Amazon Wants Investors to Own Its $8 Billion in Nvidia Chips — While Amazon Keeps Using Them
Amazon is reportedly exploring a financial maneuver that treats cutting-edge AI chips less like computer equipment and more like a commercial jet or an office building: something you don't have to own to use.
According to the Financial Times, Amazon is in discussions to shift roughly $8 billion worth of Nvidia's advanced Grace Blackwell AI chips off its own balance sheet, through a special-purpose financing vehicle.
Here's how it would work, and why it matters:
→ Under the proposed structure, outside investors would legally own the hardware
→ Amazon would continue using those same chips in its US data centers through lease arrangements
→ The deal is still under discussion — not yet a completed transaction
→ Amazon expects roughly $220 billion in capital spending this year, with much of it going toward cloud infrastructure, chips, and data centers
→ Shifting the asset off-balance-sheet frees up capital for Amazon to keep expanding elsewhere, without the chips weighing down its reported liabilities
Here's the bigger pattern this fits into: AI infrastructure has gotten so expensive that even a company with Amazon's balance sheet is looking for ways to spread the cost. When something becomes this capital-intensive, companies typically turn to the same playbook used for airplanes, real estate, and power plants — separate the ownership of the physical asset from the right to use it, and let institutional capital take the ownership risk.
If this becomes a trend among the big cloud providers, it opens up an entirely new investment category: AI chips as a financeable asset class, sitting between venture capital and traditional infrastructure finance. It also adds yet another layer of financial interdependency to an ecosystem that already includes circular deals between chipmakers, cloud providers, and AI labs.
The part worth sitting with: when hardware that depreciates as fast as AI chips becomes something investors "own" through a leasing structure, who actually carries the risk if a newer, faster chip generation makes last year's GPUs obsolete sooner than the depreciation schedule assumes?
What's your take — smart financial engineering for an unprecedented capital cycle, or a warning sign about how overstretched AI infrastructure spending has become?
#AI #Amazon #Nvidia #AIInfrastructure #TechFinance #CloudComputing
— 𝔖𝔞𝔫𝔡𝔢𝔢𝔭 ℜ𝔞𝔦𝔷𝔞