-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
Nvidia CEO Jensen Huang On How His Big Bet On A.I. Is Finally Paying Off - Full Interview
- Thread starter Purgatory
- Start date
Yup nvidia are going to be making a killing, people thought it was bad with the crypto mining, you ain't seen anything yet..... OpenAI/chatgpt is probably the equivalent to when the internet was first founded, seeing this right now in my work where pretty much every company is pouring millions into it, exciting times ahead.
Soldato
- Joined
- 26 Aug 2004
- Posts
- 5,165
- Location
- South Wales
Sadly AI mostly seems to be being utilised to enable laziness and when it doesn't work/can't cope with less usual circumstances it is often like hitting a brick wall
One I'm finding at the moment is MS's AI enhanced malware protection systems in the latest incarnation of Defender identifies my own coding/scripting projects as malware despite them being nothing of the kind and trying to quarantine, even erase from existence, files I've created myself/working on
I've had to disable the functionality and even then it can bite me in the rear when using a system with it enabled that I've forgotten is the case.
you mean sex bots?But when do we get robot companions?
Caporegime
- Joined
- 9 Nov 2009
- Posts
- 25,773
- Location
- Planet Earth
Nvidia was saying that raytracing would be the next best thing, because the creative industry was worth 100s of billions of USD in 2018 when Turing launched. Now they say mining is worthless. Yet a huge percentage of their revenue came from gaming and mining,and why they have increasing unsold inventory and collapsing GPU sales.
Difference is you don't need as many GPUs to power it as you did to generate coins for crypto. The AI market is there but it won't be as big - crypto sales we're making up half Nvidias revenue at the peak
Caporegime
- Joined
- 9 Nov 2009
- Posts
- 25,773
- Location
- Planet Earth
Plus there are far more players in AI now. Also a lot of their AI sales were to China which is under restrictions.
Last edited:
When you have all these companies worth trillions/billions paying into AI now, it will create demand and it all comes back to nvidia:
Microsoft explains how thousands of Nvidia GPUs built ChatGPT
ChatGPT has gone viral over the past few months, but it took several years, millions of dollars, and thousands of Nvidia GPUs to build the AI chatbot.
www.digitaltrends.com
NVIDIA Hopper GPUs Expand Reach as Demand for AI Grows
NVIDIA and key partners today announced the availability of new products and services featuring the NVIDIA H100 Tensor Core GPU — the world’s most powerful GPU for AI — to address rapidly growing demand for generative AI training and inference.
nvidianews.nvidia.com
And this is only just the start of the AI era, nvidia are so far ahead in the AI space right now.
Last edited:
Caporegime
- Joined
- 9 Nov 2009
- Posts
- 25,773
- Location
- Planet Earth
And these same companies are also slowly moving into developing their own hardware too,because it cuts out the middleman. There are far more companies globally trying to break into machine learning/AI and unlike graphics there are far less patents held by Nvidia/AMD that can get in the way of companies trying. MS is working with more than one company on AI:When you have all these companies worth trillions/billions paying into AI now, it will create demand and it all comes back to nvidia:
Microsoft explains how thousands of Nvidia GPUs built ChatGPT
ChatGPT has gone viral over the past few months, but it took several years, millions of dollars, and thousands of Nvidia GPUs to build the AI chatbot.
NVIDIA Hopper GPUs Expand Reach as Demand for AI Grows
NVIDIA and key partners today announced the availability of new products and services featuring the NVIDIA H100 Tensor Core GPU — the world’s most powerful GPU for AI — to address rapidly growing demand for generative AI training and inference.
And this is only just the start of the AI era, nvidia are so far ahead in the AI space right now.
