@@powerpig99 very curious as to why you think this. Do you think this is because he would be more willing to have on a larger range of guests and thus the converstion will be less focused on his usual topics of math, physics, CS, AI, etc? would love to know why!
Great talk on Dojo. LEX I just saw your over a million subscriptions congratulations. I've been listening to you for a number of years. You make us smarter, thanks
I've already watched the whole episode a I can't wait for part 3! Thank you again Lex for yet another wonderful exchange. It is a newly furnished dream of mine to share a yoga practice space with you sometime and share what I've come to learn about the world through breath and movement.
George is wrong here. Nvidia is gaining the mileage with the revenues as long as they can. However, Nvidia is using this revenue to invest in software. A full system software stack is being developed in different domains so that application developers in different fields focus on applications. This is very different from the Intel business model where software was never a source of revenue. But Nvidia is shaping up also as a software company which makes revenues by selling their hardware. That is a so unique position that it will go long long way. Even if there are knock-offs in the future, the software stack will keep Nvidia ahead of everyone. Nvidia does not need to be Intel. Nvidia will be Nvidia.
Nvidia understands that this technology will commoditize pretty soon. There is much less of vendor lock-in over the middle term then in case of the ISA architecture as with intel. They are thus doing the right thing from the business perspective. Milk it as long as you can.
I like Jeff Bezos' stated strategy of selling at the smallest margin possible to dissuade competition and to sell more units in the process. Any company putting huge margins will both sell less and then open itself up to a competitor coming in and making something approximately as good that topples you
The margin is not the issue I think because a healthy margin is good for business it allows the business to cover operating costs and invest in innovating to make the product better. Amazon can only operate like that because it is super rich, they can operate at a loss as long as they want.
@@tonytanner3048 they take the loss as they can avoid paying certain taxes as they are an online seller and a marketplace for third-party businesses. If they get hit with a tax bill they always pass it on to the third party sellers or the customer. www.forbes.com/sites/siladityaray/2020/10/14/amazon-wont-have-to-pay-uk-digital-tax-that-will-impact-third-party-sellers/
@@tonytanner3048 amazon employed that strategy even prior to the dot com bubble when they were losing money hand over fist. it has always been one of Bezos' principles. if you have a decent margin someone will come along and say they are happy to take half as much margin.
@@colstoun4762 This is a exploit in our financial system, amazon does not even sell at a margin they sell at a loss which is against free market. I am not saying a small or a big margin is the right margin but your margin should be enough to allow for operating costs and innovation. If you spend money on innovation your costs will eventually go down as efficiency increases.
@@davefroman4700 Tesla does not manufacture chips so how could they be ahead of them? Can I ask do you suffer from Tesla fanboy syndrome. Tesla does make great luxury cars though. I do assume you mean tesla is ahead in terms of ai?
ARM is an interesting approach but still has a long way to go in the market for desktops, servers and laptops. I suspect that NVIDIA is aiming towards the mobile market. Intel has an extremely poor foothold in the mobile device market. NVIDIA also has a poor foothold in the desktop market.
@@bighands69 i guess you have not seen the new Apple laptops who use arm chips faster and way less watt usages than desktop intel chips. You should look the video of coldfusion about the new chip apple made ARM IS ahead now
Lex you should interview Jensen. There are some reasons why A100 is quite expensive, including a massive amount of HBM2 and the chip is quite large so I expect the yields to be low.
1080 TI along with 1080 is still a very competent video card despite it's age...frame rates for gaming are still well above respectable rates..couple the video card with a 144 monitor and enjoy the fun.
I agreed. Tesla should sell their FSD chip to other companies so that R&D expenses will be even out for the company balance sheets. Tesla is already finished with the FSD 2.0 ship the end of this year which will be much batter so why waste older technology where you can profit from it 👍.
Tesla makes more money than they can spend intelligently. Their cash reserves keep growing even as they increase spending on growth. They are making 30% margins per car sold(including development) where combustion vehicles get negative or a few % margins on the initial sale.
