Latest / In the Loop AI / THIS WEEK IN AI: NVIDIA Dominates Google | America vs Open Source | Tesla Starlink V5

Transcript

THIS WEEK IN AI: NVIDIA Dominates Google | America vs Open Source | Tesla Starlink V5 · In the Loop AI

0:00
Ejaaz: NVIDIA just announced the key to one of AI's biggest problems, cost.
0:04
Ejaaz: Companies spend between tens to hundreds of millions of dollars every year and
0:08
Ejaaz: are running out of money.
0:09
Ejaaz: Google, for the first time yesterday, reported a negative cash flow on their
0:13
Ejaaz: quarterly earnings. They're officially losing more money than they are making.
0:17
Ejaaz: That's because they're spending so much on AI infrastructure.
0:21
Ejaaz: NVIDIA's new Vero Rubin officially got stood up yesterday and it saves you 10x
0:26
Ejaaz: on the tokens that you spend, which means you can have a 10x better model for the same cost.
0:30
Ejaaz: There's all of this and Google's new Gemini 3.6 models, which is very,
0:34
Ejaaz: very underwhelming on today's roundup.
0:37
Josh: Yeah, so we have to start with the big news of the day, which is that Vera Rubin is actually shipping.
0:42
Josh: And for those who are not familiar, Jensen came on stage at NVIDIA GTC and announced these a few months ago.
0:47
Josh: The news today is that they're finally live and they're actually operational.
0:50
Josh: And we started to get an idea of what these things look like.
0:53
Josh: And I want to preface a section with the idea that we just covered an episode
0:57
Josh: yesterday about how GPT-6 or whatever the new internal open AI model was,
1:02
Josh: they broke out of internal air-gapped containment
1:05
Josh: and hacked into a public-facing website into their production database to steal secrets.
1:11
Josh: That was on Blackwell chips. The new chips are Verorubin.
1:15
Josh: No models have been trained on Verorubin chips, but the math behind how much
1:19
Josh: more powerful they are is so unbelievably impressive.
1:22
Josh: It's like, oh my god, this is this feels like the end game it's like once these
1:25
Josh: varubin chips come online at scale what on earth are these models going to look
1:29
Josh: like if we already have fable and gpt6 class models it's going to be pretty
1:33
Josh: wild for some numbers 10x,
1:35
Josh: the efficiency which is crazy so for every single megawatt you put into a gpu
1:41
Josh: it will give you 10 times the amount of tokens this is a huge unlock for a lot of models.
1:47
Josh: The second thing is in terms of density of transistors, there's 336 billion.
1:53
Josh: That is a 62% increase over the Blackwell GB300. And this is on TSMC's three
1:58
Josh: nanometer technology, which is basically the cutting edge.
2:02
Josh: There's 22 terabytes per second of memory that is going through this whole thing.
2:07
Josh: And basically, they ran out of room for a single piece of silicon and they glued
2:11
Josh: these two maxed out dyes together and call it one.
2:13
Josh: So that's kind of where we are. is like this chip is going to be 10x more performant
2:17
Josh: per watt and it's going to have a just unbelievable baseline relative to the
2:22
Josh: gb300 a lot of people are saying about four times the baseline so
2:26
Josh: imagine what we get when these models are trained on not only four times the
2:31
Josh: baseline but also efficiency improvements in terms of algorithms like
2:34
Josh: The next generation of software built on these things is going to be a monster.
2:40
Ejaaz: And I want to translate what this means for the wider market and for NVIDIA
2:44
Ejaaz: stock, which has pretty much just been flat for the last like five months.
2:48
Ejaaz: I think this is NVIDIA's star moment for this year.
2:52
Ejaaz: Now, the reason is it's really costly to train and inference models these days.
2:58
Ejaaz: And so any way that a company, an enterprise that is spending 10 to hundreds
3:01
Ejaaz: of millions of dollars every year can save money is a big deal.
3:05
Ejaaz: Now, a big way that they can save money is infrastructure. Right now,
3:09
Ejaaz: people are paying around 50% to 60% profit margin on every dollar spent on an AI token to NVIDIA.
