NewsNTech
Memory costs for AI infrastructure have become severe enough to tip the books at three of the largest technology companies simultaneously.
Amazon, Alphabet, and Tesla each posted negative cash flow in the latest quarter as AI buildout spending accelerated beyond what operating income could offset. Meta saw its cash generation fall 91%.
The memory cost constraint The mechanism is direct.
AI accelerators require large volumes of high-bandwidth memory, and procuring that memory at scale, across multiple companies expanding in the same period, compresses supply and inflates cost.
Keep reading