RJ 8U way too premature. They should consider it or even spin out the chip unit once there are true competitors but right now the strategic importance is just too significant.
Yep Intel thought they had it all wrapped up and in the bag and were also price gouging. Then along came ryzen chips at 1/4 the price. This will happen to Nvidea but if they burn their bridges by keeping the prices high they will be done and everyone will cheer instead.
Kinda interesting that this is before the whole explosion of locally run AI applications like Stable Diffusion for images and Oobagooba for chat LLMs, driving demand way up for GPUs with 8GB to 24GB, which also happen to be consumer products many people already own, can find at Best Buy, or even buy used under $70 ( 12GB Tesla M40). But a lot of AI chip makers hadn't really anticipated the expectation for running AI models well over 1GB. Their hardware will no doubt outperform GPUs in both teraflops and memory bandwidth (equally important for AI performance) but architecture does not "scale up" like GPUs can. It seems like both an exciting and a tough time to be in the AI business.
I don’t know how reliable this is but I heard someone apparently “in the know” say that an IT hardware developer working for a major Japanese automaker had a look at the Tesla computer boards and said they couldn’t build that right now and it would take them a while to work it out. And as is well known, Tesla doesn’t stand still with their R&D, so by the time others catch up to where they are now Tesla will be further ahead again. That’s the word a few able to glimpse the tech are saying anyway. Whether it’s true or not I’m not going to pretend that I have any idea.
A car company can't make it because a car company doesn't make electronics. I think you are talking about supply chain issues and not specifically about the tesla chip. I have heard this story too. But it's about tesla tech in general and car makers not being able to shift their supply chain fast enough. (Multi year existing contract... literally 10s of thousands of suppliers)
You have PlaidML for AMD, which to me is currently broken... And maybe Apple Neural Engine in the future with CoreML but at the moment doesn't do training...
there's cerebras, groq and a ton of other companies
4 ปีที่แล้ว
Yes, and they are chip companies. The bread and butter is a software stack which NVIDIA put in place first. Now they are all finding out how fucked they are.
NVidia lives in a free market, there's no such thing as too high margins, the free market will always win. It just encourages competition if they charge high prices. There is no problem to see here.
Except then you've got to WAIT for that competition to come while you're getting buttfucked. And semiconductors are not an easy place to startup, that's why there's really only 2 companies designing GPUs.
Unless NVidia overhauls their architecture they won’t have a viable chip. The company to keep an eye out is MobilEye. They are taking a similar approach to Tesla.
It is not clear that why should elon sell dojo chip. It is their trade secret, and it must be a unique advantage of their autonomous driving tech. Intel model is different.
There's advantages to either route. For Tesla long-term, dominating the self-driving space is much more advantageous than making extra money selling FSD hardware. For George Hotz and self-driving community long-term, Tesla selling chips is more advantageous than not. Hence the differing points of view.
Tesla has already said they will license ALL of their tech. including the chips. Why? Because by the time the competition gets up to speed with them they will already be 6 revisions ahead of them. When are you guys going to wake up to the fact that Tesla has the deepest talent pool on earth that is stacked with NASA personel? Even Nvidia engineers are saying Tesla is 5-6 years ahead of anything they have on the drawing board.
Prove that this system is broken, has happened to intel and now nvidia is in danger, they just think about money.
4 ปีที่แล้ว
Producing silicon is a knife-edge between winning and going absolutely broke at the cost of the latest TSMC process, Intel found out the hard way. It's only going to get more expensive and NVIDIA can see this coming 10 years down the road. They are making money whilst they can to secure supremacy. With ARM/Mellanox they have exposure to everything.
I don't think anyone will bet against them now. What George implied about the change in culture from 2016 onward adds more weight to the notion that Nvidia is emulating Intel's footsteps... It would be a fantastic strategy if AMD continued to underdeliver, but they are certainly not doing that of late. Right now I would put my money on AMD emulating their success against Intel with Nvidia.