3:15
Ejaaz: They are like monopolies in this way. So, NVIDIA giving them a 10x efficiency
3:20
Ejaaz: increase means that you can create or run your Fable 5 model,
3:24
Ejaaz: your local open source model, Kimi K3, at one-tenth of the cost purely because of infrastructure.
3:30
Ejaaz: Now, the way that this GPU works, you can't just run one Vibra Rubin on its own.
3:35
Ejaaz: You need to run 72 of them in one server rack.
3:38
Ejaaz: And they're serviced by 32 of NVIDIA's CPUs. Now, the reason why they have so
3:43
Ejaaz: many CPUs is because when you're running these AI models these days,
3:46
Ejaaz: you're not really running one instance of an AI model. You're running many.
3:49
Ejaaz: It's called AI agents, and they need to access different tools.
3:52
Ejaaz: That's what the CPUs are for. So it's this collective thing with the software
3:55
Ejaaz: and hardware integration, the way it's set up, that makes NVIDIA's GPUs so, so effective.
4:00
Ejaaz: And so if I had to make a call here, what is their most direct competitor, Josh?
4:04
Ejaaz: It has to be Google. Google has tried to go for the NVIDIA throne with their
4:08
Ejaaz: own custom-made GPUs called TPUs, their Tensor Processing Units,
4:12
Ejaaz: and they had their quarterly earnings results
4:14
Ejaaz: Yesterday, and they made a significant chunk of money on TPUs,
4:17
Ejaaz: but they also announced that they're gonna be releasing a new chip,
4:20
Ejaaz: and I just don't think that they can catch up to NVIDIA.
4:22
Ejaaz: So NVIDIA's running away with it, and this is a huge advancement for AI models in general.
4:27
Josh: And Viti's got a hell of a lead. And one of the headlines that I do want to
4:29
Josh: finish this segment on is the amount that they're actually able to produce of
4:33
Josh: this. Because we have a lot of people that are interested in the investment
4:36
Josh: angle. We have an interesting investment angle here.
4:38
Josh: And the number is a thousand. A thousand of these racks per day.
4:43
Josh: And each, as like you mentioned, each rack contains a lot of these chips inside of it.
4:48
Josh: And that means that they are projected to make $630 billion per day.
4:54
Josh: Per quarter assuming they can get this up throw that.
4:57
Ejaaz: Into your earnings per share forecast
4:59
Josh: Whatever whatever forecasting you're doing for nvidia like perhaps think higher
5:03
Josh: that is a tremendous amount of money for a single skew like this is one of the
5:08
Josh: many products lines it's important to note also that vera rubin has cpus and
5:12
Josh: the cpus have been shipping since last month so there's
5:14
Josh: a huge business being built on this new very i think.
5:16
Ejaaz: They're on track to make like what 20 billion dollars this year just from cpus alone
5:21
Josh: A gazillion at this point i don't know Nvidia, just like there is no end in
5:24
Josh: sight to the amount of GPUs that they're going to be able to sell to people.
5:27
Josh: And this number I found pretty large. Now, you mentioned Google,
5:31
Josh: and I feel like we do have to talk about Google because they just had their
5:35
Josh: earnings report as well as announcing a lot of things.
5:38
Josh: And to preface this section, we were possibly the number one fanboys on the
5:41
Josh: internet of Google and Gemini and DeepMind.
5:44
Josh: And we were talking about them almost every day. I don't think we've had an
5:47
Josh: episode on them in the last, I don't know, two, three months.
5:50
Josh: It's been kind of rough. So are things getting better?
5:53
Ejaaz: Okay, so Google had a big week of announcements. Unfortunately,
5:57
Ejaaz: 100% of those announcements kind of sucked.
6:01
Ejaaz: So the first major announcement is they released a series or rather their new
6:05
Ejaaz: flagship AI model. It's called Gemini 3.6 Flash. This is an iteration on Gemini 3.5.
6:12
Ejaaz: And unfortunately, it's not that very good. Now, they advertise it as not being
6:17
Ejaaz: frontier, but being cheap, affordable, and cost efficient. Now,
6:21
Ejaaz: if you're saying those words, it better be worth the cost that I'm paying for
6:25
Ejaaz: it for the intelligence that I receive.
