Original | Translation |
---|---|
Call Start: 9:35 January 1, 0000 10:05 AM ET NVIDIA Corporation (NASDAQ:NVDA) UBS Global Technology Conference Call December 3, 2024 9:35 AM ET Company Participants Colette Kress - Executive Vice President and Chief Financial Officer Conference Call Participants Tim Arcuri - UBS Tim Arcuri Good morning. Thank you. I am Tim Arcuri. I am very pleased to have NVIDIA. Pleased to have Colette with us. Thank you Colette for the time. And we're just going to kick it off. And Colette, it's been an incredible past couple of years. You're still growing very fast. I'm wondering if you can talk about some of the use cases, how they've evolved? We're having enterprises adopt AI widely. | **ํตํ ์์: 9:35 January 1, 0000 10:05 AM ET** **์๋น๋์ ์ฝํผ๋ ์ด์ (NASDAQ:NVDA)** **UBS ๊ธ๋ก๋ฒ ํ ํฌ๋๋ก์ง ์ปจํผ๋ฐ์ค ์ฝ** **2024๋ 12์ 3์ผ ์ค์ 9:35 ET** **ํ์ฌ ์ฐธ์์** ์ฝ๋ ํธ ํฌ๋ ์ค - ๋ถ์ฌ์ฅ ๊ฒธ ์ต๊ณ ์ฌ๋ฌด์ฑ ์์ **์ปจํผ๋ฐ์ค ์ฝ ์ฐธ์์** ํ ์๋ฅด์ฟ ๋ฆฌ - UBS **ํ ์๋ฅด์ฟ ๋ฆฌ** ์๋ ํ์ธ์. ๊ฐ์ฌํฉ๋๋ค. ์ ๋ ํ ์๋ฅด์ฟ ๋ฆฌ์ ๋๋ค. ์๋น๋์๋ฅผ ๋ชจ์๊ฒ ๋์ด ๋งค์ฐ ๊ธฐ์ฉ๋๋ค. ์ฝ๋ ํธ์ ํจ๊ปํ ์ ์์ด์ ๊ธฐ์ฉ๋๋ค. ์๊ฐ์ ๋ด์ฃผ์ ์ ๊ฐ์ฌํฉ๋๋ค, ์ฝ๋ ํธ. ๋ฐ๋ก ์์ํ๊ฒ ์ต๋๋ค. ์ฝ๋ ํธ, ์ง๋ ๋ช ๋ ๊ฐ ์ ๋ง ๋๋ผ์ด ์ฑ๊ณผ๋ฅผ ๋ณด์ฌ์ฃผ์ จ์ต๋๋ค. ์ฌ์ ํ ๋งค์ฐ ๋น ๋ฅด๊ฒ ์ฑ์ฅํ๊ณ ๊ณ์ญ๋๋ค. ์ฌ์ฉ ์ฌ๋ก๋ค์ด ์ด๋ป๊ฒ ๋ฐ์ ํด์๋์ง, ๊ทธ๋ฆฌ๊ณ ๊ธฐ์ ๋ค์ด AI๋ฅผ ๊ด๋ฒ์ํ๊ฒ ๋์ ํ๊ณ ์๋ ์ํฉ์ ๋ํด ๋ง์ํด ์ฃผ์ค ์ ์๋์? |
How has the demand picture sort of evolved in terms of Cloud, Consumer, Internet and Enterprise and maybe if you can talk about some of the use cases that you see that are very exciting, as well? Colette Kress Okay. Let me first start with a statement that I must read. As a reminder, this presentation contains forward-looking statements and investors are advised to read our reports filed with the SEC for information related to risk, uncertainties in facing our business. Okay, really great to see everybody here and thank you so much for hosting us. Let me kind of talk about what we have seen is certainly just been a fast journey even over the last several quarters. | ํด๋ผ์ฐ๋, ์๋น์, ์ธํฐ๋ท, ๊ทธ๋ฆฌ๊ณ ์ํฐํ๋ผ์ด์ฆ ์ธก๋ฉด์์ ์์ ์ํฉ์ด ์ด๋ป๊ฒ ๋ณํํ๋์ง, ๊ทธ๋ฆฌ๊ณ ๋งค์ฐ ํฅ๋ฏธ๋ก์ด ์ฌ์ฉ ์ฌ๋ก๋ค์ ๋ํด์๋ ๋ง์ํด ์ฃผ์ค ์ ์๋์? ์ฝ๋ ํธ ํฌ๋ ์ค ์ข์ต๋๋ค. ๋จผ์ ์ ๊ฐ ๋ฐ๋์ ์ฝ์ด์ผ ํ๋ ์ฑ๋ช ์๋ถํฐ ์์ํ๊ฒ ์ต๋๋ค. ์ฐธ๊ณ ๋ก, ์ด ํ๋ ์ ํ ์ด์ ์๋ ๋ฏธ๋์์ธก์ง์ ์ด ํฌํจ๋์ด ์์ผ๋ฉฐ, ํฌ์์๋ถ๋ค๊ป์๋ ์ ํฌ ์ฌ์ ์ด ์ง๋ฉดํ ์ํ๊ณผ ๋ถํ์ค์ฑ์ ๊ดํ ์ ๋ณด๋ฅผ ์ํด SEC์ ์ ์ถํ ์ ํฌ ๋ณด๊ณ ์๋ฅผ ์ฝ์ด๋ณด์๊ธฐ ๋ฐ๋๋๋ค. ์ข์ต๋๋ค, ์ฌ๋ฌ๋ถ์ ๋ต๊ฒ ๋์ด ์ ๋ง ๊ธฐ์๊ณ ์ ํฌ๋ฅผ ์ด๋ํด ์ฃผ์ ์ ๋๋จํ ๊ฐ์ฌํฉ๋๋ค. ์ง๋ ๋ช ๋ถ๊ธฐ ๋์์๋ ์ ๋ง ๋น ๋ฅธ ์ฌ์ ์ด์๋ค๊ณ ์ ํฌ๊ฐ ๋ชฉ๊ฒฉํ ๊ฒ์ ๋ํด ๋ง์๋๋ฆฌ๊ฒ ์ต๋๋ค. |
But keep in mind, we are 30 years in terms of the business that we're doing, but we certainly are in a very important phase. When we think about the phase that we are in and what we are seeing, we do believe that the computing platform that many of us have been using and seeing for more than 20 to 30 years is here to transform for the next decades going forward. So what that means is we are seeing those concentrating on their existing computing platform, which may include general purpose computing and a shift to accelerated computing. But a new piece was also added outside