Why are people talking about Tesla doesn't need Nvidia? I mean I saw a bunch of opening job positions at Nvidia at least in Shenzhen advertise that they have a project with Tesla. Aren't Nvidia chips deployed on Tesla's cars? I'm confused now. Somebody educate me, please.
Tesla seems to want to maintain a very big lead ahead of other car companies or tech companies trying to get into autonomous vehicles. Why would they give their competitors ANY leg up whatsoever? Surely that's why they aren't selling their chips externally.
I am a capitalist through and through...HOWEVER, it does seem that companies have become SO driven by share price that the big picture gets lost. NVidia squeezing their product so hard is one example. Other examples include companies kowtowing to the demands of China. "Oh if we alienate Chine we'll lose billions!!!!!" Well...share holder value is more important than Western values in that case. It seems like absolutely NO ONE is willing (or allowed) to take a short term hit for any reason. Get that share price up TODAY! Look at the opioid crisis- nobody at Purdue pharma had the guts to ask how and why they were selling so many potent and addictive pills? I would add that this isn't something that can be solved with governmental regulation or taxation. It's about a loss of shared principles, a loss of respect for other human beings.
You mean Western values like slaughtering Indians, enslaving Africans, aggressively forcing opium into China, starving & raping Iraq, Libya, Venezuela for their oil resources, letting COVID spread like wildfire in order to save the stock markets, right? Therefore don't talk about respect for other human beings, you witless hypocrite!
@@hugohugo37 Noted that you make not the least effort to deny or defend the crimes of filthy Western capitalist imperialism, the full list of which is indeed infinite.
Nothing new, and still sticking to the same misconceptions, yet criticizing someone who has been continuously adapting from mistakes and way ahead in implementing things ahead of other’s imagination.
Why is Jensen price gouging? .... eventually the Chinese will make knock off nvidia chips....maybe that’s the answer. He is making his money while he can, and building up reserves to hold off the Chinese competition when it comes full force.
Man, It's also fault of ML frameworks developers that prioritized the use of Nvidia cuda. AMD GPUs are very fast at cranching numbers too. So why tensorflow support for AMD sucks? OpenCL implementations sucks? well go after Khronos and AMD and talk about it, don't blame only Nvidia. Start support competitors and the market will do the job and things will start have fair prices for consumers. Every Company have greedy, and this domination of Nvidia in this space is also caused by consumers that buy only from Nvidia and not the competitor even if the competitor have 1/7 of the price, why? Looks what the AMD zen cpus did to Intel? consumers start prefer ryzen cpus and Intel greedyness start to faid.
nah, not valid points, selling the battery is the exact same as selling the tpu, is helping the competition which make no sense to do. by the way everyone can make batteries as well and it's a lot easier than making tpus
Checkout this neat startup that came out of MIT. They are trying take on the ML hardware bottleneck by using light instead of electrons!! Very exciting stuff. Useful tech is probably still a a bit away tho. lightmatter.co/about.html
I enjoyed hearing this guy with a questionable record of building and running businesses telling us what the CEOs of 12-figure tech giants should do....
@@_PatrickO OpenPilot is a product, not a business. Comma.ai as a business is worth maybe 60-70M based on their last (pretty small) funding round a few years ago, which, if you're the best in class, in a multibillion dollar industry, thats not good! Hotz also isn't even CEO of Comma, and its not even clear Comma HAS a CEO right now!
You are absolutely right about your point. But sometimes people care about things other than how to make their company the most amount of money. He explicitly points out that Jensen is a founder. He can have an ideology, he doesn't necessarily have to run after only money.
Google is like that girl that looks really good from a distance but ugly close up.
4 ปีที่แล้ว
NVIDIA is making hay whilst the sun shines. Loads of AI chip startups targeting them, silicon only getting more expensive with TSMC nodes. They are hedging now with ARM and Mellanox acquisitions
Lex Fridman is shaping up to be the Joe Rogan of the programming world. I mean that in the best way possible!