6:27
Ejaaz: Fortunately, it's dumber than ChatGPT 5.6 Luna, which is their worst model of
6:33
Ejaaz: their frontier labs that they release.
6:35
Ejaaz: And it gets worse, Josh. It's worse than GLM-5. It has 5.2, not even 5.3, GLM-5.2 and Kimi-K3.
6:43
Ejaaz: The open source models from China that have a fraction of the budget that Google,
6:47
Ejaaz: remember, Google is spending $200 billion this year.
6:52
Ejaaz: These Chinese AI labs are spending one hundredth of that and they're able to
6:57
Ejaaz: pull off a better model and they're open sourcing it for everyone to run at
6:59
Ejaaz: a much more cost-effective way.
7:01
Ejaaz: So the question I have for Google is, what earth are you doing?
7:05
Ejaaz: These models are just terrible. Sorry.
7:07
Josh: Yeah, it's disappointing to see because, you know, Google was kind of crushing
7:11
Josh: it. They were winning across the board and they were working on these world
7:14
Josh: models. Like Google was the world model company and they had all of these individual pillars.
7:18
Josh: And now when we look at this chart, I mean, they're not even in the middle.
7:22
Josh: They're at the back of the pack.
7:24
Josh: I think that's probably most embarrassing is like, dude, you're getting beaten by meta. Come on now.
7:28
Josh: And this seems to be the trend. And I get upset with myself a little bit for
7:32
Josh: like, they fooled me once and then they fooled me again,
7:35
Josh: where Google has had this period of time where like, hey, they literally invented
7:39
Josh: the transformer architecture that every single AI language model runs on,
7:42
Josh: but they weren't able to turn it into a product.
7:44
Josh: Then the CEO, the co-founder or the old co-founder, I guess,
7:48
Josh: Sergey Brin, he comes back into the office.
7:50
Josh: He whips the company into shape. He gets them right back at the frontier.
7:53
Josh: And then we're kind of going through this fall off again, where perhaps like
7:56
Josh: culturally at Google, there is this deeply rooted inability to actually convert
8:00
Josh: these ideas to compelling products
8:02
Josh: and to do so at the scale required to compete with Frontier AI Labs.
8:05
Josh: And I think that's the moment of time that we're at now where Google very much
8:08
Josh: feels like they are in crisis mode, given the things that they have released so far.
8:13
Josh: And relative to the things in the market, it's like, it's tough to find a good
8:16
Josh: thing to say about what's gone down this week. There is some promise and some hope.
8:21
Josh: In the sense that while I was reading through everything from Google this week,
8:24
Josh: they have this new chip architecture that they're trying to build.
8:28
Josh: And that seems like it could possibly give them an edge.
8:31
Josh: I mean, when we think about the GPU, CPU architecture world,
8:35
Josh: there is like general GPUs with NVIDIA, but then there's accelerators like the
8:38
Josh: TPUs and like the custom ASICs and like the etched chips that have LLM architecture baked right in.
8:44
Josh: So what is the deal with this new Google chip that seems to be competing directly with that?
8:49
Ejaaz: Yeah. So it's called or codenamed frozen v2 and it's basically an iteration
8:56
Ejaaz: of their existing chip architecture that they've built tensor processing units
9:00
Ejaaz: or tpus now if we look at their quarterly earnings which we're about to in a second
9:04
Ejaaz: they made quite a decent chunk of money selling their tpus to external customers
9:08
Ejaaz: now this is a big jump from the previous quarter because google primarily built
9:12
Ejaaz: and used tpus to train their own gemini models what they've done the change
9:15
Ejaaz: between q1 and q2 is they've started selling it to
9:19
Ejaaz: other vendors, such as Anthropic and other AI labs to train their models to
9:24
Ejaaz: create a new revenue business line for them.
9:26
Ejaaz: Now, frozen chip v2 is meant to be six to eight times more efficient than TPUs.
9:32
Ejaaz: And if you remember, the original pitch of TPUs was more efficient training
9:36
Ejaaz: and inference for your AI model. So they seem to have revived a new chip architecture
9:41
Ejaaz: and they want to build it.
9:42
Ejaaz: Now, the bad news is this thing isn't coming online until 2028.
9:46
Ejaaz: Vera Rubin that we just discussed came online yesterday.
