Market reports say Amazon’s AWS is expected to receive $5 billion worth of Nvidia GB300 racks. But so far, the only evidence is a brief report, and the scale of the purchase, delivery arrangements, and whether the deal has been finalized all remain unverified.
My view is that even if the report is accurate, the arrival of the racks does not mean AI spending has already paid off. Another report, citing research data, said that in September, the average weekly trading volume of credit default swaps for seven companies, including Amazon, exceeded $9 billion, compared with less than $2 billion in July. These figures are the combined total for all seven companies and should not be taken as evidence that Amazon’s own credit risk has risen.
Public discussion has also mentioned that Amazon will report earnings next week, though the exact date still needs to be verified. The next things to watch are whether the company officially confirms the rack arrangements and whether AWS revenue and cash flow can justify the spending. If the order is confirmed but the returns remain unclear, will the market focus first on expanding computing capacity or on calculating the cost of funding?
My view is that even if the report is accurate, the arrival of the racks does not mean AI spending has already paid off. Another report, citing research data, said that in September, the average weekly trading volume of credit default swaps for seven companies, including Amazon, exceeded $9 billion, compared with less than $2 billion in July. These figures are the combined total for all seven companies and should not be taken as evidence that Amazon’s own credit risk has risen.
Public discussion has also mentioned that Amazon will report earnings next week, though the exact date still needs to be verified. The next things to watch are whether the company officially confirms the rack arrangements and whether AWS revenue and cash flow can justify the spending. If the order is confirmed but the returns remain unclear, will the market focus first on expanding computing capacity or on calculating the cost of funding?