Bro he’s joe Rogan of the intellectual bros
Als long he does not fuck up his studio I am all for it :)
LOL
I actually feel the popular he gets, the less intellectually stimulating his contents become. Sort of inevitable though.
@@powerpig99 very curious as to why you think this. Do you think this is because he would be more willing to have on a larger range of guests and thus the converstion will be less focused on his usual topics of math, physics, CS, AI, etc? would love to know why!
Man, Hotz is a great interview.
He is interesting and entertaining.
great talk. George's comments on Tesla, NVidia, Google and Intel makes a lot of sense. thanks.
Great talk on Dojo. LEX I just saw your over a million subscriptions congratulations. I've been listening to you for a number of years. You make us smarter, thanks
6:28 Nailed it!!!!!!!!!!
I've already watched the whole episode a I can't wait for part 3! Thank you again Lex for yet another wonderful exchange. It is a newly furnished dream of mine to share a yoga practice space with you sometime and share what I've come to learn about the world through breath and movement.
George is wrong here. Nvidia is gaining the mileage with the revenues as long as they can. However, Nvidia is using this revenue to invest in software. A full system software stack is being developed in different domains so that application developers in different fields focus on applications. This is very different from the Intel business model where software was never a source of revenue. But Nvidia is shaping up also as a software company which makes revenues by selling their hardware. That is a so unique position that it will go long long way. Even if there are knock-offs in the future, the software stack will keep Nvidia ahead of everyone.
Nvidia does not need to be Intel. Nvidia will be Nvidia.
Nvidia understands that this technology will commoditize pretty soon. There is much less of vendor lock-in over the middle term then in case of the ISA architecture as with intel. They are thus doing the right thing from the business perspective. Milk it as long as you can.
I like Jeff Bezos' stated strategy of selling at the smallest margin possible to dissuade competition and to sell more units in the process. Any company putting huge margins will both sell less and then open itself up to a competitor coming in and making something approximately as good that topples you
The margin is not the issue I think because a healthy margin is good for business it allows the business to cover operating costs and invest in innovating to make the product better. Amazon can only operate like that because it is super rich, they can operate at a loss as long as they want.
@@tonytanner3048 they take the loss as they can avoid paying certain taxes as they are an online seller and a marketplace for third-party businesses.
If they get hit with a tax bill they always pass it on to the third party sellers or the customer.
www.forbes.com/sites/siladityaray/2020/10/14/amazon-wont-have-to-pay-uk-digital-tax-that-will-impact-third-party-sellers/
The east is coming for them
@@tonytanner3048 amazon employed that strategy even prior to the dot com bubble when they were losing money hand over fist. it has always been one of Bezos' principles. if you have a decent margin someone will come along and say they are happy to take half as much margin.
@@colstoun4762 This is a exploit in our financial system, amazon does not even sell at a margin they sell at a loss which is against free market. I am not saying a small or a big margin is the right margin but your margin should be enough to allow for operating costs and innovation. If you spend money on innovation your costs will eventually go down as efficiency increases.
3:45 well, NVIDIA bought ARM I think they positioning themselves for this.
I thought they signalled to buy ARM, didn't think the deal actually went through.
Think again. They already admitted Tesla is 5-6 years ahead of them. And Tesla does not sit still and say "Thats good enough"
@@davefroman4700
Tesla does not manufacture chips so how could they be ahead of them?
Can I ask do you suffer from Tesla fanboy syndrome.
Tesla does make great luxury cars though. I do assume you mean tesla is ahead in terms of ai?
ARM is an interesting approach but still has a long way to go in the market for desktops, servers and laptops.
I suspect that NVIDIA is aiming towards the mobile market. Intel has an extremely poor foothold in the mobile device market.
NVIDIA also has a poor foothold in the desktop market.