9:50
Ejaaz: And it has 10 times more efficiency than the last predecessor of NVIDIA,
9:54
Ejaaz: which the TPUs couldn't keep up with.
9:56
Ejaaz: So my question to you is, even if they achieve six to eight times more efficiency
10:00
Ejaaz: in two and a half years time, who the hell cares?
10:03
Ejaaz: Because NVIDIA will have a better chip architecture by then that is massively more efficient.
10:08
Ejaaz: So I don't see a way that Google capitalizes or keeps up or catches up,
10:13
Ejaaz: rather, with NVIDIA on the chip side of things.
10:16
Ejaaz: Now, just briefly on the model side of things, I can't emphasize enough how
10:22
Ejaaz: bad it is that Google hasn't delivered a frontier model in the last,
10:25
Ejaaz: what is it, four months, right?
10:28
Ejaaz: They have all the data in the world. Gmail, Search, Browser, Android, G Suite.
10:36
Ejaaz: They can use that data to create an amazing model. Sucks.
10:39
Ejaaz: Coding models, where they should have diverted all their compute to,
10:42
Ejaaz: absolutely failed. What were they doing instead?
10:44
Ejaaz: Oh, let's just sell all our compute to other vendors. Okay, but what if they
10:47
Ejaaz: weren't selling their compute to other vendors? What were they doing?
10:50
Ejaaz: They were distributing it amongst teams.
10:52
Ejaaz: Teams literally had to fight internally to get compute. DeepMind had to fight for the compute.
10:57
Ejaaz: Now, if you compare that to Anthropic, if you compare that to OpenAI,
11:00
Ejaaz: it is a completely different philosophy. It's like, give the researchers the
11:03
Ejaaz: compute because they're going to build a better model and that model eventually,
11:06
Ejaaz: down the line, will bring every other business unit up.
11:09
Ejaaz: And Google fumbled that back. A lot of critics are going to respond to this
11:13
Ejaaz: and say, well, Google had an amazing revenue quarter. Yeah, for now.
11:17
Ejaaz: Like, what does that look like in like five months from now?
11:20
Ejaaz: And there's a reason why Google stock is down 5% as we're recording this.
11:23
Ejaaz: Because even though they had an amazing record quarter, no one cares.
11:26
Josh: Yeah, I think that's the idea is a lot of earnings now, although they are making
11:30
Josh: a lot of money, they are up. Oh, I think revenue is up 24% year over year.
11:33
Josh: They're still growing fairly quickly, but no one really cares about growth.
11:37
Josh: That is expected. There is this AI golden era where if you are serving any sort
11:40
Josh: of tokens, you are making money on them with a fairly large margin.
11:44
Josh: The thing people are interested in is that looking forward, like what is the forward look?
11:47
Josh: What is their chance of capturing this new wave of AI token economics?
11:52
Josh: And their probabilities continue to go down. And we see this in the stock price.
11:56
Josh: Like, look at this chart. This is so ugly. It's so bad. It's really tough to see.
12:00
Josh: And I think it's because like you read the numbers now, you're like,
12:02
Josh: okay, great. Google search totally
12:04
Josh: isn't dead. It's actually doing much better than it ever was before.
12:06
Josh: I think what it's like up 17% on the year or something like that, which is really great.
12:11
Josh: But when you look forward, like,
12:13
Josh: what is the odds that Google is at the frontier at 12 months from now?
12:17
Josh: I think that number is lower and it's lower than it has been before.
12:20
Josh: And that is discouraging to investors. So I'm wishing Google all the best.
12:23
Josh: I really am. I freaking love this company and I want them to win so bad.
12:27
Josh: It's just, it's difficult to get excited when there's no tools that you actually
12:31
Josh: want to use because everything else that's on the market is so far superior.
12:35
Ejaaz: So moving on, a big debate this week, Josh, we'll cover this in a few episodes
12:39
Ejaaz: this week, is the China versus USA, open source versus closed source AI debate.
12:45
Ejaaz: Now, for those of you who are unfamiliar, definitely go check out some of our previous episodes.
12:48
Ejaaz: But basically, China released Kimi K3 and soon to be a series of open source
12:54
Ejaaz: models that are 90% capable of the frontier models that the US have made.
