00:00Ed, let me bring you into the conversation. What difference does it make for Intel now?
00:04Yeah, to refactor means basically in layman's terms to modernize and improve the chip manufacturing process.
00:10And so that is Intel's pitch, that they would come in with the knowledge and expertise of having scaled output
00:16across various nodes.
00:18The difficulty is that right now there is a lot of unknown about the TerraFab project.
00:24It's not reflected in Tesla's full year capital expenditures forecast, for example.
00:28And their goal of one terawatt of compute capacity is basically many factors greater than all of the current global
00:36capacity from TSMC and Samsung for lead edge nodes, lead edge manufacturing of AI chips.
00:43And so there is a great question mark of how they'll do it.
00:45But like analysts have estimated you're talking over an aggregate period of like trillions of dollars of spending to get
00:52to that scale in the end.
00:53So if this was something that we were chasing and it's interesting that Intel notes in its post, Intel, that
01:00Elon Musk visited with them over the weekend in Santa Clara.
01:04Very interesting indeed.
01:06Ed, you know, Intel has obviously sold a stake or President Trump and the United States have bought a stake
01:13there.
01:13So they're, you know, in bed together.
01:16I'm never sure about Elon Musk, but at least at one point he was very deeply in bed with President
01:21Trump.
01:22Is this whole sort of click going to help Intel recover from its incredible fall to get back to what
01:33it once was?
01:35Well, no, because if you think about the pathway of an AI chip through a fab, it goes through many
01:42processes and touches many hands, many machines.
01:45And those machines come from LAM, KLA, applied materials and ASML ultimately when it comes to lithography.
01:53Intel may well help the TerraFab project, which is a JV between Tesla, SpaceX and SAI, with the planning, right?
02:01The structure of this project, but the spending is going to be on others.
02:06It's going to be on the chip equipment makers, not necessarily on Intel.
02:09And Intel's story to date very simply is that it has not been able to scale or find customers other
02:16than itself for its lead edge processes.
02:19There isn't any evidence that its next generation nodes have high volume throughput yet.
02:26And so that's not immediately clear that they'll be able to translate what they will say is expertise and experience
02:32to Elon Musk's project anyway.
Comments