@@bighands69 i guess you have not seen the new Apple laptops who use arm chips faster and way less watt usages than desktop intel chips. You should look the video of coldfusion about the new chip apple made ARM IS ahead now
these clips are the best.
Lex you should interview Jensen. There are some reasons why A100 is quite expensive, including a massive amount of HBM2 and the chip is quite large so I expect the yields to be low.
it's 826mm^2. an A6000 is 628mm^2 at half the cost and 48GB GDDR6. am not sure that the 80gb HBM2 deserves a 2x markup.
1080 TI along with 1080 is still a very competent video card despite it's age...frame rates for gaming are still well above respectable rates..couple the video card with a 144 monitor and enjoy the fun.
Now I feel old
I agreed. Tesla should sell their FSD chip to other companies so that R&D expenses will be even out for the company balance sheets. Tesla is already finished with the FSD 2.0 ship the end of this year which will be much batter so why waste older technology where you can profit from it 👍.
Tesla makes more money than they can spend intelligently. Their cash reserves keep growing even as they increase spending on growth. They are making 30% margins per car sold(including development) where combustion vehicles get negative or a few % margins on the initial sale.
RJ 8U way too premature. They should consider it or even spin out the chip unit once there are true competitors but right now the strategic importance is just too significant.
Lex should do some Country songs about the coming Singularity. He has the voice for it.
Yep Intel thought they had it all wrapped up and in the bag and were also price gouging. Then along came ryzen chips at 1/4 the price.
This will happen to Nvidea but if they burn their bridges by keeping the prices high they will be done and everyone will cheer instead.
AMD for the win!
Yesterday Nvidia crossed 2 trillion dollar market cap and their share price increased more than 400% between Jan 2023 to Feb 2024
Kinda interesting that this is before the whole explosion of locally run AI applications like Stable Diffusion for images and Oobagooba for chat LLMs, driving demand way up for GPUs with 8GB to 24GB, which also happen to be consumer products many people already own, can find at Best Buy, or even buy used under $70 ( 12GB Tesla M40).
But a lot of AI chip makers hadn't really anticipated the expectation for running AI models well over 1GB. Their hardware will no doubt outperform GPUs in both teraflops and memory bandwidth (equally important for AI performance) but architecture does not "scale up" like GPUs can. It seems like both an exciting and a tough time to be in the AI business.
I don’t know how reliable this is but I heard someone apparently “in the know” say that an IT hardware developer working for a major Japanese automaker had a look at the Tesla computer boards and said they couldn’t build that right now and it would take them a while to work it out. And as is well known, Tesla doesn’t stand still with their R&D, so by the time others catch up to where they are now Tesla will be further ahead again. That’s the word a few able to glimpse the tech are saying anyway. Whether it’s true or not I’m not going to pretend that I have any idea.
A car company can't make it because a car company doesn't make electronics. I think you are talking about supply chain issues and not specifically about the tesla chip. I have heard this story too. But it's about tesla tech in general and car makers not being able to shift their supply chain fast enough. (Multi year existing contract... literally 10s of thousands of suppliers)
amd is always there, waiting for us to enter, waiting to enter us
Production tip: Never use a swivel chair in video.
Tbh I didnt even notice until you mentioned it. Not sure how many people did and if they were bothered by it.
You have PlaidML for AMD, which to me is currently broken... And maybe Apple Neural Engine in the future with CoreML but at the moment doesn't do training...
MLIR maybe is the best shot to support not only AMD gpus but also fpga.
there's cerebras, groq and a ton of other companies
Yes, and they are chip companies. The bread and butter is a software stack which NVIDIA put in place first. Now they are all finding out how fucked they are.
NVidia lives in a free market, there's no such thing as too high margins, the free market will always win. It just encourages competition if they charge high prices. There is no problem to see here.
Except then you've got to WAIT for that competition to come while you're getting buttfucked. And semiconductors are not an easy place to startup, that's why there's really only 2 companies designing GPUs.
thank you
love your shows, it's sad that everyone just expect china will rip off, it's a real factor
Unless NVidia overhauls their architecture they won’t have a viable chip. The company to keep an eye out is MobilEye. They are taking a similar approach to Tesla.