12:59
Ejaaz: So we're talking about Fable 5, GPT 5.6, but at a fraction of the cost,
13:04
Ejaaz: and you can host it on your own server, which means that it could be even cheaper.
13:07
Ejaaz: And so the question all these companies are asking, similarly to the new Vera Rubin update is,
13:13
Ejaaz: Why should I pay all this money to access Fable or GPT 5.6 when I can just spend
13:18
Ejaaz: a lot less? And yeah, maybe the Chinese models are a little slower,
13:21
Ejaaz: but I get to own my data and I need to, and I basically save a ton of money. So why would I do this?
13:27
Ejaaz: And so the US has responded and said, hmm, these Chinese labs,
13:31
Ejaaz: I think have distilled a lot of proprietary information from US American labs.
13:35
Ejaaz: And so we are now considering banning open source models. Now,
13:39
Ejaaz: I just want to repeat that for a second.
13:41
Ejaaz: The capitalist freedom country is considering potentially banning open source models.
13:47
Ejaaz: And meanwhile, the communist country is like, hey, open source is a strategy
13:50
Ejaaz: going forwards. We had a comment from Liang Wenfeng from DeepSeek.
13:54
Ejaaz: He said, we're going to remain open source for as long as we can,
13:56
Ejaaz: because that is what we believe in.
13:58
Ejaaz: So it's just weird dichotomy. And if US does end up banning open source,
14:03
Ejaaz: it's going to mess up a lot of young startups in the US who are currently relying
14:07
Ejaaz: on open source models to build their products because they can't afford the
14:10
Ejaaz: expensive American Frontier Labs.
14:13
Josh: Yeah, this doesn't make any sense to me at all. I don't think like I'm not even
14:16
Josh: sure this is going to be possible. It seems to me like this is posturing for
14:19
Josh: probably setting up some sort of legislative plan or some sort of form of action.
14:25
Josh: It's not actually going to happen because when you think about what it means
14:28
Josh: to ban these Chinese models, this open source code, I mean, ignore the idea
14:32
Josh: that they're Chinese. Just think of it as open source code. This is just binary numbers in a page.
14:37
Josh: This is a near impossibility in this world to ban when everyone is actively
14:41
Josh: seeking it. I mean, you think about internal secrets at Frontier AI Labs and
14:45
Josh: how quickly those leak out.
14:47
Josh: If this is out in the public, there's no way that people aren't going to clone it and run it locally.
14:52
Josh: And who's going to stop them from doing that? It's like this is this is not really a practicality.
14:56
Josh: And when you think about I mean, the country as a whole, just from an ethical
14:59
Josh: standpoint, open source code very much is reminiscent of open and free speech.
15:04
Josh: And just like banning that seems like a slippery slope that is very problematic.
15:10
Josh: So I am hopeful and feeling fairly certain that this is posturing for something larger.
15:17
Josh: Just kind of signaling an intention versus actually taking action on it.
15:20
Josh: Because I mean, I would hope that they are, you know, reasonable enough to recognize
15:24
Josh: that banning Chinese open source models is just never going to happen.
15:28
Josh: I mean, if the code is online, we are going to get it, we are going to use it,
15:31
Josh: we are going to run it locally.
15:33
Josh: The fact that it runs locally is even more against it, because it's easier to
15:36
Josh: just take it on a computer and run it air gapped, and no one can ever discover it.
15:39
Josh: So I think it's it seems like we'll see, we'll keep monitoring this one and see like what the larger
15:45
Josh: idea behind this is but i don't think it's actually banning because that just
15:49
Josh: goes against so many ethical things that we've set up so much of like just practical
15:53
Josh: um the way this works it's just it doesn't make sense.
15:56
Ejaaz: I think what's happening at the core of all of this is there's a big fear in
16:00
Ejaaz: the u.s that china is catching up and
16:02
Ejaaz: originally it was like china's three years behind both in model development
16:06
Ejaaz: and chip architecture and you need both of those things you need compute but
16:08
Ejaaz: china has a lot of compute in order to build frontier models.
16:12
Ejaaz: Recently, they've caught up a lot quicker than we expected. The models are getting better.
