies further down the supply chain. AI cloud providers are committing to massive infrastructure projects years before the associated revenue is fully realized. Some have secured enormous customer contracts while still facing high borrowing costs, operating losses and incomplete financing.

Nvidia benefits when those companies expand, so it has strong reasons to help them raise money. The financial exposure created by doing so is becoming increasingly significant.

Our judgment is that Nvidia's existing growth is real, but its future growth is becoming more dependent on financing the infrastructure that creates demand for its technology.

The next major test will be whether Nvidia-backed cloud providers can generate enough cash from their customers to finance further expansion on their own. If they can, Nvidia's investments may prove exceptionally valuable. If they cannot, the company could end up carrying a much larger share of the AI boom's financial risk than investors originally expected.