Amazon wants to sell about $8 billion worth of Nvidia chips to outside investors — and keep using every one of them.

According to The Financial Times, Amazon has discussed selling thousands of Nvidia Grace Blackwell chips that already run inside more than a dozen of its data centers.

The new owners wouldn’t take possession of the hardware. Instead, Amazon would lease the chips back and continue using them to run AI.

By selling chips it already owns and leasing them back, Amazon could recover billions in cash and put that money toward its next round of AI spending.

The deal would work through a special-purpose vehicle, or SPV — a separate company created specifically for the transaction. Outside investors would put money into the SPV, which would buy the chips from Amazon. Amazon would then pay the SPV to keep using them.

Amazon gets cash and keeps the computing power. Investors collect the lease payments and own the hardware.

But the hardware poses a problem.

Nvidia says its advanced GPUs can generate revenue for up to a decade. Banks take a more conservative view, typically underwriting GPUs on a three-to-four-year depreciation schedule.

Nvidia keeps producing faster, more efficient hardware, which can lower the price customers will pay to use older chips — and the price investors could recover by selling them.

That leaves investors putting billions of dollars into hardware without much history to tell them what those chips will be worth several years from now.

The potential Amazon deal represents just one piece of a much larger financing boom.

Companies plan to spend more than $5 trillion on AI infrastructure through 2030, from chips and data centers to cooling systems and power generation. Big Tech and traditional banks won’t provide all of that capital.

So Wall Street is turning to another source of capital: private credit.

Your Rundown for Monday, October 5, 2026...

After SaaS, Private Credit Bets on AI

Private credit simply means loans made by investment firms and funds instead of traditional banks. The industry has grown rapidly by lending to companies that banks might avoid or by structuring loans banks don’t want to hold.

Private-credit loans to software-as-a-service companies, or SaaS, grew from about $8 billion in 2015 to more than $500 billion by the end of 2025, according to the Bank for International Settlements.

The appeal seemed straightforward. SaaS companies sell software through monthly or annual subscriptions. Customers tend to stick around, producing recurring revenue that lenders could count on when borrowers made their debt payments.

Then AI disrupted that model.

Generative AI can write code, automate tasks and perform jobs that once required specialized software. That has put pressure on some software companies — and the loans behind them.

Blue Owl’s technology-focused private-credit fund, for instance, received $1.1 billion in withdrawal requests for the third quarter of 2026, equal to 39% of the fund’s shares.

But Blue Owl allows investors to withdraw only about 5% of the fund’s shares each quarter. Investors have now spent three straight quarters trying to pull large amounts of money from the fund.

Blue Owl (NYSE: OWL) is also publicly traded, which means ordinary stock investors can own shares of one of the country’s largest private-credit managers.

And now private credit wants a major role financing AI itself.

Carlyle, one of the world’s largest private-credit managers, estimates private-credit firms could lend roughly $1 trillion to finance AI infrastructure.

Last week, Carlyle published a paper called “Financing AI Without a SaaS Repeat.”

The firm warns that AI lending could prove even more speculative than software lending. Lenders must evaluate new financing structures, rapidly changing technology and projects that have never faced a serious economic downturn.

Carlyle research chief Jason Thomas has compared the approach some lenders take toward data centers with mortgage lending before the 2008 financial crisis.

If lenders overestimate the value of AI chips, data centers or the revenue they generate, funds could take losses. Investors could seek withdrawals, lenders could tighten their standards and heavily indebted companies could struggle to refinance.

Higher financing costs could then delay data-center projects and raise the cost of the AI buildout.

Private credit poured hundreds of billions into software because recurring revenue looked dependable. AI exposed the risk in that assumption. Now the same lenders are preparing to finance the technology that disrupted it.

Jim Rickards warned Strategic Intelligence readers about the private-credit crisis back in May, before the industry’s latest push into AI.

Jim’s advice to investors: “Review your 401(k)s and IRAs for exposure to private credit and reduce it where possible.”

Some exposure is easy to spot. Blue Owl, for example, trades publicly under the ticker OWL. But private-credit risk can also show up through the banks that finance the industry.

Jim pointed out that Wells Fargo, Bank of America, PNC, Citigroup, JPMorgan Chase and Goldman Sachs all have exposure to private credit.

That doesn’t mean investors should dump their bank stocks. But losses in private credit could eventually affect the banks that finance the industry — and their shareholders..

Market Rundown for Monday, Oct. 5, 2026

S&P 500 futures are slightly in the red at 7,774.

Oil is down 1.10% to $90 for a barrel of WTI.

Gold is up 0.50% to $4,183.20 per ounce.

And Bitcoin is up 0.75% to $86,075.