of just focusing on accelerated computing and that was the focus on AI and the focus on Generative AI. | ํ์ง๋ง ๋ช ์ฌํ์ ์ผ ํ ์ ์, ์ ํฌ๊ฐ ํ๊ณ ์๋ ์ฌ์ ์ธก๋ฉด์์๋ 30๋ ์ ์ญ์ฌ๋ฅผ ๊ฐ์ง๊ณ ์์ง๋ง, ํ์ฌ ๋งค์ฐ ์ค์ํ ๊ตญ๋ฉด์ ์๋ค๋ ๊ฒ์ ๋๋ค. ์ ํฌ๊ฐ ์ฒํด ์๋ ์ด ๊ตญ๋ฉด๊ณผ ์ ํฌ๊ฐ ๋ชฉ๊ฒฉํ๊ณ ์๋ ์ํฉ์ ์๊ฐํด๋ณด๋ฉด, ์ ํฌ ์ค ๋ง์ ์ฌ๋๋ค์ด 20๋ ์์ 30๋ ์ด์ ์ฌ์ฉํ๊ณ ์ง์ผ๋ด ์จ ์ปดํจํ ํ๋ซํผ์ด ์์ผ๋ก ์์ญ ๋ ๊ฐ ๋ณํ๋ฅผ ๋ง์ดํ ๊ฒ์ด๋ผ๊ณ ํ์ ํฉ๋๋ค. ์ด๊ฒ์ด ์๋ฏธํ๋ ๋ฐ๋, ๋ฒ์ฉ ์ปดํจํ ์ ํฌํจํ ์ ์๋ ๊ธฐ์กด ์ปดํจํ ํ๋ซํผ์ ์ง์คํ๋ ๊ฒ์์ ๊ฐ์ ์ปดํจํ ์ผ๋ก์ ์ ํ์ ์ง์คํ๋ ๋ชจ์ต์ ๋ณด๊ณ ์๋ค๋ ๊ฒ์ ๋๋ค. ํ์ง๋ง ๊ฐ์ ์ปดํจํ ์๋ง ์ง์คํ๋ ๊ฒ ์ธ์ ์๋ก์ด ์์๊ฐ ์ถ๊ฐ๋์๋๋ฐ, ๊ทธ๊ฒ์ ๋ฐ๋ก AI์ ๋ํ ์ง์ค, ๊ทธ๋ฆฌ๊ณ ์์ฑํ AI์ ๋ํ ์ง์ค์ ๋๋ค. |
The new use cases that we have seen of course over the last several quarters is the size of models. If you recall, before the very onset of Generative AI, we talked about large language models. And the importance that they were there for much of the work in terms of the consumer Internet companies, the recommender engines and model sizes continue to get larger. Right now the work in terms of foundational models are very key in terms of the work that you see happening. But you also see the scaling of those models as they focus on many different types of foundational models and multi-modal types of models being built. | ์ง๋ ๋ช ๋ถ๊ธฐ ๋์ ์ ํฌ๊ฐ ๋ชฉ๊ฒฉํ ์๋ก์ด ์ฌ์ฉ ์ฌ๋ก๋ ๋ฌผ๋ก ๋ชจ๋ธ์ ๊ท๋ชจ์ ๋๋ค. ๊ธฐ์ตํ์๊ฒ ์ง๋ง, ์์ฑํ AI๊ฐ ๋ณธ๊ฒฉ์ ์ผ๋ก ์์๋๊ธฐ ์ ์ ์ ํฌ๋ ๋๊ท๋ชจ ์ธ์ด ๋ชจ๋ธ์ ๋ํด ์ด์ผ๊ธฐํ์ต๋๋ค. ๊ทธ๋ฆฌ๊ณ ์ด๋ฌํ ๋ชจ๋ธ๋ค์ด ์๋น์ ์ธํฐ๋ท ๊ธฐ์ ๋ค์ ์ ๋ฌด, ์ถ์ฒ ์์ง ์ธก๋ฉด์์ ๋งค์ฐ ์ค์ํ์ผ๋ฉฐ, ๋ชจ๋ธ ํฌ๊ธฐ๋ ๊ณ์ํด์ ์ปค์ง๊ณ ์์ต๋๋ค. ํ์ฌ ํ์ด๋ฐ์ด์ ๋ชจ๋ธ ๊ด๋ จ ์์ ์ ์ฌ๋ฌ๋ถ์ด ๋ณด๊ณ ๊ณ์ ์ ๋ฌด ์ธก๋ฉด์์ ๋งค์ฐ ํต์ฌ์ ์ ๋๋ค. ํ์ง๋ง ๋ํ ๋ค์ํ ์ ํ์ ํ์ด๋ฐ์ด์ ๋ชจ๋ธ๊ณผ ๋ฉํฐ๋ชจ๋ฌ ์ ํ์ ๋ชจ๋ธ ๊ตฌ์ถ์ ์ง์คํ๋ฉด์ ์ด๋ฌํ ๋ชจ๋ธ๋ค์ ํ์ฅ๋ ๋ณด๊ณ ๊ณ์ค ๊ฒ์ ๋๋ค. |
The next phase of this transition is also focusing on that inferencing phase. And you'll see more and more work in terms of the types of systems that we are bringing do support that inferencing model after we have developed the large language models, all of our different types of customers are unique. We are working from everything from the start-ups to the research, the CSPs are some of the most important that are standing up compute so fast for the end enterprises that need to use that. We are working with consumer Internet companies, but more importantly, we are working globally in terms of supporting this initiative. | ์ด ์ ํ์ ๋ค์ ๋จ๊ณ๋ ์ถ๋ก (inferencing) ๋จ๊ณ์ ์ง์คํ๋ ๊ฒ์ด๊ธฐ๋ ํฉ๋๋ค. ๋๊ท๋ชจ ์ธ์ด ๋ชจ๋ธ์ ๊ฐ๋ฐํ ํ ์ถ๋ก ๋ชจ๋ธ์ ์ง์ํ๊ธฐ ์ํด ์ฐ๋ฆฌ๊ฐ ๋์ ํ๊ณ ์๋ ์์คํ ์ ํ ์ธก๋ฉด์์ ๋์ฑ ๋ง์ ์์ ์ ๋ณด์๊ฒ ๋ ๊ฒ์ ๋๋ค. ์ฐ๋ฆฌ์ ๋ค์ํ ๊ณ ๊ฐ ์ ํ์ ๋ชจ๋ ๊ณ ์ ํฉ๋๋ค. ์ฐ๋ฆฌ๋ ์คํํธ์ ๋ถํฐ ์ฐ๊ตฌ๊ธฐ๊ด๊น์ง ๋ชจ๋ ๋ถ์ผ์์ ์์ ํ๊ณ ์์ผ๋ฉฐ, CSP(ํด๋ผ์ฐ๋ ์๋น์ค ์ ๊ณต์ ์ฒด)๋ค์ ์ต์ข ๊ธฐ์ ๋ค์ด ์ฌ์ฉํ ์ ์๋๋ก ์ปดํจํ ์ ๋งค์ฐ ๋น ๋ฅด๊ฒ ๊ตฌ์ถํ๊ณ ์์ด ๊ฐ์ฅ ์ค์ํ ๊ณ ๊ฐ ์ค ์ผ๋ถ์ ๋๋ค. ์ฐ๋ฆฌ๋ ์๋น์ ์ธํฐ๋ท ๊ธฐ์ ๋ค๊ณผ๋ ํ๋ ฅํ๊ณ ์์ง๋ง, ๋ ์ค์ํ ๊ฒ์ ์ด๋ฌํ ์ด๋์ ํฐ๋ธ๋ฅผ ์ง์ํ๊ธฐ ์ํด ์ ์ธ๊ณ์ ์ผ๋ก ์์ ํ๊ณ ์๋ค๋ ์ ์ ๋๋ค. |
