Nvidia (NVDA) and Broadcom (AVGO) already make billions selling the technology powering the artificial intelligence boom.
Now they’re increasingly helping customers find the money to buy it.
Bank of America (BAC) says major chip suppliers are taking on an unexpected role as “credit intermediaries,” helping remove some of the financial risks that can make massive AI data centers difficult or expensive to finance.
Nvidia is not twiddling its thumbs.
The chip giant says it’s partnering with Apollo Global Management (APO), BlackRock (BLK), Blackstone (BX), Brookfield, Goldman Sachs (GS), and KKR (KKR) to raise over $500 billion in third-party capital for AI infrastructure.
The collaborations aim to build specialized pools of finance that Nvidia’s clients may use at reasonable rates, it added.
That’s a hell of a lot of cash.
But the sector could need it.
Bank of America forecasts that AI capital expenditures might reach over $5 trillion between 2026 and 2030, and the sector could need about $1.2 trillion of external finance to back the increase.
That provides an intriguing trade-off for investors in both NVDA and Broadcom.
More infrastructure funding for consumers might keep chip demand soaring.
It may also shift some of the financial risk from the AI boom back to the firms selling the technology.
Nvidia and Broadcom move deeper into AI financing
The financial problem begins with a basic mismatch.
AI infrastructure demands a lot of money upfront, whereas most of the income anticipated from such data centers comes later.
This challenge is particularly relevant for firms that don’t produce the large cash flows like Microsoft (MSFT), Amazon (AMZN), Alphabet (GOOGL) or Meta Platforms (META).
Companies like Nvidia and Broadcom are progressively helping to close that gap, Bank of America argues.
They are not necessarily becoming typical lenders.
Instead, suppliers may help finance via minimum revenue commitments, take-or-pay contracts, residual value guarantees, and other arrangements that lessen the risk that lenders would otherwise have to assume.
Nvidia’s own actions increasingly corroborate that theory.
The company’s $500 billion fundraising push aims to transform Nvidia’s processing infrastructure into an asset that institutional investors can finance like other huge infrastructure projects.
Here’s how Nvidia CEO Jensen Huang explained the change:
“We began by building chips; today, we are helping create a new class of productive, investable infrastructure: AI factories.”
In August, Nvidia went one step farther.
The business will offer finance assistance for land, electricity, and construction at SB Energy’s PORTS-Pike Technology Campus in Ohio.
Related: Relatively unknown cloud provider gets Nvidia’s stamp of approval
The first rollout will provide 4.25 gigawatts of AI factory capacity, and OpenAI is expected to use the computer infrastructure. Nvidia also said it was investing $1.5bn in SB Energy.
Those aren’t small commitments.
They illustrate how far Nvidia has moved beyond merely shipping graphics processing units.
Broadcom shows how guarantees can cut AI borrowing costs
Broadcom is another example of how supplier assistance might alter the economics of an AI finance arrangement.
Bank of America said Broadcom’s AI XPV Platform has secured senior notes and the remaining value of chips covering $31 billion of an initial $35 billion loan package arranged with Apollo and Blackstone.
That was nearly 87% of the debt package.
But more importantly, the guaranty seems to have made the loan far less expensive.
The guaranteed part was priced at 5.75%, against 8.5% for an unsecured second lien, Bank of America stated.
Put simply, Broadcom’s backing helped lenders accept a lower interest rate. For Broadcom, supporting those projects has a clear commercial logic. Its AI business is already expanding at a remarkable pace.
Broadcom on Sept. 2 reported $16.7 billion in AI semiconductor sales for its fiscal third quarter, up 221% from a year earlier and 54% from the prior quarter.
CEO Hock Tan said Broadcom expects AI semiconductor revenue to reach about $21.7 billion in its fiscal fourth quarter, an increase of 236% year-over-year.
That financial strength helps explain why chip suppliers might play a greater role in financing.
A supplier may gain twice if backing a finance package helps develop another huge AI campus.
It makes the project feasible.
Then it offers pricey hardware to fill it.
Big Tech spending shows how massive the AI arms race has become
Bank of America’s forecast of $5 trillion seems outrageous.
But the expenditure in recent weeks by the world’s biggest technological corporations says otherwise.
In its fiscal fourth quarter, Microsoft said it spent $41 billion in capital expenditures, with around two-thirds of it on shorter-lived equipment, mostly CPUs and GPUs.
The business also booked $5.6 billion in financing leases, mostly for major data center facilities.
Meta is spending at a similarly breakneck rate.
Facebook’s parent now expects capital expenditures to be between $130 billion and $145 billion in 2026, narrowing its previous forecast of $125 billion to $145 billion.
Meta has said its spending is being driven by infrastructure investments needed to support its AI efforts and core business.
Amazon’s numbers illustrate just how fast those investments can eat up cash.
Operating cash flow increased 33% to $161.4 billion for the 12 months ended June 30, the company said.
