Amazon is exploring a deal to move about $8 billion of advanced Nvidia chips into a new vehicle and lease them back, a move that underscores how aggressively Big Tech is financing the AI buildout even as the cash burn keeps rising. The plan would shift thousands of Nvidia Grace Blackwell chips into a special-purpose vehicle, with Amazon then renting the hardware back while outside investors fund the structure through debt. Amazon has been talking with investors in recent weeks, but the effort remains exploratory and no closing has been announced.
The deal, reported by Reuters and Financial Times, would give Amazon a way to strengthen its balance sheet while keeping the compute it needs to power AWS. That matters because AI infrastructure is becoming one of the most expensive race tracks in corporate America. Amazon expects to spend over $200 billion in capital expenditure this year, mostly on AWS chips and data centres, according to the fact pack. Turning a slice of that hardware into a financing vehicle may help ease pressure on the company’s own books without slowing the pace of investment.
The structure would place the chips into an SPV that raises money from outside investors through debt issuance. Amazon would also offer an equity stake of up to 10% in the vehicle, according to Reuters. In practical terms, the company would be converting owned or leased hardware into a separate financing asset while preserving its access to the machines. That kind of move is not a sign that Amazon is pulling back from AI. It is evidence that the spending campaign is getting large enough to inspire more creative funding tools.
The chips at the center of the plan are not sitting in a warehouse. Reuters said they are installed in more than a dozen US data centres across five states, including Nevada and Virginia. That detail matters because it suggests the proposed deal is tied directly to working infrastructure, not just inventory on a balance sheet. Amazon has already paid, or is already on the hook, for hardware that is in use. Packaging those chips into an SPV could let it keep that capacity online while changing how the asset is financed.
Amazon is hardly alone in pushing AI capex higher, but this proposal shows where the stress point is likely to emerge: the balance sheet. The company is trying to support an enormous buildout while also maintaining flexibility for the rest of the business. Offloading hardware into a vehicle can be read as a form of balance sheet engineering, one aimed at protecting liquidity and possibly improving optics around asset intensity. For investors, the message is simple: the AI boom is now big enough to require financial structures once reserved for more exotic assets.
That does not mean the market should assume distress. Amazon remains one of the strongest cash-generating companies in tech, and the move is being described as a way to strengthen its balance sheet, not as an emergency capital raise. But the structure does reveal a key tension in the AI trade. Companies want the upside of owning next-generation compute, yet the costs are so high that ownership itself can become a financing problem. Amazon appears to be testing a middle path: keep the chips, outsource part of the funding burden.
The timing is also notable. Amazon has held talks with investors in recent weeks, which indicates the concept is still being market-tested. There is no announced filing, no closing date and no vote. In other words, this is a story about corporate intent, not completed execution. That distinction matters because the market often prices in AI infrastructure ambitions long before the accounting catches up. Here, the accounting is the story. The company’s appetite for chips is large enough that investors are being asked to help carry the load.
Nvidia sits at the center of the plan because its Grace Blackwell chips are the hardware Amazon wants to finance through the vehicle. That places Nvidia in a familiar but important role: the company is not just selling shovels for the AI gold rush, it is also becoming part of the asset base that others may securitize, lease and refinance. The fact pack does not show any direct market reaction in Amazon or Nvidia shares, so there is no verified price move to tie to the news. Even so, the signal to the market is hard to miss.
For Nvidia, the headline is about demand durability. Amazon is so committed to AI infrastructure that it is exploring ways to fund a stockpile of Nvidia chips with outside capital. That is the kind of behavior that speaks to sustained appetite for high-end accelerators. But it also suggests a more complex future for the ecosystem. If hyperscalers increasingly finance chips through special-purpose vehicles, the AI supply chain may start to look less like simple product sales and more like a layered credit market built around compute.
That is where the drama gets bigger than one company. A financing structure around Nvidia hardware raises questions about how much of the AI buildout will be carried on corporate balance sheets and how much will be pushed into structured vehicles that sit beside them. Amazon’s plan, if it advances, would be an early example of that shift. The company would still control the assets operationally, but investors would help fund them through debt and equity exposure in the SPV. The economics of AI infrastructure would become more fragmented, and perhaps more opaque.
There is no verified stock move tied to the report, which limits how far the market reaction story can be pushed. Still, the absence of an immediate price shock does not make the disclosure less important. If anything, it shows how normalized giant AI spending has become. A company contemplating an $8 billion chip financing structure can surface as a strategic financing note rather than a crisis headline. That is a remarkable change from just a few years ago, when hardware capex on this scale would have seemed extraordinary on its own.
Amazon and Nvidia did not immediately respond to Reuters for comment, so the public record remains thin on details. The exact $8 billion figure is attributed to unnamed FT sources and has not been independently confirmed by either company. One syndicated version also offers a conflicting read on the equity split in the vehicle, with differing descriptions of whether Amazon would retain any stake. For now, the cleanest conclusion is that Amazon is exploring a complex financing tool to support a massive AI investment cycle, and the details are still moving.
What matters most is that Amazon is not backing away from the AI arms race. It is trying to make the race easier to fund. If the transaction advances, it could become a template for other hyperscalers facing the same problem: how to keep buying premium chips and building data centers without letting capex swamp the balance sheet. For now, the chips remain in Amazon’s data centers, the investors remain in talks and the structure remains under discussion. The next move will show whether this is a one-off financial workaround or a new way to pay for AI at scale.