16:16
Ejaaz: They're owning their own chip architecture. In Xi Jinping's speech last week,
16:20
Ejaaz: There was also a co-announcement from Huawei, which basically said,
16:24
Ejaaz: hey, we have this new chip rack server, and every single Chinese frontier lab is going to use it.
16:28
Ejaaz: And then followed by that, we have Chipu, one of their biggest AI labs,
16:31
Ejaaz: announced that they've just completed the construction of a gigantic data center
16:34
Ejaaz: that is only going to run Chinese-made chips.
16:37
Ejaaz: No NVIDIA chips in sight. Now, if you rewind like three months ago,
16:41
Ejaaz: NVIDIA, Jensen really wanted to sell China chips because they will then rely
16:45
Ejaaz: on American frontier chip architecture and therefore we could kind of like regulate
16:49
Ejaaz: and keep an eye on these things.
16:50
Ejaaz: That is no more. So I can understand why the Trump cabinet is getting a bit
16:54
Ejaaz: worried and I'm interested to see kind of what process they take going forwards. Yeah.
16:58
Ejaaz: Moving on, Josh, there is a trend that is developing this week,
17:03
Ejaaz: a weird one that not a lot of people expected.
17:06
Ejaaz: Over the last three days, three different companies, three major companies,
17:10
Ejaaz: announced that they're creating what's known as a model routing platform.
17:14
Ejaaz: Now, if you want to know what that is, when you send a prompt,
17:17
Ejaaz: typically you're using a ChatGPT subscription. So it goes to ChatGPT or using
17:20
Ejaaz: a Claude subscription. It goes to Claude.
17:22
Ejaaz: A model routing platform takes your prompt and decides which parts of your prompt
17:27
Ejaaz: to send to different models for different types of function.
17:30
Ejaaz: And the reason why they do this is, one, to give you a better answer output
17:34
Ejaaz: than sending it to a single model, and two, to save you a heck ton of money.
17:37
Ejaaz: That's what Cursor did for coding, and it's what led to them getting it acquired
17:41
Ejaaz: for $60 billion. And Cursor themselves is coming out with a new platform called
17:45
Ejaaz: Cursor Router, which is a more generalized platform for any kind of LLM prompt.
17:49
Ejaaz: The other two companies that also announced that they're building a similar
17:52
Ejaaz: product is Meta and Ramp. Ramp is in the finance game.
17:56
Ejaaz: And Meta, as you know, looked at OpenRouter, looked at Ramp and was like,
18:00
Ejaaz: I need to do this internally because I'm spending tens of billions of dollars
18:03
Ejaaz: every single year to build a better AI model. Why don't I just create a cheaper,
18:07
Ejaaz: effective routing platform? It's interesting to see.
18:10
Josh: Yeah, the interesting thing, I mean, Cursor is claiming frontier quality at 60% lower cost.
18:14
Josh: And if that is true, that seems like a pretty compelling argument for a lot
18:18
Josh: of labs to come and use this or for a lot of users to come and use this.
18:21
Josh: Now, I find it interesting that we're seeing this from Cursor,
18:24
Josh: from Ramp, from Meta, and we're not seeing this from OpenAI,
18:27
Josh: from Anthropic, from Google, because they very clearly have this suite of models
18:30
Josh: that would imply that they have the ability to do this.
18:32
Josh: I mean, we have like the Sol Luna Terra from OpenAI and GPT,
18:37
Josh: where it's like very clear that they would benefit from a router.
18:40
Josh: So you have to imagine they're kind of working on this. And I wonder if this
18:43
Josh: is an instance, kind of like the early Internet era, where...
18:46
Josh: Your feature or your company becomes a feature and if cursor router is not able
18:52
Josh: to kind of sustain a moat of an audience
18:54
Josh: what's stopping chat gpt and open ai from launching their own and doing this
18:58
Josh: own router with their own tokens and as they get more efficient they're able to ride it
19:01
Josh: more more effectively and they have more data than any other company because
19:05
Josh: they have all the queries that they can just
19:07
Josh: train off of and imply this so it's exciting to see this now and i think it's
19:11
Josh: incredibly important for companies who are trying to save money to use this
19:14
Josh: type of technology but man it's going to be tough to compete i think it's like well,
19:20
Josh: you're kind of going like you're going up against the big guys here and like
19:23
Josh: maybe this works for now but i mean you're a one feature away from if people
19:27
Josh: just not really caring anymore.
