News from the IT community on October 2nd: The world's richest man, Elon Musk, posted a tweet on the X platform today (October 2nd), announcing that in order to allocate sufficient memory for the mass production of Optimus robots, he plans to reduce the memory usage of Tesla's AI5 and AI6 chips.

This adjustment can be traced back to Micron Technology's predictions. During its latest financial report conference call, Micron Technology's Chief Financial Officer pointed out that humanoid robots will become a new driver for memory demand: each device will require more than 200GB of DRAM as well as several TB NAND of flash memory. Even if the artificial intelligence bubble bursts, memory prices are unlikely to fall.

Netizen @DirtyTesLa shared this speculation, pointing out that Musk previously believed that the number of robots would exceed that of humans. If we calculate based on 10 billion robots, then that would require 2 trillion GB of memory, yet currently the world produces only about 45 billion GB of DRAM per year. IT attaches the relevant screenshots below:

Elon Musk then retweeted this tweet, stating that in order to reserve enough memory for the production of Optimus, the company has decided to reduce the memory configuration of Tesla's AI5 chips and Tesla's AI6 chips:

• AI5: RAM has been reduced from the original planned 144GB to 72GB, which is a 50% decrease. LP5
• AI6: RAM has been reduced from the original planned 216GB to 144GB, which is a decrease of about 33%. LP6

He also stated that memory bandwidth is more likely to become a performance bottleneck than the total capacity, and since the bandwidth remains unchanged, the expected impact on the performance of Optimus is "negligible."