Legendary game developer John Carmack recently stated on Twitter that competing against TSMC head on was pretty much impossible. But he then asked his followers how less impossible it would be for a big tech company to make its own bespoke chips, rather than competing against the general purpose companies. Cue one small reply from a certain Elon Musk: “Tesla could do it, but I sure hope we don’t have to.”
Now, just to be clear on something, Musk’s tweet wasn’t about Tesla trying to become the next TSMC. After all, if the likes of Samsung, GlobalFoundries, and even Intel struggle, then somebody stepping into the battlefield from scratch stands zero chance. Instead, he was referring to Carmack’s point that not only is competing against TSMC nigh on impossible, it’s equally challenging for a company to build its own chip factory, just to make processors specifically for its products.
At the moment, Tesla relies on Samsung’s production lines for the main chips used to handle its car’s Autopilot feature. In theory, it could buy and hire all of the equipment and staff required to have its own fabrication plant and stop outsourcing chip production. All of the latest processors in Telsa cars are designed in-house and it’s only the manufacturing of them that isn’t done internally.
This is no different to how AMD and Nvidia operate. Both companies design everything themselves, then rely on little foundries for building test chips, and TSMC and Samsung to manufacture the processor at the enormous scale required to sell millions of chips worldwide, each year.
Tesla could do it, but I sure hope we don’t have toMarch 24, 2024
Musk’s claim that “Tesla could do it” isn’t perhaps entirely hyperbole, despite Carmack inferring that making very specific chips isn’t really any less impossible than making them en masse for the general market. Although it’s not the same level of complexity, Tesla has switched from buying Li-ion batteries from parties such as Panasonic, to building huge factories to make its own.
But what about the “I sure hope we don’t have to” comment? Well that’s quite obvious: Nobody in their right mind would want to spend billions of dollars on something that has the distinct potential of not working right for years, when you can just contract a highly experienced and cost-effective business to do it for you.
However, given the increasing concern over the reliance of the Korean/Taiwanese/Chinese supply chain and massive importance of processors in today’s economy, there’s always the risk that Musk’s hand may be forced.
With the US Chips Act offering billions of dollars in loans and grants to increase the amount of chip production in the States, I do wonder if Musk is seriously considering spending a few dollars of his own and stepping into that market.
After all, if he was happy to spend $44 billion on acquiring Twitter, what’s another 50-odd billion dollars on a foundry to such a person?
www.pcgamer.com