저자: JensenHuang,SassineGhazi · 외국인
요약 보기
엔비디아와 시놉시스가 확대 파트너십 체결, 엔비디아 2조원 규모(20억달러) 투자. 시놉시스 설계·엔지니어링 툴을 GPU 가속 및 피지컬 AI/옴니버스 기반으로 전환, 설계 시장 확대 주장. 엔비디아는 중국 매출 미포함이며 중국은 향후 보너스 기회로 언급. 사회자는 12월 19일 BOJ 금리인상 가능성과 엔캐리 청산에 따른 미국 증시 변동성을 경고. 해당 영상의 엔비디아 관련 예측은 주로 장기 성장 전망.
다듬은 글
Joining us now exclusively and this is very exciting and timely Nvidia founder and CEO Jensen Wong and synopsis CEO Sashin Gazi and both these gentlemen I think have a lot to say about the larger issues beyond what we've been dealing with gentlemen thank you for coming on so much on spoke on the street it's this timely moment
hey Jim hey nice to see you thank you great to be here all right so Jess I'm going to take it head on the first word that I saw uh when I was reading the partnership was expand expanded. So if it's expanded, explain to me why this is a big deal we're spending some time on for the uh world's investing populace.
This is a huge deal. The partnership we're announcing today is about revolutionizing one of the most computensive industries in the world, design and engineering. uh Synopsis is pivoting across their company to transform their software and all the tools that the the industry the world's been using for some 35 years to be GPU accelerated on NVIDIA.
They're going to pivot the company to build on top of NVIDIA CUDA physics physical AI as well as Omniverse so that we could revolutionize all of these tools from EDA to SDA system design automation to CAE to computer aed drug discovery. This is going to expand the market of computing into the world of design and engineering for the very first time.
This partnership is going to be revolutionary for that entire industry, expanding their TAM, expanding the capabilities. You know what's really amazing, Jim? This is really the culmination of everything I showed you when you visited Nvidia years ago. It's taken this long for us to create the software stack necessary for Synopsis and the rest of the EDA industry and SDA industry in order for them to accelerate the software that they've historically only ran on CPUs.
And as a result, we're able to do simulations at a speed and scale unimaginable in the past so that we could do basically the entire engineering work inside a computer in a digital twin before we have to build it at all. So the the type of products we can invent and the quality that we could do and the speed that we could do it at is going to be extraordinary.
Right? I think that when you first introduced this to me, you explained to me over and over again because I was so fascinated by by what could happen with a chat GBT with the prompts was Jim that's business to consumer and it's really really important but you better start thinking enterprise. Enterprise is you were using a number that is 10 times the size of what I was thinking about.
Are you still standing by that? Absolutely. If you look at what we're doing today, if you look we've been partnering with Synopsis for a very long time. You know, Nvidia was built on a foundation of design tools from Synopsis and of course many other companies like Cadence and and uh uh Seammens and Dau.
These are very important companies that the entire engineering industry is built on top of as a result of what we're able to do today. We're able to bring GPU accelerated computing NVs CUDA into the world's industrial sector for the very first time. This is this is uh physics physical AI. This is AI that obeys the laws of physics.
That AI that interacts with the physical world. This is extremely complex stuff. It's taken us years to be able to do this. And now if you look at the world's industrial sector, it's measured in trillions of dollars. So for both Nvidia and Synopsis and the entire industry, this is a giant expansion of opportunity.
All right. So, let me speak to you about this system because when I spoke with you, the thing that was important was you were about to close the answers deal and and autos are another gigantic market. I'm trying to put things in perspective. You get the money from Nvidia, it's not exclusive. That's important.
But the partnership matters a great deal that when I look at the engineering that you do in aerospace, in automotive, in industrial, uh, I say to myself, wait a second, should I really be concerned that Gemini 3 may win the battle against perplexity? Is am I small thinking if I'm focused just on that niche?
So Jim, Synopsis' roots have been providing essential solutions for semiconductor design and verification. With the ENS acquisition, it expands our time for every engineering workload. Think about any product that you just mentioned, be it automotive, robotics, drones, machinery for industries. They're getting so sophisticated.
They're going to become AIdriven. Meaning to design these systems, you need sophisticated engineering solutions in order to deliver these systems on time, deal with the complexity at a cost that is affordable. So now with the answers portfolio it expanded our time significantly. In order to deliver these solutions and delivering it in in time for the customer to be able to simulate you need acceleration be it a GPU be it AI to change the workflow digital twin of the whole system you need to re-engineer engineering in order to deliver to these sophisticated uh future products.