Nothing that we've seen before in terms of the speed and the understanding of what's going to be in front of us. So, those are some of the things that we're seeing today. Tim Arcuri And I think, we all worry that eventually we're going to build too much capacity. And I think your view is that we're nowhere close to that. So, maybe can you just speak to sort of how much visibility you have and maybe just the demand picture relative to what you're able to supply? | ์ด์ ์ ๋ณธ ์ ์ด ์๋ ์๋์ ์์ผ๋ก ๋ฌด์์ด ํผ์ณ์ง์ง์ ๋ํ ์ดํด ์์ค์
๋๋ค. ์ด๊ฒ๋ค์ด ์ค๋ ์ฐ๋ฆฌ๊ฐ ๋ณด๊ณ ์๋ ๋ช ๊ฐ์ง ์ํฉ๋ค์
๋๋ค. Tim Arcuri ๊ทธ๋ฆฌ๊ณ ์ ํฌ ๋ชจ๋๋ ๊ฒฐ๊ตญ ๋๋ฌด ๋ง์ ์์ฐ๋ฅ๋ ฅ์ ๊ตฌ์ถํ๊ฒ ๋ ๊ฒ์ ์ฐ๋ คํ๊ณ ์๋ค๊ณ ์๊ฐํฉ๋๋ค. ๊ทธ๋ฐ๋ฐ ๊ทํ์ ๊ฒฌํด๋ก๋ ์ฐ๋ฆฌ๊ฐ ๊ทธ๋ฐ ์ํฉ๊ณผ๋ ๊ฑฐ๋ฆฌ๊ฐ ๋ฉ๋ค๊ณ ๋ณด์๋ ๊ฒ ๊ฐ์ต๋๋ค. ๊ทธ๋ ๋ค๋ฉด ๊ทํ๊ฐ ์ผ๋ง๋ ๋ง์ ๊ฐ์์ฑ์ ํ๋ณดํ๊ณ ๊ณ์ ์ง, ๊ทธ๋ฆฌ๊ณ ๊ณต๊ธ ๊ฐ๋ฅํ ๋ฌผ๋ ๋๋น ์์ ์ํฉ์ ๋ํด ๋ง์ํด ์ฃผ์ค ์ ์์๊น์? |
Colette Kress Yeah, when we look at the quarters in the past and our scaling, our number one goal in terms of scaling is working with both all of our partners, downstreams with our customers, but also in terms of bringing in supply. No, we are not near any point in time, now where we are seeing any type of a slowdown. What we are seeing right now is demand continues to be fueled a lot due to the size of models, the complexity of inferencing and we are still getting ready for our next architecture, Blackwell. | ์ฝ๋ ํธ ํฌ๋ ์ค ๋ค, ๊ณผ๊ฑฐ ๋ถ๊ธฐ๋ค๊ณผ ์ ํฌ์ ํ์ฅ์ ์ดํด๋ณผ ๋, ํ์ฅ ์ธก๋ฉด์์ ์ ํฌ์ ์ต์ฐ์ ๋ชฉํ๋ ๋ชจ๋ ํํธ๋๋ค, ๊ณ ๊ฐ๋ค๊ณผ์ ๋ค์ด์คํธ๋ฆผ ํ๋ ฅ๋ฟ๋ง ์๋๋ผ ๊ณต๊ธ ํ๋ณด ์ธก๋ฉด์์๋ ํจ๊ป ์์ ํ๋ ๊ฒ์ ๋๋ค. ์๋์, ์ ํฌ๋ ํ์ฌ ์ด๋ค ์ข ๋ฅ์ ๋ํ๋ฅผ ๋ณด๊ณ ์๋ ์์ ์ ์ ํ ๊ฐ๊น์ด ์์ง ์์ต๋๋ค. ์ง๊ธ ์ ํฌ๊ฐ ๋ณด๊ณ ์๋ ๊ฒ์ ๋ชจ๋ธ์ ๊ท๋ชจ, ์ถ๋ก ์ ๋ณต์ก์ฑ์ผ๋ก ์ธํด ์์๊ฐ ๊ณ์ํด์ ํฌ๊ฒ ์ฆ๊ฐํ๊ณ ์๋ค๋ ๊ฒ์ด๋ฉฐ, ์ ํฌ๋ ์ฌ์ ํ ์ฐจ์ธ๋ ์ํคํ ์ฒ์ธ ๋ธ๋์ฐ์ ์ค๋นํ๊ณ ์์ต๋๋ค. |
And right now what we see in terms of our Blackwell, which will be here this quarter is also probably a supply constraint that is going to take us well into our next fiscal year for several quarters from now. So, no, we don't see a slowdown. We continue to see tremendous demand and interest, particularly for our new architecture that's coming out. Tim Arcuri Well, just on that point, so youโve put up a great quarter. Youโre shipping Blackwell in January and you are actually shipping more Blackwell than you thought you would three months ago. And at the same time, we do hear a lot of chop in the supply chain. You hear a lot of articles written and things like that. | ๊ทธ๋ฆฌ๊ณ ํ์ฌ ์ด๋ฒ ๋ถ๊ธฐ์ ์ถ์๋ ์์ ์ธ ์ ํฌ Blackwell์ ๊ฒฝ์ฐ์๋ ์๋ง ๊ณต๊ธ ์ ์ฝ์ด ์์ ๊ฒ์ผ๋ก ๋ณด์ด๋ฉฐ, ์ด๋ ์ง๊ธ์ผ๋ก๋ถํฐ ๋ช ๋ถ๊ธฐ ํ์ธ ๋ค์ ํ๊ณ์ฐ๋๊น์ง ์ง์๋ ๊ฒ์ผ๋ก ์์๋ฉ๋๋ค. ๋ฐ๋ผ์ ์๋์, ์ ํฌ๋ ๋ํ๋ฅผ ๋ณด์ง ์์ต๋๋ค. ํนํ ์ ํฌ๊ฐ ์ถ์ํ๋ ์๋ก์ด ์ํคํ
์ฒ์ ๋ํ ์์ฒญ๋ ์์์ ๊ด์ฌ์ ์ง์์ ์ผ๋ก ๋ณด๊ณ ์์ต๋๋ค. Tim Arcuri ๋ฐ๋ก ๊ทธ ์ ์ ๋ํด์ ๋ง์๋๋ฆฌ๋ฉด, ํ๋ฅญํ ๋ถ๊ธฐ ์ค์ ์ ๊ธฐ๋กํ์ จ๊ณ , 1์์ Blackwell์ ์ถํํ์ค ์์ ์ด๋ฉฐ ์ค์ ๋ก 3๊ฐ์ ์ ์ ์์ํ๋ ๊ฒ๋ณด๋ค ๋ ๋ง์ Blackwell์ ์ถํํ๊ณ ๊ณ์ญ๋๋ค. ๋์์ ๊ณต๊ธ๋ง์์ ๋ง์ ํผ๋์ด ์๋ค๋ ์์์ ๋ฃ๊ณ ์์ต๋๋ค. ๊ด๋ จ ๊ธฐ์ฌ๋ค๋ ๋ง์ด ๋์ค๊ณ ์๊ณ ์. |