It is not clear that why should elon sell dojo chip. It is their trade secret, and it must be a unique advantage of their autonomous driving tech. Intel model is different.
There's advantages to either route. For Tesla long-term, dominating the self-driving space is much more advantageous than making extra money selling FSD hardware. For George Hotz and self-driving community long-term, Tesla selling chips is more advantageous than not. Hence the differing points of view.
Tesla has already said they will license ALL of their tech. including the chips. Why? Because by the time the competition gets up to speed with them they will already be 6 revisions ahead of them. When are you guys going to wake up to the fact that Tesla has the deepest talent pool on earth that is stacked with NASA personel? Even Nvidia engineers are saying Tesla is 5-6 years ahead of anything they have on the drawing board.
it's hard how to process
Prove that this system is broken, has happened to intel and now nvidia is in danger, they just think about money.
Producing silicon is a knife-edge between winning and going absolutely broke at the cost of the latest TSMC process, Intel found out the hard way. It's only going to get more expensive and NVIDIA can see this coming 10 years down the road. They are making money whilst they can to secure supremacy. With ARM/Mellanox they have exposure to everything.
AMD & Xilinx is going to rebalance the ecosystem.
I don't think anyone will bet against them now.
What George implied about the change in culture from 2016 onward adds more weight to the notion that Nvidia is emulating Intel's footsteps... It would be a fantastic strategy if AMD continued to underdeliver, but they are certainly not doing that of late.
Right now I would put my money on AMD emulating their success against Intel with Nvidia.
I guess, AMD could neutralize the system, but they don't seems going on that path way.
With zero sw , I doubt. They win against Intel by taking the Intel SW but for Nvidia there is no code to steal.
You guys need to go and look at ROCM.
I have been hearing how AMD is going to dominate for decades. Maybe someday they will get there.
can't do that I respect my community
Google TPUs also are not as programmable as GPUs.
Why are people talking about Tesla doesn't need Nvidia? I mean I saw a bunch of opening job positions at Nvidia at least in Shenzhen advertise that they have a project with Tesla. Aren't Nvidia chips deployed on Tesla's cars? I'm confused now. Somebody educate me, please.
Tesla seems to want to maintain a very big lead ahead of other car companies or tech companies trying to get into autonomous vehicles. Why would they give their competitors ANY leg up whatsoever? Surely that's why they aren't selling their chips externally.
And it's probably why they opened up their charging ports. It locks everyone into their ecosystem whilst still mainting a competitive edge.
I am a capitalist through and through...HOWEVER, it does seem that companies have become SO driven by share price that the big picture gets lost. NVidia squeezing their product so hard is one example. Other examples include companies kowtowing to the demands of China. "Oh if we alienate Chine we'll lose billions!!!!!" Well...share holder value is more important than Western values in that case. It seems like absolutely NO ONE is willing (or allowed) to take a short term hit for any reason. Get that share price up TODAY! Look at the opioid crisis- nobody at Purdue pharma had the guts to ask how and why they were selling so many potent and addictive pills? I would add that this isn't something that can be solved with governmental regulation or taxation. It's about a loss of shared principles, a loss of respect for other human beings.
You mean Western values like slaughtering Indians, enslaving Africans, aggressively forcing opium into China, starving & raping Iraq, Libya, Venezuela for their oil resources, letting COVID spread like wildfire in order to save the stock markets, right?
Therefore don't talk about respect for other human beings, you witless hypocrite!
@@foobarrel9046 Yikes.
@@hugohugo37 Noted that you make not the least effort to deny or defend the crimes of filthy Western capitalist imperialism, the full list of which is indeed infinite.
@@foobarrel9046 boo hoo
Didn't know TPU had such restrictions :/
Tesla dojo chips are NPU not TPU..
love you, Lex
0:48 Help me please I’m really close... I hope.