Synopsys AI Chip Design Hits Major Milestone With Azure Cloud, STMicro And SK Hynix
Electronic Design Automation tools (EDA) bellwether, Synopsys, recently chalked up its 100th AI-designed chip tape-out with top industry players like STMicroelectronics, SK Hynix and Microsoft stepping out in endorsement of the company’s DSO.ai (Design Space Optimization) place and route technology.
www.forbes.com
Microsoft is working on its own chips for the last few years:
Microsoft is hiring engineers to work on A.I. chip design for its cloud
Job postings show Microsoft is designing custom silicon that could speed up artificial intelligence processing in its Azure cloud.
www.cnbc.com
The big problem,is the last revenue breakdown,had Nvidia loosing revenue in both gaming and professional visualisation/VFX. For the latter,this is what they said when Turing was released WRT to VFX:
NVIDIA Reinvents Computer Graphics with Turing Architecture
NVIDIA today reinvented computer graphics with the launch of the NVIDIA Turing GPU architecture. The greatest leap since the invention of the CUDA GPU in 2006, Turing features new RT Cores to accelerate ray tracing and new Tensor Cores for AI inferencing which, together for the first time, make...
nvidianews.nvidia.com
Nvidia in 2018 at the Turing launch said:
It doesn't seemed to have worked out,despite being the "leader" in raytracing. Where is all these billions of USD from VFX? Last quarter it was under $300 million revenue.
So they are betting on an area,where lots of companies are trying to get into. Just think of smartphones - BOTH Apple and Google are designing custom SOCs which have machine learning capabilities.
Last edited:
And these same companies are also slowly moving into developing their own hardware too,because it cuts out the middleman. There are far more companies globally trying to break into machine learning/AI and unlike graphics there are far less patents held by Nvidia/AMD that can get in the way of companies trying. MS is working with more than one company on AI:
Synopsys AI Chip Design Hits Major Milestone With Azure Cloud, STMicro And SK Hynix
Electronic Design Automation tools (EDA) bellwether, Synopsys, recently chalked up its 100th AI-designed chip tape-out with top industry players like STMicroelectronics, SK Hynix and Microsoft stepping out in endorsement of the company’s DSO.ai (Design Space Optimization) place and route technology.
Microsoft is working on its own chips for the last few years:
Microsoft is hiring engineers to work on A.I. chip design for its cloud
Job postings show Microsoft is designing custom silicon that could speed up artificial intelligence processing in its Azure cloud.
The big problem,is the last revenue breakdown,had Nvidia loosing revenue in both gaming and professional visualisation/VFX. For the latter,this is what they said when Turing was released WRT to VFX:
NVIDIA Reinvents Computer Graphics with Turing Architecture
NVIDIA today reinvented computer graphics with the launch of the NVIDIA Turing GPU architecture. The greatest leap since the invention of the CUDA GPU in 2006, Turing features new RT Cores to accelerate ray tracing and new Tensor Cores for AI inferencing which, together for the first time, make...
It doesn't seemed to have worked out,despite being the "leader" in raytracing. Where is all these billions of USD from VFX? Last quarter it was under $300 million revenue.
So they are betting on an area,where lots of companies are trying to get into. Just think of smartphones - BOTH Apple and Google are designing custom SOCs which have machine learning capabilities.
No doubt there will be more companies to choose from than just nvidia going forward but it's not going to be easy for them competing against nvidia, nvidia have poured an absolute **** ton of money into AI/machine learning and most importantly, they have been at it for years so have the experience here and know the market very well now, in the corporate world, especially with big global companies, things move very very very very slowly.
Maybe in a few years/decades, we'll have better/more substantial competition but it isn't going to happen overnight, so, much like the gpu market, nvidia will reign in this space for a good while too imo.
Apple are superb with their SOCs but that's apple we're talking about..... Google have been trying for a while and they're lacklustre, when it comes to SOCs, look at snapdragon and samsungs exynos, even after all these years, samsungs exynos just pails compared to snapdragon.
I reckon intel will give them a run for their money though.....
Caporegime
- Joined
- 9 Nov 2009
- Posts
- 25,773
- Location
- Planet Earth
The thing is companies like MS/Google also have extensive experience in the software area too,and the biggest issue is Nvidia wanting mega margins. This will be a driver to cut out the middlemen. This is why Apple designs its own SOCs,and literally bankrolls TSMC newer generation process nodes,and gets first dibs. Then you have the big issue,that even for AI,China was one of the biggest growth markets for Nvidia,but the new US trade restrictions put a damper on that. So China will be forced to make its own designs.