Can you just speak to how Blackwell is different than prior product cycles? Colette Kress Yes, our Blackwell architecture is unique. What we are doing here is building at a datacenter scale. Don't get us wrong, we have been working in terms of platforms for many years, not just at the chip level and something that is end-to-end to scale. But when you look at our Hopper architecture, our Hopper architecture is for that rack scale and the work that we have done. Essentially what you're seeing with our Blackwell architecture, some of them we will completely build inside of our supply chain get it ready, get it stood up, take it down, dip it and they will stand it up together. | ๋ธ๋์ฐ์ด ์ด์ ์ ํ ์ฌ์ดํด๊ณผ ์ด๋ป๊ฒ ๋ค๋ฅธ์ง ๋ง์ํด ์ฃผ์ค ์ ์๋์? ์ฝ๋ ํธ ํฌ๋ ์ค: ๋ค, ์ ํฌ ๋ธ๋์ฐ ์ํคํ ์ฒ๋ ๋ ํนํฉ๋๋ค. ์ ํฌ๊ฐ ์ฌ๊ธฐ์ ํ๊ณ ์๋ ๊ฒ์ ๋ฐ์ดํฐ์ผํฐ ๊ท๋ชจ๋ก ๊ตฌ์ถํ๋ ๊ฒ์ ๋๋ค. ์คํดํ์ง ๋ง์๊ธฐ ๋ฐ๋๋๋ค. ์ ํฌ๋ ์๋ ๊ฐ ํ๋ซํผ ์ธก๋ฉด์์ ์์ ํด์์ต๋๋ค. ๋จ์ํ ์นฉ ์์ค์ด ์๋๋ผ ํ์ฅ ๊ฐ๋ฅํ ์๋ํฌ์๋ ์๋ฃจ์ ์ผ๋ก ๋ง์ ๋๋ค. ํ์ง๋ง ์ ํฌ ํธํผ ์ํคํ ์ฒ๋ฅผ ๋ณด์๋ฉด, ํธํผ ์ํคํ ์ฒ๋ ๋ ๊ท๋ชจ์ ์ ํฌ๊ฐ ์ํํ ์์ ์ ์ํ ๊ฒ์ ๋๋ค. ๋ณธ์ง์ ์ผ๋ก ๋ธ๋์ฐ ์ํคํ ์ฒ์์ ๋ณด์๋ ๊ฒ์, ์ผ๋ถ๋ ์ ํฌ๊ฐ ๊ณต๊ธ๋ง ๋ด์์ ์์ ํ ๊ตฌ์ถํ์ฌ ์ค๋นํ๊ณ , ์ค์นํ๊ณ , ํด์ฒดํ๊ณ , ํ ์คํธํ ํ ํจ๊ป ๋ค์ ์ค์นํ ๊ฒ์ ๋๋ค. |
And what we are doing is a greater portion of choices for the customers depending on where they are in their lifecycle. And what we mean by that is datacenters are complex and the types of things that they do to make them the most efficient, each of them are at different stages. So you have the opportunity to choose between a lot of different options of what we're doing. That means we can do liquid cool. We can do continued in terms of air-cooled. You can incorporate an ARM CPU, but also x86 if you want. There's many different networking options that we have, whether that's InfiniBand and Ethernet and many different other switch choices in there. | ๊ทธ๋ฆฌ๊ณ ์ ํฌ๊ฐ ํ๊ณ ์๋ ๊ฒ์ ๊ณ ๊ฐ๋ค์ด ๊ทธ๋ค์ ๋ผ์ดํ์ฌ์ดํด์์ ์ด๋ ๋จ๊ณ์ ์๋์ง์ ๋ฐ๋ผ ๋ ๋ง์ ์ ํ๊ถ์ ์ ๊ณตํ๋ ๊ฒ์ ๋๋ค. ์ด๊ฒ์ด ์๋ฏธํ๋ ๋ฐ๋ ๋ฐ์ดํฐ์ผํฐ๋ ๋ณต์กํ๊ณ , ๋ฐ์ดํฐ์ผํฐ๋ฅผ ๊ฐ์ฅ ํจ์จ์ ์ผ๋ก ๋ง๋ค๊ธฐ ์ํด ํ๋ ์์ ๋ค์ ์ ํ์ด ๊ฐ๊ฐ ๋ค๋ฅธ ๋จ๊ณ์ ์๋ค๋ ๊ฒ์ ๋๋ค. ๋ฐ๋ผ์ ์ ํฌ๊ฐ ์ ๊ณตํ๋ ๋ค์ํ ์ต์ ๋ค ์ค์์ ์ ํํ ์ ์๋ ๊ธฐํ๊ฐ ์์ต๋๋ค. ์ด๋ ์ก์ฒด ๋๊ฐ์ ํ ์๋ ์๊ณ , ๊ณต๊ธฐ ๋๊ฐ์ ๊ณ์ ์ฌ์ฉํ ์๋ ์๋ค๋ ์๋ฏธ์ ๋๋ค. ARM CPU๋ฅผ ํตํฉํ ์๋ ์์ง๋ง, ์ํ๋ค๋ฉด x86๋ ์ฌ์ฉํ ์ ์์ต๋๋ค. ์ ํฌ๊ฐ ๋ณด์ ํ ๋คํธ์ํน ์ต์ ๋ ๋งค์ฐ ๋ค์ํฉ๋๋ค. InfiniBand์ ์ด๋๋ท์ ๋น๋กฏํด ๊ทธ ์์์ ์ ํํ ์ ์๋ ๋ค์ํ ์ค์์น ์ต์ ๋ค์ด ์์ต๋๋ค. |
That decision is with the customer and many different configurations. When we think about where we are right now in Blackwell for this quarter, all is done in terms of the chip. The chip is fine. The chip and the work that we have done has moved quite well. Right now, we are standing up configurations for so many of our different customers. You will all see pictures on the Internet. Pictures with happy faces as they get excited for standing up the first one and getting ready to put the whole datacenter and all of the different racks together in terms of that. That's where we stand today. Tim Arcuri Great. | ํด๋น ๊ฒฐ์ ์ ๊ณ ๊ฐ๊ณผ ๋ค์ํ ๊ตฌ์ฑ์ ๋ฌ๋ ค ์์ต๋๋ค. ์ด๋ฒ ๋ถ๊ธฐ Blackwell์ ํ์ฌ ์ํฉ์ ์๊ฐํด๋ณด๋ฉด, ์นฉ ์ธก๋ฉด์์๋ ๋ชจ๋ ๊ฒ์ด ์๋ฃ๋์์ต๋๋ค. ์นฉ์ ๋ฌธ์ ์๊ณ , ์ ํฌ๊ฐ ์ํํ ์์
๋ ์๋นํ ์์กฐ๋กญ๊ฒ ์งํ๋์์ต๋๋ค. ํ์ฌ ์ ํฌ๋ ๋ค์ํ ๊ณ ๊ฐ๋ค์ ์ํ ๊ตฌ์ฑ ์ค์ ์ ์งํํ๊ณ ์์ต๋๋ค. ์ฌ๋ฌ๋ถ๋ค๋ ์ธํฐ๋ท์์ ์ฌ์ง๋ค์ ๋ณด์ค ์ ์์ ๊ฒ์
๋๋ค. ์ฒซ ๋ฒ์งธ ์์คํ
์ ๊ตฌ์ถํ๋ฉฐ ๊ธฐ๋ปํ๊ณ , ์ ์ฒด ๋ฐ์ดํฐ์ผํฐ์ ๋ชจ๋ ๋ค์ํ ๋๋ค์ ํจ๊ป ๊ตฌ์ฑํ ์ค๋น๋ฅผ ํ๋ ํ๋ณตํ ํ์ ์ ์ฌ์ง๋ค์ ๋ง์
๋๋ค. ์ด๊ฒ์ด ์ค๋ ํ์ฌ ์ ํฌ๊ฐ ์ ์๋ ์์น์
๋๋ค. Tim Arcuri: ์ข์ต๋๋ค. |
