#26
Why build your own chips (AI5, Terafab)?
Interviews
7
Answers
8
X posts
12
Span
2025 to 2026
First asked Sep 7, 2025, most recently Oct 20, 2026. Excerpts below are from machine transcripts with filler words (um, uh, you know) and stutters removed; … marks a cut. They are not checked word for word, so use the link to hear the exact words, or the transcript link for the raw text. Timestamps: 0 cross-checked against YouTube captions, 7 from the archive transcript only, 1 checked against our transcript of the X replay. X has no link-to-time option, so X links open at the start.
How the answer evolved · oldest first
Sep 7, 2025 · All-In Summit · asked by Host
Q: Is it a potential long term competitor? What is it?
So at Tesla, we basically had two different chip programs, one Dojo and one Dojo on the training side. And then what we call AI4, it's just our inference chip that the AI force was currently shipping in all vehicles. And we're finalized finalizing the design of AI5, which will be an immense jump from AI4. By some metrics, the improvement in AI5 will be 40 times better than AI4.
Nov 14, 2025 · Baron Investment Conference · asked by Ron Baron
Q: What does that mean? What do you do?
It's Saturdays but sometimes Sundays. Usually recent weekends spent Sundays too. Yeah, just the AI5 chip which is going to be a great chip. You know all of Tesla hinges on that chip. That's the chip that goes into our next generation of self driving cars. And it's also essential for the Optimus robot. So that chip program was in bad shape. It wasn't closing because it's quite an ambitious chip design and it really …
Nov 14, 2025 · Baron Investment Conference · asked by Ron Baron
Q: Is your thought to have a partner to do it by ourselves?
Yeah, first of all I have immense respect for TSMC and Samsung and we've worked with both TSMC and Samsung at Tesla and at SpaceX. So TSMC and Samsung are great companies and we want them to make our trip as quickly as they can and scale up to as high as possible volume that they're comfortable doing but it doesn't appear to be fast enough. And you know when I ask how long will it take to from start to finish to get …
Jan 28, 2026 · Tesla Q4 and full year 2025 earnings call · asked by Analyst
Q: If you think about batteries into chips into memory and the efficiency of the system and what sort of advantages you Think you'll end up getting out of some of these purpose built devices that you'll end up integrating into multiple end …
Well, really all we're trying to do is make sure that we can scale to very high volume with autonomous vehicles, with human robots, and that we address geopolitical risk, which I think there's so many companies out there that are asleep with the switch with regard to geopolitical risk, or they just have their head in the sand and hope nothing bad will happen. I'm way more paranoid than that. I always think of Andy …
Apr 22, 2026 · Tesla Q1 2026 earnings call · asked by Analyst
Q: And how long is it going to take to ramp to get to yield that achieves that type of economic parity?
No, I mean Terafab is not some sort of mechanism to generate leverage over our chip suppliers. It's just literally we don't see a path to having enough, sufficient quantity of AI chips down the road as we scale production to high levels. Just the rate at which the industry is growing in logic, but even more so in memory is just doesn't we just anticipate hitting the wall if we don't make chips ourselves. So that's …
Jun 8, 2026 · SpaceX Technical Talk: AI Satellites at Scale · asked by Moderator
Q: Aside from just the size, what's going to make this unique different from any other chip building operation on the planet?
Well I think over time there's going to be a lot of technology evolution with the tariff app. But fundamentally it's about scale. So even if there were no fundamental technology breakthroughs and you simply, you could simply scale the existing chip making technology with a lot of difficulty to a terawatt of chip output per year. That's if you look at it just from the logic die standpoint, that's a, that's like …
Oct 10, 2026 · X Takeover 2026 · asked by Host (Tesla Owners Silicon Valley)
Q: What is something you have learned about chip making that you didn't appreciate maybe early on?