Wow. Re-engineer engineering. I I I'm looking at Carl. It's so important to try to put everything in perspective because Jensen, if you were here and we didn't have the synopsis, I would be saying, are you worried that the spend uh on AI for the data center is now well exceeding the cash flows for several of the key companies for Oracle and Open AI and that perhaps the investment in open AI that you've made is a little maybe should be a little more iffy.
Uh are these the kinds of questions? Is the capex unsustainable for these kinds of for the hyperscalers? Are these the kinds of questions that are really just part of the horse race and not part of the fourth industrial revolution that you've taught me about? Most people only see the tip of the iceberg and they see a partial part of the picture as you know.
You know, I think what's going on at the foundation, we're going through a platform shift from classical general purpose computing running on CPUs to a new way of doing computing, accelerated computing running on GPUs. that old way of doing is going to continue to to uh uh exist of course but the the world is shifting to this new way of doing computing that the cash flow of the world's hyperscalers are at the core really about supporting this platform shift with or without agentic AI with or without chat bots the world's hyperscalers would have made this platform shift anyhow because this is a much more efficient way of doing computing the general uh general purpose computing has really run its course.
Moors law as you know is slowing tremendously. And so the world needs a a more powerful more capable way of doing computing going forward. And so at the at the foundation is that the second thing that to realize is that AI is not just chat bots. Chatbots are incredible. Cognitive co cognition AI is very important.
But AI covers the entire world of physical AI, industrial AI, robotics, sciences, digital biology. AI is going to transform every single industry. This is the part that we've been talking about for a very long time. You and I have been talking about this literally for years. It's just less visible because consumers don't have to use engineering tools that we use from Synopsis and Cadence and others.
We don't, you know, our companies use these tools because it's missionritical to us. These industrial applications, these industrial tools where accelerated computing and AI are now in the process of revolutionizing are foundational to our industries. It's not optional to us. It's missionritical to us.
And now for the first time, we're at the tipping point of that and we're going to revolutionize the industrial software industry with it. And just to give you a sense, Jim, what what Jensen's referring to, you're taking a workload that may run for 2 3 weeks and reducing it to hours. And that's the value we're delivering to the customers through this partnership and accelerating the software on GPU from Nvidia.
Right. Although Jensen, I'm sure you can sympathize with viewers who are trying to keep up with just your deal flow. We're we're trying to keep up with your deal flow. Um, and I guess that you must I'm wondering your communication strategy. How do you keep people afloat when they can't be expected to understand that view from 40,000 ft?
Well, we're just going to keep trying harder. Carl, I think that the the important concept here is that accelerated computing and AI is revolutionizing every single industry. Of course, it started in the consumer segment. And the reason why cons starts in the consumer segment is because getting the answer 90% right is a pretty magical thing.
But the industrial segment that 10% you don't get right becomes mission critical. And so it takes longer for us to create the tools, create the technology necessary for the industrial se sectors, the enterprise segments to be able to adopt this technology. But the time is now here. This platform shift that happened starting about 10 years ago starting in consumers is now in the process of happening across the world's industries.
Let me give you one statistic. I think this is actually about the same time that I met Jim. During that year 2016, the world's scientific supercomputers was 90% CPUs and 10% GPUs. This year it's 90% GPUs and 10% CPUs. The platform shift has happened. Now that's in scientific computing. What synapsis does, what this industry does with engineering is founded on science, founded on physics.
And so principled physics, we're in the process of making a shift in this industry. Now this is one of the most computing intensive industries in the world. And now during this transition, we're now doing the platform shift for this entire industry and elevating it with AI. We're going to expand the market really tremendously for this industry.
It's really exciting time. Okay. So, let me let me just follow up on that gentleman because uh business to consumer is very very big, but what I've been concerned about is Open AI is almost entirely right now business to consumer. We like Google so much. Obviously, they're winning the horse race right now.
How do we convince people and you did it to me but I'm struggling with others that businessto business is really the nirvana of what you're doing and the business consumer while very exciting to white collar very exciting to people who want to play with it and get answers that's it's not a sideeshow but it's it's much smaller than than what we should be thinking about
well consumer it is is exciting you know it it moves quickly. Uh, new technologies could be introduced quickly because as I mentioned, getting 90% right in recommending a movie or 90% right in recommending an ad or 90% right in recommending, you know, the next item in your in your basket is plenty delightful.