The big issue,is that Nvidia bet massively on VFX,automative,tablets,etc but in the end it was gaming/mining which propped up over half their revenues during the pandemic. Think how much money Nvidia put into all these other areas over the last 15 years?
So the big issue is Nvidia isn't still diversified against companies coming in and eating share in various areas. Graphics is by far the easier place for them to fight in. AMD is a CPU company which spends most of its R and D on CPUs and both Nvidia/AMD own a ton of patents which makes it harder for new entrants to get into the market. This is not such an issue with machine learning/AI and you need to consider the amount of money going into collectively,is more than Nvidia spends on R and D. Many of these companies have extensive experience on the software side and a number might even have state support. Nvidia trying to kill its Golden Gaming Goose by jacking up pricing to get some short term gain,is still perplexing.
Last edited:
The thing with mining, every Dick and Tom could do it with very little experience or technical knowledge. a lot of the apps like nicehash was pretty much install, click and go. You only had to tune the gpu's for a better result but for the average Tom, most didn't bother.
AI is a lot different i think
Surely you have been around long enough to not take what a CEO says to talk-up shareprices cannot be taken seriously?!
Also, very convenient that Nvidia do not have to break down revenue from all the various sources. When a CEO makes mistake it is important that they and their CFO can bury any bad news in case the shareholder start asking questions (which they seldom do when things are good even if some of those "analysts" must suspect some smoke and mirrors).
As it happens, Intel and AMD play a similar games. They all do. AMD did let slip how big Sony is to them so we sort were able to work some things out (console margins aren't that slim - even if for the PS5 Sony's revised clock targets and resulting poor yields probably wasted tons of wafers when they were super scarce; if AMD had know in advance they should have made a bigger SOC at saner clocks for the PS5)
half?? try again! more like ALL
I'll believe it's a revolution when I see it. Too many latest and greatest ideas that soak up lots of funding and then come to nothing or turn out to be overhyped. Too easy these days to generate momentum and before you know it money is pouring in, all because money is pouring in! I know in my work there's so much that an AI will never work out as very few people know how anything works, relies on that history of having worked somewhere for a long time. I'll be long gone by the time they work it all out 
Soldato
- Joined
- 21 Jul 2005
- Posts
- 21,111
- Location
- Officially least sunny location -Ronskistats
Absolutely this. He pushed the tech to hoodwink gamers that it was needed, when the other (sleight of) hand laughingly propped up the R&D/time needed to ballast the AI segment (them lovely tensor cores) and now seems to be ditching the gaming portion to the periphery. If he wanted to deliver his inexpensive GPU's as he was captured saying as a major goal he seems to be taking it in the polar opposite direction right now.
Caporegime
- Joined
- 9 Nov 2009
- Posts
- 25,773
- Location
- Planet Earth
It's not the first time either. When Nvidia started increasing prices with Kepler(by introducing the Titan tier of dGPUs),they spent loads on subsidising Tegra SOCs for their foray into tablets/automotive areas,and people here made excuses for it. It was the same when Intel was making more and more expensive quad cores,segmented features on motherboards,etc as they were subsidising Atom,etc. I told PCMR they were being used as a higher margin market to subsidise forays into other markets,and they need to stop making excuses all the time for prices increases by any company(that also will include AMD too).
There was never actually a market to expand into. They were already servicing creatives before Turing launched, with its dedicated RT cores.
Some backstory, GPU offline raytracers has been around since the maxwell GPU days (900 series). So creatives who could fit their scenes within the VRAM of a GPU were already buying GPUs to do their rendering on (I know of a person who had 7 GTX 1080s in his system), because it was faster than CPU rendering. The people who weren't rendering on their GPUs before Turing launched were not doing because GPU rendering was too slow and they needed RT cores to make it faster. They were not rendering on GPUs because of a lack of VRAM.
So there was never really a market to expand to, it was just a normal release cycle. Something that Nvidia is almost certainly aware of. Just some clever CEO spin.