The semiconductor scarcity in the course of the pandemic was actual, with costs rendering them virtually
“untouchable.” Though that’s historical past at this level, OpenAI CEO Sam Altman is taking part in the lengthy recreation. A brand new Bloomberg report reveals that Altman is trying to elevate “billions” for organising a sequence of AI chip-making factories.
Altman believes that there’s solely a lot that industry-leading foundries reminiscent of TSMC, Intel, and Samsung can do with regards to making AI-centric chipsets sooner or later. That is why, he desires to arrange a ‘community of AI chip factories’, in accordance with the report.
That will instantly place OpenAI within the huge leagues’ competitors, proper alongside Google, Microsoft, Amazon, and the like. These giants have been making their personal chipsets for a while now, as a substitute of utilizing off-the-shelf processors.
Furthermore, firms like Google taking a full-on customized route for his or her chipsets solely proves that the AI {industry} is reaching new heights each single day and requires such experiments. Beforehand, Google used Samsung’s foundry to construct their Tensor chipsets. Now, with the next-gen Tensor G5, they’ll most likely be dropping them for TSMC.
Nonetheless, that isn’t what Sam Altman desires to do, from the appears to be like of it. The report solely suggests that he’s trying to elevate cash to arrange an impartial AI chip-making big with out having to depend on anybody. He believes that AI applied sciences are on their approach to going wall-t0-wall impartial in supporting their very personal semiconductor provide chain.
Altman in Lively Discussions With Buyers
In line with the report, Altman is already in discussions with buyers like Abu Dhabi-based G42 and SoftBank Group. The dialogue is primarily targeted on not solely assembly the present AI chip scarcity but in addition trying to meet any future necessities.
Furthermore, this undertaking is a far cry from being something however substantial. The report suggests funding discussions with G42 alone are going north of $8 billion. Fabrication nodes for chip architectures are getting smaller by the day, and mass-producing a 2nm or 3nm class node might price roughly $30 billion proper now.
The bottomline is that the expertise required for contemporary fabrications is dear, with a single device costing means over $100 million at the moment. What do you suppose? Is that this transfer going to make Altman the AI chip messiah, or will Nvidia reign supreme? Tell us what you suppose within the feedback down beneath.