And there's just a lot going on with the product cadence and you have B100, you have B200, you have the racks with [TB], then you have Blackwell Ultra coming about six months later. So, is there a risk that customers wait, because you have products coming so quickly? Colette Kress No, when you think about what is necessary in designing the datacenter, it does takes planning. And what we have seen in, at least, the last five to ten years is more work with all of our different customers as they plan every six months they're planning. What is here? What am I going to build? And they need to be ready for compute at that time that they are working in terms of in projects. | ๊ทธ๋ฆฌ๊ณ ์ ํ ์ถ์ ์ผ์ ์ ๋ง์ ์ผ๋ค์ด ์์ต๋๋ค. B100๋ ์๊ณ , B200๋ ์๊ณ , [TB]๊ฐ ํฌํจ๋ ๋๋ ์๊ณ , ๊ทธ ๋ค์์๋ ์ฝ 6๊ฐ์ ํ์ Blackwell Ultra๊ฐ ์ถ์๋ ์์ ์
๋๋ค. ๊ทธ๋ ๋ค๋ฉด ์ ํ๋ค์ด ๋๋ฌด ๋น ๋ฅด๊ฒ ์ถ์๋๊ธฐ ๋๋ฌธ์ ๊ณ ๊ฐ๋ค์ด ๊ธฐ๋ค๋ฆด ์ํ์ด ์์๊น์? ์ฝ๋ ํธ ํฌ๋ ์ค ์๋์, ๋ฐ์ดํฐ์ผํฐ๋ฅผ ์ค๊ณํ๋ ๋ฐ ํ์ํ ๊ฒ๋ค์ ์๊ฐํด๋ณด๋ฉด, ๊ณํ์ด ํ์ํฉ๋๋ค. ๊ทธ๋ฆฌ๊ณ ์ ์ด๋ ์ง๋ 5๋ ์์ 10๋ ๋์ ์ฐ๋ฆฌ๊ฐ ๋ณธ ๊ฒ์ ๋ชจ๋ ๋ค์ํ ๊ณ ๊ฐ๋ค๊ณผ ๋ ๋ง์ ํ์ ์ ํ๊ณ ์๋ค๋ ๊ฒ์ ๋๋ค. ๊ทธ๋ค์ด 6๊ฐ์๋ง๋ค ๊ณํ์ ์ธ์ธ ๋ ๋ง์ด์ฃ . ์ง๊ธ ๋ฌด์์ด ์๋๊ฐ? ๋ด๊ฐ ๋ฌด์์ ๊ตฌ์ถํ ๊ฒ์ธ๊ฐ? ๊ทธ๋ฆฌ๊ณ ๊ทธ๋ค์ด ํ๋ก์ ํธ์์ ์์ ํ๋ ๊ทธ ์์ ์ ์ปดํจํ ์ ์ํ ์ค๋น๊ฐ ๋์ด ์์ด์ผ ํฉ๋๋ค. |
So given that there are things that are still short in supply, we are still serving with an amazing configuration Hopper 200 for our customers. That is an opportunity for them to begin some of the work that they are doing with an HGX system and they havenโt yet even touched a GPU yet, just because the demand for that has been so strong over. The folks have worked with us helping understand how they build out the datacenters. The datacenters often have already been procured. They just know they need to fit what will be inside of those. That's why you see us with two architectures. | ๋ฐ๋ผ์ ์ฌ์ ํ ๊ณต๊ธ์ด ๋ถ์กฑํ ์ ํ๋ค์ด ์๋ ์ํฉ์์, ์ ํฌ๋ ๊ณ ๊ฐ๋ค์๊ฒ ๋๋ผ์ด ๊ตฌ์ฑ์ Hopper 200์ ์ ๊ณตํ๊ณ ์์ต๋๋ค. ์ด๋ ๊ณ ๊ฐ๋ค์ด HGX ์์คํ ์ผ๋ก ์์ ์ ์์ํ ์ ์๋ ๊ธฐํ๊ฐ ๋๋ฉฐ, ์์๊ฐ ๋งค์ฐ ๊ฐํ๊ธฐ ๋๋ฌธ์ ์์ง GPU์กฐ์ฐจ ๊ฑด๋๋ฆฌ์ง ๋ชปํ ์ํฉ์ ๋๋ค. ์ ํฌ์ ํจ๊ป ์์ ํ๋ ๋ถ๋ค์ ๋ฐ์ดํฐ์ผํฐ ๊ตฌ์ถ ๋ฐฉ๋ฒ์ ์ดํดํ๋ ๋ฐ ๋์์ ์ฃผ๊ณ ์์ต๋๋ค. ๋ฐ์ดํฐ์ผํฐ๋ ์ด๋ฏธ ์กฐ๋ฌ๋ ๊ฒฝ์ฐ๊ฐ ๋ง์ผ๋ฉฐ, ๊ทธ๋ค์ ๋จ์ง ๊ทธ ์์ ๋ฌด์์ ๋ฃ์์ง๋ง ์๋ฉด ๋๋ ์ํฉ์ ๋๋ค. ์ด๊ฒ์ด ์ ํฌ๊ฐ ๋ ๊ฐ์ง ์ํคํ ์ฒ๋ฅผ ๋ณด์ ํ๊ณ ์๋ ์ด์ ์ ๋๋ค. |
Additionally, as you know, we will do Vera Rubin going forward that again will be a discussion that we will have with customers that says here is a potential offerings. How can we help you think through what you may need going forward? Tim Arcuri But I guess maybe it's a bit similar to the H100 and H200 where customers didnโt wait and that went all great. So I guess there is evidence out there where customers did not wait when you had a product come out quickly after those Vera products? Colette Kress The customers are eager. Wait is a strong word, I would say, they call every day asking in terms of when they can see the compute and we are working feverishly for that new supply. | ์ถ๊ฐ์ ์ผ๋ก, ์์๋ค์ํผ ์์ผ๋ก Vera Rubin์ ์ถ์ํ ์์ ์ธ๋ฐ, ์ด ์ญ์ ๊ณ ๊ฐ๋ค๊ณผ ๋
ผ์ํ ์ฌ์์ด ๋ ๊ฒ์
๋๋ค. ์ ์ฌ์ ์ธ ์ ํ ์ ๊ณต์ ๋ํด ์ค๋ช
ํ๊ณ , ํฅํ ํ์ํ ์ ์๋ ๊ฒ๋ค์ ๋ํด ์ด๋ป๊ฒ ์๊ฐํด๋ณผ ์ ์๋๋ก ๋์ธ ์ ์๋์ง์ ๋ํ ๋