However, in the world of designing cars, planes, factories, you know, building Nvidia chips, you've got to be perfect. And the reason for that is because we have so much at stake. And so this industry requires the technology to be extraordinary. It takes longer time, but when it happens, the size of the industry is is incredibly large.
As you as you know, the world's 100 trillion dollar industry is largely industrial and enterprise to enterprise. These are enterprise, you know, serious industrial enterprise applications that are now being transformed by this platform shift. But all right, so am I supposed to believe that what'll happen at a lot of the Dell conference call where Jeff Clark talked exactly about AI factories that we're going to see a new shift that we should be thinking about how much uh GM is spending, how much BMW is spending, how much Zemens is spending, how much the big uh Korean shipyards are spending.
Is that what we should now add to the mix and not just perplexity against anthropic? Exactly. If you think of all the companies you just mentioned and the percentage of R&D they invest in automation in software etc is going to become exponential driven by the complexity and the sophistication of what they're building.
So if they were spending 2% of their revenue in R&D dedicated to the software uh automation and the technology is going to get significantly higher due to complexity in order for them to remain competitive and delivering for the future. All right and then Jensen I know here's a here's an easy way to think about it.
You know, one of the things that's really important is almost all industrial companies, companies that make things like Nvidia, like GM, like Boeing, these are companies that spend probably hundreds of hundreds of millions, maybe very low billions of dollars in engineering software tools. However, the amount that they spend in prototyping all of those products is easily 10 20 times higher.
It's the it's absolutely the same case here at Nvidia. We spent a few hundred million dollars in EDA tools and design tools. However, we spend billions of dollars in prototyping it in the future. We're going to prototype all of these products digitally so that we don't waste any money when we build it physically.
In order to build in order to prototype things digitally, simulated inside a digital twin. This is all the things I've been talking to you about. All of a sudden, the market opportunity increases by a factor of 10, 100. Okay. So last question. Right now we spent a lot of time talking about Google and what they're doing uh with their own their their own TPU and we talk about whether Metal do it but one thing that you taught me was you like to think 20 years forward and then backward.
I I'm thinking about Vera Rubin. I had to go read that Richard Feman book because you mentioned he's a pretty funny guy. But I mean, when I look at the Verir Rubin and the Fineman, am I gonna be saying, you know what, that was a great thing those other guys had during that ninemonth window that Jensen allowed him to have.
You know, listen, we've been uh we've been competing with AS6 now for quite a long time, and Google has AS6 for a long time. Uh Nvidia's Nvidia and they did a great job as you know, and I've always been complimentary. What NVIDIA does is much more versatile. Our technology is much more funible. Notice the announcement today.
Accelerating Synopsis design tools, the EDA, the SDA, the CAE tools requires a computer architecture like CUDA. It's not available for AS6. And so Nvidia can address markets that are much much broader, not just chatbots. We are also everywhere. We're in every cloud. We're in every single OEM. We're on prem and we're also at the edge.
And so the NVIDIA opportunity is much much larger. And you're now seeing a real tangible experience example of an opportunity that we could do with our platform that nobody else can. And so this is a very big opportunity for us, huge opportunity to reinvent the EDA industry, the design and engineering industry for synopsis and the rest of the industry and a growth opportunity for us because this is so computationally intensive.
I was just thinking the classic book is surely you're joking, Mr. Fineman. Maybe maybe it should be surely you're joking Mr. Wong because of Jensen. Yes, Jensen really helped me. Jensen, I didn't get to ask you about how none of your revenue includes China. Should I have just included that as a boiler point and that you can't talk about what's going on?
Kind of a coda. Well, you know, I would I would like everybody to just assume that China is a bonus opportunity in the future. At the moment, we got plenty of our demand is really strong as you know. Uh demand is really skyrocketing and and over the next couple of years, we've got a lot of demand we have to go serve.
And if uh if China comes along, which I believe is going to be in the best interest in the United States as well as China and the rest of the world, uh that's going to be a huge bonus opportunity for us. All right. Want to thank Sajin Gazi from Synopsis, a great company that was a more than a $30 billion acquisition they made of answers.