Well, I still have a tremendous amount to learn about chip making. I don't know how to make chips, to be clear. I have no idea how to make chips. But we'll figure it out. Go start somewhere. I had no idea how to make cars or rockets as well. And they say making chips is one of the hardest things to do in the world. I don't think it's easy, but I also look at the equipment and I'm like, okay, there's not a ton of …
Oct 20, 2026 · SpaceX Q3 2026 Earnings Call · asked by Adam Jonas (Morgan Stanley)
Q: Where do you see the biggest areas to improve their processes and supply chain when you do it your way?
I'll tell you what our internal expectations are … because we don't know if we will actually achieve this or not, but what we're aiming for internally is to make our first sort of test chip functional in approximately May or June of next year. … our aspiration is to reach volume production around Q3 of '28 at Terafab at Gibbons Creek, the main Terafab location.
Watch on X at 0:44:41 ↗Timestamp checked against our transcript of the X replay
In his own posts on X
@elonmusk · Nov 4, 2025
Slightly different versions of the Tesla AI5 chip will be made at TSMC and Samsung simply because they translate designs to physical form differently, but the goal is that our AI software works identically. We will have samples and maybe a small number of units in 2026, but high volume production is only possible in 2027. AI6 will use the same fabs, but achieve roughly 2X performance. Aiming for a fast follow to AI5, so hopefully mid 2028 for volume production of AI6. AI7 will need different fabs, as it is more adventurous.
View on X ↗@elonmusk · Jan 19, 2026
Solving AI5 was existential to Tesla, which is why I had to focus both the teams on that chip and I’ve personally spent every Saturday for several months working on it. This will be a very capable chip. Roughly Hopper class as single SoC and Blackwell as dual, but it costs peanuts and uses much less power. Now that AI5 is looking good, we have some headroom to work on Dojo3 too.
View on X ↗@elonmusk · Apr 15, 2026
Congrats to the @teslasi chip design team on taping out AI5! AI6, Dojo3 & other exciting chips in work. https://t.co/hm54TdIzBx
View on X ↗@elonmusk · Aug 6, 2026
Terafab Texas will be the largest and most valuable building on Earth by far. And it will be stunningly beautiful.
View on X ↗@elonmusk · Aug 6, 2026
@teslaownersSV My very approximate guess is that Terafab AI compute output would be ~25% for Tesla Optimus and ~75% for AI spacecraft
View on X ↗@elonmusk · Aug 7, 2026
Terafab will be 50 times the size of the Pentagon when complete
View on X ↗@elonmusk · Aug 29, 2026
@l_lucullus @wintonARK Yes, Terafab will be over a trillion dollars long-term if it achieves a terawatt/year of silicon, but that would still be an incredible ROI
View on X ↗@elonmusk · Oct 1, 2026
@DirtyTesLa We cut our RAM in half for the Tesla AI5 chip (now 72GB of LP5) and 1/3 for AI6 (now 144GB of LP6). This was the only way to get enough volume for Optimus production and greatly reduces cost. As it turns out, we think this will have a negligible effect on Optimus performance, as memory bandwidth is a bigger limiting factor than total memory storage (bandwidth was held constant).
View on X ↗@elonmusk · Oct 2, 2026
@DirtyTesLa Actually, we decided to nudge AI5 up a little to 96GB, as Tesla would otherwise be the only company using the min RAM version of LP5
View on X ↗@elonmusk · Oct 7, 2026
@herbertong @thejefflutz No, we will build and run the fab. Let there be ZERO doubt about that. Maybe TSMC subleases part of the Terafab if they want, but nothing more than that.
View on X ↗@elonmusk · Oct 12, 2026
Join the Terafab project. The value of the output of this massive structure will rank in the top 20 of country GDPs!
View on X ↗@elonmusk · Oct 13, 2026
@XFreeze Yes [asked if Terafab is a massive step toward Kardashev Type II]
View on X ↗
Legend: Timestamp cross-checked; Timestamp from archive, not cross-checked; Unverified: caption-only timestamp; Timestamp checked against our transcript of the X replay; Unverified timestamp: from archive transcript, not checked against the X video. Report an error →