ผ์ ๋ง์
๋๋ค. **Tim Arcuri** ํ์ง๋ง H100๊ณผ H200์ ๊ฒฝ์ฐ์ ๋ค์ ๋น์ทํ ๊ฒ ๊ฐ์๋ฐ, ๋น์ ๊ณ ๊ฐ๋ค์ ๊ธฐ๋ค๋ฆฌ์ง ์์๊ณ ๋ชจ๋ ๊ฒ์ด ์์กฐ๋กญ๊ฒ ์งํ๋์์ต๋๋ค. ๋ฐ๋ผ์ Vera ์ ํ๋ค ์ดํ ๋น ๋ฅด๊ฒ ์ถ์๋๋ ์ ํ์ ๋ํด ๊ณ ๊ฐ๋ค์ด ๊ธฐ๋ค๋ฆฌ์ง ์์๋ ์ฌ๋ก๊ฐ ์๋ค๊ณ ๋ณผ ์ ์์ง ์์๊น์? **Colette Kress** ๊ณ ๊ฐ๋ค์ ๋งค์ฐ ์ด๋งํ๊ณ ์์ต๋๋ค. '๊ธฐ๋ค๋ฆฐ๋ค'๋ ํํ์ ๊ฐํ ํํ์ด๋ผ๊ณ ์๊ฐํ๋๋ฐ, ๊ณ ๊ฐ๋ค์ ์ธ์ ์ปดํจํ ์ ํ์ ๋ณผ ์ ์๋์ง ๋งค์ผ ์ ํ๋ฅผ ๊ฑธ์ด ๋ฌธ์ํ๊ณ ์์ผ๋ฉฐ, ์ ํฌ๋ ์๋ก์ด ๊ณต๊ธ์ ์ํด ์ด์ฌํ ์์ ํ๊ณ ์์ต๋๋ค. |
But what the importance is, is remember, this is a journey that's probably going to take place for two decades. Everybody will get onboard. Our architectures allows an end-to-end scaling approach for them to do whatever they need to in the world of accelerated computing and Ai. And we're a very strong candidate to help them, not only with that infrastructure, but also with the software. Tim Arcuri Thank you. So I wanted to ask you the cash question. We actually talked about this last night. You've generated about $56 billion in the cash flow over the past four quarters. I heard you are generating about $120 billion in calendar โ25. | ํ์ง๋ง ์ค์ํ ๊ฒ์, ์ด๊ฒ์ด ์๋ง๋ 20๋
๊ฐ ์ง์๋ ์ฌ์ ์ด๋ผ๋ ์ ์ ๊ธฐ์ตํด์ผ ํ๋ค๋ ๊ฒ์
๋๋ค. ๋ชจ๋ ๊ธฐ์
๋ค์ด ์ฐธ์ฌํ๊ฒ ๋ ๊ฒ์
๋๋ค. ์ฐ๋ฆฌ์ ์ํคํ
์ฒ๋ ๊ทธ๋ค์ด ๊ฐ์ ์ปดํจํ
๊ณผ AI ๋ถ์ผ์์ ํ์๋ก ํ๋ ๋ชจ๋ ๊ฒ์ ์ํํ ์ ์๋๋ก ์๋ํฌ์๋ ํ์ฅ ์ ๊ทผ๋ฒ์ ์ ๊ณตํฉ๋๋ค. ๊ทธ๋ฆฌ๊ณ ์ฐ๋ฆฌ๋ ์ธํ๋ผ๋ฟ๋ง ์๋๋ผ ์ํํธ์จ์ด ์ธก๋ฉด์์๋ ๊ทธ๋ค์ ๋์ธ ์ ์๋ ๋งค์ฐ ๊ฐ๋ ฅํ ํ๋ณด์
๋๋ค. ํ ์๋ฅด์ฟ ๋ฆฌ(Tim Arcuri) ๊ฐ์ฌํฉ๋๋ค. ํ๊ธ์ ๋ํ ์ง๋ฌธ์ ๋๋ฆฌ๊ณ ์ถ์ต๋๋ค. ์ค์ ๋ก ์ด์ ฏ๋ฐค์ ์ด์ ๋ํด ์ด์ผ๊ธฐ๋ฅผ ๋๋ด๋๋ฐ์. ์ง๋ 4๋ถ๊ธฐ ๋์ ์ฝ 560์ต ๋ฌ๋ฌ์ ํ๊ธํ๋ฆ์ ์ฐฝ์ถํ์ จ์ต๋๋ค. 2025๋ ์ญ๋ ๊ธฐ์ค์ผ๋ก๋ ์ฝ 1,200์ต ๋ฌ๋ฌ๋ฅผ ์ฐฝ์ถํ ๊ฒ์ผ๋ก ๋ค์์ต๋๋ค. |
And even if I assume that you buy back about $50 billion per year, you're going to end up with a $100 billion in cash at the end of next year and potentially $200 billion in cash at the end of 2026. These are obviously massive, massive numbers even Apple, which got to $100 billion in cash ended up working that down. How do you think about what you're going to do with the cash? Colette Kress Our work in terms of our cash and our use of cash is some of the most important things that any company has to look at. We do spend with a great group of people, helping things through our strategy and our needs. | ๊ทธ๋ฆฌ๊ณ ์ฐ๊ฐ ์ฝ 500์ต ๋ฌ๋ฌ์ ์์ฌ์ฃผ ๋งค์
์ ๊ฐ์ ํ๋๋ผ๋, ๋ด๋
๋ง์๋ 1,000์ต ๋ฌ๋ฌ์ ํ๊ธ์, ๊ทธ๋ฆฌ๊ณ 2026๋
๋ง์๋ ์ ์ฌ์ ์ผ๋ก 2,000์ต ๋ฌ๋ฌ์ ํ๊ธ์ ๋ณด์ ํ๊ฒ ๋ ๊ฒ์
๋๋ค. ์ด๋ ๋ถ๋ช
ํ ์์ฒญ๋, ์์ฒญ๋ ๊ท๋ชจ์ ์์น์
๋๋ค. ์ฌ์ง์ด 1,000์ต ๋ฌ๋ฌ์ ํ๊ธ์ ๋ณด์ ํ๋ ์ ํ์กฐ์ฐจ๋ ๊ฒฐ๊ตญ ๊ทธ ๊ท๋ชจ๋ฅผ ์ค์ฌ๋๊ฐ์ต๋๋ค. ์ด๋ฌํ ํ๊ธ์ ์ด๋ป๊ฒ ํ์ฉํ ๊ณํ์ธ์ง์ ๋ํด ์ด๋ป๊ฒ ์๊ฐํ๊ณ ๊ณ์ ๊ฐ์? ์ฝ๋ ํธ ํฌ๋ ์ค: ํ๊ธ๊ณผ ํ๊ธ ํ์ฉ์ ๋ํ ์ฐ๋ฆฌ์ ์ ๋ฌด๋ ์ด๋ค ํ์ฌ๋ ์ดํด๋ด์ผ ํ ๊ฐ์ฅ ์ค์ํ ์ฌ์ ์ค ํ๋์ ๋๋ค. ์ฐ๋ฆฌ๋ ์ฐ๋ฆฌ์ ์ ๋ต๊ณผ ํ์์ฌํญ์ ํตํด ๋์์ ์ฃผ๋ ํ๋ฅญํ ์ธ์ฌ๋ค๊ณผ ํจ๊ป ์ง์ถํ๊ณ ์์ต๋๋ค. |