And Jensen Wong comes on and Jensen I I can't thank you enough for giving us the perspective that you gave me almost a decade ago that business consumers very interesting focus on business to business. Thank you so much, gentlemen. &. I'm going to talk about Nvidia in a moment, but really quick, I want to mention something that's very important to the entire market, and I want you to be prepared for what's ahead.
Much of the market was in the red early in today's trading session after the Japanese 10-year yield surged notably higher. I'm bringing this to your attention as a heads up. You may remember that the last two times the Bank of Japan raised interest rates. US stocks reacted negatively shortly thereafter, both in August of 2024, and then again in early 2025.
Why am I mentioning this now? Because many market participants expect the Bank of Japan to raise interest rates again at their next meeting on December 19th. Some are calling for the BOJ to wait until January to raise rates, but we'll see what happens. I'm not going to get into all the details of the yen carry trade or Japan's debt situation right now because that would take an entire video to explain.
The very simplified short version is that Japanese interest rates have been very low for a very long time and that resulted in many large firms borrowing large amounts of money from Japan and then using that money to purchase US assets including US stocks. And so if the Bank of Japan raises rates and the yen rises, some of those US assets will have to be sold.
Again, that is a very surface level and simplified explanation due to a lack of time, but that's essentially what's happening. Be prepared for some potential volatility this month ahead of the BOJ meeting on December 19th and also after the meeting if they raise interest rates because that would likely have an effect on US stocks.
Now, let's talk about Nvidia. Today, Nvidia announced an expanded partnership with Synopsis as well as a $2 billion investment in the company. The deal entails a multi-year collaboration spanning NVIDIA, CUDA, accelerated computing, Aentic and physical AI, and Omniverse digital twins, opening new market opportunities across engineering.
The companies will also collaborate in engineering and marketing activities to further adoption of GPU accelerated engineering solutions. This collaboration will enable R&D teams to design, simulate, and verify intelligent products with greater precision, speed, and at a lower cost. This partnership is a big deal.
I've been pounding the table in recent months about Nvidia's physical AI business because I genuinely don't think it's getting enough attention right now. And I'll talk more about that in a moment. Really quick, I want to remind you that Nvidia is set to present at the UBS Global Technology and AI conference this Tuesday, December 2nd, at 6:35 a.m.
Pacific, 9:35 a.m. Eastern. This event will likely , given that this event is geared toward the financial community. I expect the topic of GPU depreciation and useful life to be brought up given all the recent talk from well-known short sellers. As a reminder, I disagree with the premise that the hyperscalers are understating GPU depreciation in order to artificially boost earnings.
Rather, I think the hyperscalers have simply come to the realization that the useful life of the hardware is longer than what they originally anticipated due mainly to two factors. Nvidia's continuous improvements of older generation hardware through software. And second, the world is severely compute constrained in the face of very strong AI demand.
Since the world is compute constrained, developers will use whatever compute they can get their hands on, including hardware that under normal circumstances would be considered older generation. This extends the useful life of the hardware. There's a reason why Nvidia, a 100 GPUs from 6 years ago, are still running at full utilization today.
There's a reason why Core Weave is able to rent out H100 capacity at 95% of the original rental price. The world is computed and there is simply not enough supply to meet the demand. That is fundamentally opposite of a bubble. A bubble is characterized by a surplus of supply and a shortage of demand.
Today, we're seeing the complete opposite. Today, there's a shortage of supply in the face of very strong demand. At the peak of the dotcom bubble, most of the fiber that had been laid was dark. Today, there are no dark GPUs. Every GPU is spoken for as the world simply cannot get enough compute to meet the demand, and that will likely continue for at least a few years.
As I've said for many months now, . Jensen appears to agree with this given his recent commentary about agentic AI. And then after that, I think the next major surge in compute demand will be fueled by physical AI. And physical AI is where the NVIDIA synopsis partnership comes into focus. Today, most analysts are hyperfocused on Nvidia's data center business, and I don't think they are adequately factoring in growth from some of Nvidia's other business segments into their long-term NVIDIA estimates.
Nvidia CFO nailed it on the earnings call when she said that physical AI addresses . I a may eventually be larger than their current data center business. This is not getting enough attention right now, but I think it will get a lot of attention in 2026 and 2027. Additionally, I think we will see many new use cases emerge as a result of physical AI and that should increase the need for compute even further.
Comments 0