- According to people familiar with the matter Sam Altman is shopping to secure funding to build AI chips.
- Reports suggest the OpenAI chief exec is looking to secure as much as $7 trillion in funding.
- The funding would be used to establish semiconductor fabrication facilities that would be run by the likes of TSMC.
One of the scenes from David Fincher’s The Social Network that stuck with us was when Jesse Eisenberg, portraying Mark Zuckerberg, had a meltdown about Facebook going down. This is still seemingly a big deal for the real Zuckerberg because, as you might be aware, Facebook – now Meta – runs mostly on its own bespoke hardware.
This makes sense as while hyperscalers are reliable, using your hardware means you are in control.
We bring this up because OpenAI chief executive officer Sam Altman is reportedly shopping around for investors to help him fund the production of advanced chips to power artificial intelligence.
Since ChatGPT hit the market there has been surge in demand for hardware capable of sustaining the computing requirements of AI. This has seen Nvidia and Intel’s bottom lines burgeoning to unthinkable levels and clearly, Altman wants a slice of that pie.
As reported by the Wall Street Journal, via Al Jazeera, according to people familiar with the matter, Altman is trying to raise up to $7 trillion to fund the production of AI chips. Per the report, the pitch is that Altman would use the funding to build foundries that would then be run by the likes of the Taiwan Semiconductor Manufacturing Company (TSMC).
Not only would this give Altman access to what has become a hugely lucrative chip market but also means that OpenAI wouldn’t be subject to the peaks and troughs of chip production.
Would it work? We don’t see why it wouldn’t but $7 trillion is a lot of money to go asking for even if Altman has reportedly approached the likes of the United Arab Emirates officials, SoftBank and TSMC executives for this funding.
As much as the plan to build one’s own chips makes sense – see Apple’s success with its in-house processors – developing chips for one purpose is something that hasn’t worked in the past. During the crypto boom of the late 2010s, GPU manufacturers began selling cards that were intended to be used exclusively for mining cryptocurrency. These GPUs were short-lived but it highlighted that like Alton Brown, people in the technology sector are resistant to investing in a unitasker.
A chip for AI sounds good right now but will it still be a good idea in five years? A semiconductor fabrication facility can take five or more years to build and over the last decade we’ve seen a number of revolutionary technologies fizzle into obscurity.
With that said, we wish Altman the best of luck in this endeavour because if he pulls it off, it will be rather monumental not only for OpenAI but the AI sector as a whole.
[Image – Brian Kostiuk on Unsplash]