Your first thing that you're going to consider is what is that cash that we need for innovation and all sorts of support just our work in the scaling that we are doing. That's going to be the first pieces. But we are continuing and know that innovation and everything that we are doing in R&D is going to be an important part of that. Now we can do that from both learnings from companies. We can also think about that in terms of our work of bringing on great teams in some M&A form that they come onboard. That's a great opportunity for us to do and we will continue to work also in that area. | ๋จผ์ ๊ณ ๋ คํด์ผ ํ ๊ฒ์ ํ์ ๊ณผ ํ์ฌ ์งํ ์ค์ธ ํ์ฅ ์์ ์ ์ง์ํ๋ ๋ฐ ํ์ํ ํ๊ธ์ด ์ผ๋ง์ธ์ง์ ๋๋ค. ์ด๊ฒ์ด ์ฒซ ๋ฒ์งธ ์์๊ฐ ๋ ๊ฒ์ ๋๋ค. ํ์ง๋ง ์ ํฌ๋ ๊ณ์ํด์ ํ์ ๊ณผ R&D์์ ์ํํ๊ณ ์๋ ๋ชจ๋ ํ๋์ด ๊ทธ ์ค์ํ ๋ถ๋ถ์ด ๋ ๊ฒ์์ ์๊ณ ์์ต๋๋ค. ์ด์ ์ ํฌ๋ ๊ธฐ์ ๋ค๋ก๋ถํฐ์ ํ์ต์ ํตํด ์ด๋ฅผ ์ํํ ์ ์์ต๋๋ค. ๋ํ M&A ํํ๋ก ์ฐ์ํ ํ๋ค์ ์์ ํ์ฌ ํฉ๋ฅ์ํค๋ ์ ํฌ์ ์์ ์ธก๋ฉด์์๋ ์๊ฐํด๋ณผ ์ ์์ต๋๋ค. ์ด๋ ์ ํฌ์๊ฒ ํ๋ฅญํ ๊ธฐํ์ด๋ฉฐ, ํด๋น ์์ญ์์๋ ๊ณ์ ์์ ํด ๋๊ฐ ๊ฒ์ ๋๋ค. |
Then it leads to thinking about new types of business models that we may want to add and focus on in new areas of AI. And the support that we can do not only for, let's just say, building out software, but building out full systems for others and we'll be investing in that. After we determine those types of investments, our work is, in terms of returning to shareholders, it always will be and our focus on that is a combination of share repurchases and dividends and you'll continue to see this, watch this very carefully. We're not a fan of excess cash. So we are going to watch this carefully and you'll continue for the next quarters. Tim Arcuri Great. Thank you. | ๊ทธ๋ฌ๋ฉด ์ฐ๋ฆฌ๊ฐ AI์ ์๋ก์ด ์์ญ์์ ์ถ๊ฐํ๊ณ ์ง์คํ๊ณ ์ ํ๋ ์๋ก์ด ์ ํ์ ๋น์ฆ๋์ค ๋ชจ๋ธ์ ๋ํด ์๊ฐํ๊ฒ ๋ฉ๋๋ค. ๊ทธ๋ฆฌ๊ณ ์ํํธ์จ์ด ๊ตฌ์ถ๋ฟ๋ง ์๋๋ผ ๋ค๋ฅธ ๊ธฐ์
๋ค์ ์ํ ์์ ํ ์์คํ
๊ตฌ์ถ์ ์ง์ํ ์ ์๋ ๋ถ๋ถ์ ํฌ์ํ ์์ ์
๋๋ค. ์ด๋ฌํ ์ ํ์ ํฌ์๋ฅผ ๊ฒฐ์ ํ ํ, ์ฃผ์ฃผ ํ์ ์ธก๋ฉด์์ ์ฐ๋ฆฌ์ ์
๋ฌด๋ ํญ์ ๊ทธ๋์๋ฏ์ด ์์ฌ์ฃผ ๋งค์
๊ณผ ๋ฐฐ๋น์ ์กฐํฉ์ ์ค์ ์ ๋๊ณ ์์ผ๋ฉฐ, ์ฌ๋ฌ๋ถ๊ป์๋ ์ด๋ฅผ ๊ณ์ํด์ ๋งค์ฐ ์ฃผ์ ๊น๊ฒ ์ง์ผ๋ณด์ค ๊ฒ์
๋๋ค. ์ฐ๋ฆฌ๋ ๊ณผ๋ํ ํ๊ธ์ ์ ํธํ์ง ์์ต๋๋ค. ๋ฐ๋ผ์ ์ด๋ฅผ ์ ์คํ ๋ชจ๋ํฐ๋งํ ๊ฒ์ด๋ฉฐ, ํฅํ ๋ถ๊ธฐ์๋ ๊ณ์ ์ง์ผ๋ด ์ฃผ์๊ธฐ ๋ฐ๋๋๋ค. Tim Arcuri ๊ฐ์ฌํฉ๋๋ค. |
As we talk about gross margin for a minute, you did say that gross margin comes down in the fiscal Q1 of next year as Blackwell ramps, you said it comes on to low 70s, which you then clarified as 71 through 72.5 on the call. But it comes back up after that and it comes back up to the mid-70s as you get to the end of the year. Can you talk about sort of how confident you are in that? And with Rubin coming after that, there are some people that think, โWell, gross margin will be under pressure once again, once Rubin begins to ramp.โ So, can you talk about how confident you are that you can maintain mid-70s over time? Colette Kress Okay. | ์ ์ ๋งค์ถ์ด์ด์ต๋ฅ ์ ๋ํด ์ด์ผ๊ธฐํด๋ณด๊ฒ ์ต๋๋ค. ๋ง์ํ์ ๋ฐ๋ก๋ Blackwell์ด ์ฆ์ฐ๋๋ฉด์ ๋ด๋
ํ๊ณ์ฐ๋ 1๋ถ๊ธฐ์ ๋งค์ถ์ด์ด์ต๋ฅ ์ด ํ๋ฝํ์ฌ 70% ์ด๋ฐ๋๋ก ๋จ์ด์ง๋ค๊ณ ํ์
จ๋๋ฐ, ํตํ์์ ์ด๋ฅผ 71%์์ 72.5% ์ฌ์ด๋ก ๊ตฌ์ฒดํํด ์ฃผ์
จ์ต๋๋ค. ํ์ง๋ง ๊ทธ ์ดํ ๋ค์ ์์นํ์ฌ ์ฐ๋ง๊น์ง 70% ์ค๋ฐ๋๋ก ํ๋ณต๋๋ค๊ณ ํ์
จ์ต๋๋ค. ์ด์ ๋ํ ํ์ ์ ๋๋ ์ด๋ค์ง ๋ง์ํด ์ฃผ์ค ์ ์๋์? ๊ทธ๋ฆฌ๊ณ ๊ทธ ์ดํ Rubin์ด ์ถ์๋๋ฉด, ์ผ๋ถ์์๋ "Rubin์ด ์ฆ์ฐ์ ์์ํ๋ฉด ๋งค์ถ์ด์ด์ต๋ฅ ์ด ๋ค์ ์๋ฐ์ ๋ฐ์ ๊ฒ"์ด๋ผ๊ณ ์๊ฐํ๊ณ ์์ต๋๋ค. ๋ฐ๋ผ์ ์ฅ๊ธฐ์ ์ผ๋ก 70% ์ค๋ฐ๋๋ฅผ ์ ์งํ ์ ์๋ค๋ ํ์ ์ด ์ด๋ ์ ๋์ธ์ง ๋ง์ํด ์ฃผ์ค ์ ์๋์? ์ฝ๋ ํธ ํฌ๋ ์ค: ๋ค. |