Nothing new, and still sticking to the same misconceptions, yet criticizing someone who has been continuously adapting from mistakes and way ahead in implementing things ahead of other’s imagination.
Who are you talking about?
The guy in the dark suit looks like he’s going to fall asleep at any moment and is thinking about what he’d like to do next.
Why is Jensen price gouging? .... eventually the Chinese will make knock off nvidia chips....maybe that’s the answer. He is making his money while he can, and building up reserves to hold off the Chinese competition when it comes full force.
Their high share price has allowed them to buy Arm
Lisa Su will take over....
Are you also here to understand why Elon said "Dojo for DOGE" ?
Perhaps AMD may step up. Lisa Su for the rescue.
Man, It's also fault of ML frameworks developers that prioritized the use of Nvidia cuda. AMD GPUs are very fast at cranching numbers too. So why tensorflow support for AMD sucks? OpenCL implementations sucks? well go after Khronos and AMD and talk about it, don't blame only Nvidia. Start support competitors and the market will do the job and things will start have fair prices for consumers. Every Company have greedy, and this domination of Nvidia in this space is also caused by consumers that buy only from Nvidia and not the competitor even if the competitor have 1/7 of the price, why? Looks what the AMD zen cpus did to Intel? consumers start prefer ryzen cpus and Intel greedyness start to faid.
Why people support Nvidia? Can we blame AMD drivers?
Brainchip’s Akida is coming to snatch their lunch.
nah, not valid points, selling the battery is the exact same as selling the tpu, is helping the competition which make no sense to do. by the way everyone can make batteries as well and it's a lot easier than making tpus
5:53 This guy is a fool "keep the simple batteries but get rid of the awesomely complex thing thats hard to copy".
Checkout this neat startup that came out of MIT. They are trying take on the ML hardware bottleneck by using light instead of electrons!! Very exciting stuff. Useful tech is probably still a a bit away tho. lightmatter.co/about.html
Tesla does it all they don't need Nvidia they don't need Google you guys don't know about it this is the best we should know this
I enjoyed hearing this guy with a questionable record of building and running businesses telling us what the CEOs of 12-figure tech giants should do....
Questionable? LOL. Go try openpilot. Its the best self driving out there until tesla cleans up their beta. Waymo and cruise are way behind.
@@_PatrickO OpenPilot is a product, not a business. Comma.ai as a business is worth maybe 60-70M based on their last (pretty small) funding round a few years ago, which, if you're the best in class, in a multibillion dollar industry, thats not good! Hotz also isn't even CEO of Comma, and its not even clear Comma HAS a CEO right now!
You are absolutely right about your point. But sometimes people care about things other than how to make their company the most amount of money. He explicitly points out that Jensen is a founder. He can have an ideology, he doesn't necessarily have to run after only money.
watching and listening to this on my old P5820 W-2123 with an 5800XT
Google is like that girl that looks really good from a distance but ugly close up.
NVIDIA is making hay whilst the sun shines. Loads of AI chip startups targeting them, silicon only getting more expensive with TSMC nodes. They are hedging now with ARM and Mellanox acquisitions
Like these people heard about nvidia for the first time in 2016. They were doing same shit since they were created.
The lady doth protest too much, methinks.
NO !
Xilinx 🤫
I watching only my podcast
This has not aged very well…
Intel arc hpg
Hotz’s critique didnt age well
Jensen is building money to buy a company
First
Who cares
The legacy automakers DONT have a ton of money, that’s why the don’t invest the money to set up EV production lines.
Google.de
shame.....its AMD
Uhm...what? What do you mean by that, could you please use more words rather than half a sentence?
Nvidia will become like intel. Dont want a Ryzen situation
This guy is as arrogant as Trevor Milton, despite his (much) bigger competence !!!
Lex stop staring out the window. Look at your guest
AMD>NVIDIA
No serious Machine Learning developer said that ever.