But free cash flow slipped to an outflow of $7.6 billion from an inflow of $18.2 billion a year ago.
Amazon said the decline was driven primarily by a year-over-year increase in purchases of property and equipment of $66.1 billion.
That spending is paying off for Nvidia.
More Nvidia:
- Nvidia just made a move Wall Street wasn’t ready for
- Nvidia just locked down deal that changes AI race
- Nvidia stock is doing something it hasn’t done in years
The chipmaker reported Aug. 26 that fiscal second-quarter revenue doubled to $96.2 billion and Data Center revenue rose 117% to $89 billion.
Simply put, hyperscalers are pouring extraordinary amounts of cash into infrastructure, and Nvidia and Broadcom are making increasingly extraordinary amounts of revenue off of that spending.
Financing is then the mechanism that can help sustain that cycle.
The $5 trillion AI forecast may not be as wild as it sounds
Bank of America’s basic assumption is that almost $5 trillion in cumulative AI capex will occur between 2026 and 2030.
About $3 trillion of it is for spending to increase capacity.
Of that spending, around $1.4 trillion is expected to go on chips specifically, and about $2.1 trillion in total on information technology equipment.
Eventually, overall IT investment might exceed $4.1 trillion as companies replace and upgrade equipment.
Independent predictions point to a comparably huge build-out.
Global data center consumption might climb from around 82 gigawatts in 2025 to roughly 220 gigawatts by 2030, according to McKinsey.
McKinsey forecasts that data centers might need around $6.7 trillion in total worldwide capital investment through 2030, with about $5.2 trillion of that related to infrastructure supporting AI workloads.
Power might be as vital as cash.
Energy demand from global data centers will more than quadruple to almost 945 terawatt hours by 2030, the International Energy Agency estimates.
That’s a little more juice than Japan uses now.
The government anticipates data centers in the U.S. will account for roughly half the rise in power use by the end of the decade.
Those forecasts reveal what is abnormal about this AI cycle. This isn’t just the latest software investment cycle. It needs land, buildings, chips, cooling systems, transmission infrastructure, and a lot of electricity. And it all has to be paid for. And much of it has to be paid for before the projects start to make money back.
Bank of America sees a $1.2 trillion AI funding problem
Bank of America studied eight large businesses that are scaling up in AI and cloud: Microsoft, Amazon, Alphabet, Meta, Oracle (ORCL), SpaceX, CoreWeave (CRWV) and Nebius Group (NBIS).
The bank forecasts they may provide around $6.6 trillion in operational cash flow compared with about $6.9 trillion in capital spending through 2030.
That leaves a modest $300 billion shortfall, at first blush.
The difficulty is the unequal distribution of the funds.
Amazon, Alphabet, Meta, and Microsoft are forecast to produce about $5.5 trillion in operational cash flow versus $5.2 trillion in capital expenditure.
The other four corporations have around $1.7 trillion of capital expenditures vs $1 trillion of operational cash flow.
So you get a deficit of something like $700 billion.
And even the giants that can finance infrastructure themselves may borrow.
Borrowing may help conserve funds for acquisitions, dividends, share repurchases and other investments.
That means the overall external capital need may be as high as $1.2 trillion through 2030, according to Bank of America, with debt taking most of the hit.
The bank sees the after-tax cost of borrowing at about 5%, compared to a cost of equity above 11%.
If the full $1.2 trillion were debt-funded, it’s almost $300 billion per year in new credit issuance.
Bank of America thinks the credit markets can manage it.
Wall Street is already getting ready to try, as shown in the effort by Nvidia to raise more than $500 billion from some of the world’s biggest sources of cash.
Nvidia and Broadcom investors face a new kind of AI risk
The lure for chipmakers is obvious. More financing means more data centers. More data centers, more chips. But the financial guarantees do not remove the risk. They pass it on.
If a supplier is backing revenue, assets, or elements of a financing package, it could be exposed if projects fall short, customers struggle, or the value of AI hardware drops faster than expected.
And that’s why the trend is relevant for Nvidia and Broadcom investors.
The big question of the AI boom has been how much hardware the hyperscalers will buy most of the time.
Shareholders may increasingly ask how far suppliers will go to ensure those customers can afford it.
That’s a whole different type of relationship.
That change is especially evident in the August deals from Nvidia.
The company that symbolized the AI boom by selling GPUs now works with some of the world’s biggest asset managers and investment banks to make AI computing infrastructure something global capital markets can finance.
If finance keeps the AI building going, it would be very bullish.
It might also generate hazards that didn’t exist when Nvidia and Broadcom were just chipmakers.
For investors, the next phase of the AI boom might mean looking beyond revenue, margins, and GPU shipments.
They’ll also have to scrutinize who’s funding the boom and who bears the risk if the trillions being spent don’t pay off the way everyone anticipates.
Related: Jensen Huang just answered Michael Burry’s Nvidia bear case