19:28
Ejaaz: I think so i think so um okay i think to round up the docket today josh something's going on with
19:35
Josh: Oh this is so sick wait yes can we talk about tesla yeah.
19:38
Ejaaz: What's going on tell me what's going on
19:39
Josh: Oh my god okay so first starlink v5 if you've used Starlink before,
19:43
Josh: this is like a big satellite dish, sits on your roof, can get you connection
19:46
Josh: anywhere in the world. If you've ever traveled to a country that isn't very
19:49
Josh: well industrialized, you've seen these on roofs everywhere.
19:52
Josh: It's how everyone gets their internet. This new version, Starlink V5, is about 40% smaller.
19:56
Josh: It is significantly more efficient in terms of how much you can get per watt.
20:01
Josh: And more importantly, it is being integrated into every single cyber cab being
20:05
Josh: produced ever, which is unbelievable because now there's this super cluster
20:09
Josh: of computers that can run anywhere in the world that will be connected to,
20:14
Josh: internet. So if there's ever a problem, if the car ever needs help from anyone,
20:18
Josh: if it ever needs to share data back to the mainframe, it's able to do that using
20:22
Josh: this integrated Starlink antenna.
20:24
Josh: And we're entering this world between this and the Starlink mobile satellites
20:27
Josh: that are coming with Starlink V3 on Starship, where you will never go anywhere
20:32
Josh: on planet Earth and not have connection to the internet.
20:35
Josh: And that to me seems like so cool and really powerful. And when you think about
20:39
Josh: how much compute is going on these cybercabs, training cluster,
20:42
Josh: I don't know, just saying.
20:44
Ejaaz: So my biggest bull case for Tesla has got nothing to do with cars and honestly
20:49
Ejaaz: nothing to do with robots.
20:50
Ejaaz: It's to do with this decentralized network that he's building that he can lean
20:54
Ejaaz: on to train models, to inference prompts, and just to kind of like power this new AI economy.
21:00
Ejaaz: If that sounds like a lot of jargon, it's because this is new.
21:03
Ejaaz: This is net new and we haven't ever seen anything like this.
21:05
Ejaaz: But Elon, who is the master of hyperscaling any kind of infrastructure target
21:09
Ejaaz: that he sets his eyes on is the guy to do this. And I think Tesla,
21:13
Ejaaz: It's the largest network of robots. If you want to argue like what a robot is
21:16
Ejaaz: and whether robots are roaming around today in the world, Tesla is a prime example of that.
21:20
Ejaaz: And so I'm excited to see this happen. I'm so, so pumped.
21:23
Ejaaz: Sorry, I know this is a layman thing to say, but to just have internet wherever
21:26
Ejaaz: I go, I'm tired of getting on like a Delta flight and having crappy internet.
21:32
Ejaaz: Just allow me to stream, allow me to watch videos, allow me to call my friends.
21:35
Ejaaz: That would be amazing. So I'm excited to see this come out.
21:38
Ejaaz: But I believe that is the end of the docket for this week. That's it.
21:41
Ejaaz: Yeah, if you're listening to this, You are all officially caught up.
21:45
Ejaaz: I've heard rumors that there's some potentially new features and models coming
21:48
Ejaaz: out. You'll hear about all of that next week, I think on Tuesday.
21:52
Ejaaz: But aside from that, if you're listening to us on YouTube, please subscribe.
21:55
Ejaaz: Please leave us a comment. Give us a rating.
21:57
Ejaaz: Turn on notifications. It helps us out massively. If you're listening to us
22:00
Ejaaz: on Spotify or Apple, leave us a comment. Give us a rating. It helps us out massively.
22:04
Ejaaz: And I think that's it, Josh. Is there anything else?
22:07
Josh: That's everything. Thank you so much for joining us for another amazing week
22:10
Josh: of AI Frontier Technology.
22:12
Josh: Next week buckle up the rumor mill's been going crazy i think by then we're
22:15
Josh: going to have a lot of news so i'm excited to talk about it here on the show
22:18
Josh: don't miss it we'll see you guys next week have a great weekend see you guys.