So Blackwell is unique as we've discussed in terms of its different configurations and we're standing up quite many different ones that you are going to see go to market, even in this quarter. We're not just shipping one version. There's going to be several. So therefore the volume at this time is quite small. As we continue to scale this throughout the year, we will be able to improve the gross margins once we get into the scaling of all of our different system configurations that we have. When we think about going forward though and the Vera Rubin, little far out there to go through, we still have to run through an analysis in terms of the TCO and things that we would do. | ๋ฐ๋ผ์ ๋ธ๋์ฐ์ ์์ ๋ ผ์ํ ๋ฐ์ ๊ฐ์ด ๋ค์ํ ๊ตฌ์ฑ์์ ๋ ํนํ ํน์ง์ ๊ฐ์ง๊ณ ์์ผ๋ฉฐ, ์ด๋ฒ ๋ถ๊ธฐ์๋ ์์ฅ์ ์ถ์๋ ์ฌ๋ฌ ๊ฐ์ง ๋ค๋ฅธ ๊ตฌ์ฑ๋ค์ ์ค๋นํ๊ณ ์์ต๋๋ค. ์ ํฌ๋ ๋จ์ผ ๋ฒ์ ๋ง ์ถํํ๋ ๊ฒ์ด ์๋๋๋ค. ์ฌ๋ฌ ๋ฒ์ ์ด ์์ ์์ ์ ๋๋ค. ๋ฐ๋ผ์ ํ์ฌ๋ก์๋ ๋ฌผ๋์ด ์๋นํ ์ ์ต๋๋ค. ์ฌํด ๋ด๋ด ์ด๋ฅผ ์ง์์ ์ผ๋ก ํ๋ํด ๋๊ฐ๋ฉด์, ์ ํฌ๊ฐ ๋ณด์ ํ ๋ชจ๋ ๋ค์ํ ์์คํ ๊ตฌ์ฑ์ ๊ท๋ชจ ํ๋์ ์ง์ ํ๊ฒ ๋๋ฉด ์ด ๋ง์ง์ ๊ฐ์ ํ ์ ์์ ๊ฒ์ ๋๋ค. ํ์ง๋ง ์์ผ๋ก ๋ฒ ๋ผ ๋ฃจ๋น์ ๊ณ ๋ คํ ๋๋ ์์ง ๊ฒํ ํ๊ธฐ์๋ ๋ค์ ๋จผ ์ด์ผ๊ธฐ์ด๋ฉฐ, ์ฌ์ ํ TCO(์ด์์ ๋น์ฉ) ์ธก๋ฉด์์ ๋ถ์์ ์งํํด์ผ ํ๊ณ ์ ํฌ๊ฐ ํด์ผ ํ ์ผ๋ค์ด ์์ต๋๋ค. |
So we'll put that off in terms of a little later in terms of what we see. But I'd say, we're in a unique position right now just with Blackwell and what we're seeing. Tim Arcuri And how have you been able to move gross margin up so much? Because when Hopper launched, I remember during the early phases of Hopper gross margin was in the mid-60s, and now you're going to end up in the mid-70s. So, how have you moved gross margin up so much? Colette Kress Okay. There are many different things when we are determining the value that we have provided to customers. It's not just about performance. It's not just about performance of the chips. | ๋ฐ๋ผ์ ์ ํฌ๊ฐ ๋ณด๊ณ ์๋ ์ํฉ์ ๋ํด์๋ ์กฐ๊ธ ๋์ค์ ๋ง์๋๋ฆฌ๊ฒ ์ต๋๋ค. ํ์ง๋ง ํ์ฌ Blackwell๊ณผ ์ ํฌ๊ฐ ๋ณด๊ณ ์๋ ์ํฉ์ ๊ณ ๋ คํ ๋, ์ ํฌ๋ ๋
ํนํ ์์น์ ์๋ค๊ณ ๋ง์๋๋ฆด ์ ์์ต๋๋ค. **Tim Arcuri** ๊ทธ๋ฆฌ๊ณ ์ด๋ป๊ฒ ๋งค์ถ์ด์ด์ต๋ฅ ์ ๊ทธ๋ ๊ฒ ๋ง์ด ๋์ด์ฌ๋ฆด ์ ์์๋์? Hopper๊ฐ ์ถ์๋์์ ๋, Hopper ์ด๊ธฐ ๋จ๊ณ์์ ๋งค์ถ์ด์ด์ต๋ฅ ์ด 60% ์ค๋ฐ๋์๋ ๊ฒ์ผ๋ก ๊ธฐ์ตํ๋๋ฐ, ์ด์ ๋ 70% ์ค๋ฐ๋๋ก ๋๋ ๊ฒ ๊ฐ์ต๋๋ค. ๊ทธ๋ ๋ค๋ฉด ์ด๋ป๊ฒ ๋งค์ถ์ด์ด์ต๋ฅ ์ ๊ทธ๋ ๊ฒ ๋ง์ด ํฅ์์ํฌ ์ ์์๋์? **Colette Kress** ๋ค. ์ ํฌ๊ฐ ๊ณ ๊ฐ์๊ฒ ์ ๊ณตํ ๊ฐ์น๋ฅผ ๊ฒฐ์ ํ ๋ ๊ณ ๋ คํ๋ ๋ค์ํ ์์๋ค์ด ์์ต๋๋ค. ๋จ์ํ ์ฑ๋ฅ๋ง์ ๋ฌธ์ ๊ฐ ์๋๋๋ค. ์นฉ์ ์ฑ๋ฅ๋ง์ ๋ฌธ์ ๊ฐ ์๋๋๋ค. |
It is about the end-to-end solution and what the customer is able to do to find the lowest TCO for them to determine. That helps determine how we go to market and use a certain price of that piece. That TCO value is essentially looking at the full end-to-end. How would you complete the software? How would you complete your full data-center architecture? Would you need other teams? It's more than just looking at the different components. And it is not a situation where itโs the components and the cost plus model. | ์ด๋ ์๋ํฌ์๋ ์๋ฃจ์ ๊ณผ ๊ณ ๊ฐ์ด ์์ ๋ค์๊ฒ ๊ฐ์ฅ ๋ฎ์ TCO๋ฅผ ์ฐพ๊ธฐ ์ํด ํ ์ ์๋ ๊ฒ์ ๊ดํ ๋ด์ฉ์ ๋๋ค. ์ด๊ฒ์ด ์ฐ๋ฆฌ๊ฐ ์์ฅ์ ์ง์ถํ๋ ๋ฐฉ์๊ณผ ํด๋น ๋ถ๋ถ์ ํน์ ๊ฐ๊ฒฉ์ ์ฌ์ฉํ๋ ๋ฐฉ๋ฒ์ ๊ฒฐ์ ํ๋ ๋ฐ ๋์์ด ๋ฉ๋๋ค. ๊ทธ TCO ๊ฐ์น๋ ๋ณธ์ง์ ์ผ๋ก ์์ ํ ์๋ํฌ์๋๋ฅผ ์ดํด๋ณด๋ ๊ฒ์ ๋๋ค. ์ํํธ์จ์ด๋ฅผ ์ด๋ป๊ฒ ์์ฑํ ๊ฒ์ธ๊ฐ? ์ ์ฒด ๋ฐ์ดํฐ์ผํฐ ์ํคํ ์ฒ๋ฅผ ์ด๋ป๊ฒ ์์ฑํ ๊ฒ์ธ๊ฐ? ๋ค๋ฅธ ํ๋ค์ด ํ์ํ๊ฐ? ์ด๋ ๋จ์ํ ์๋ก ๋ค๋ฅธ ๊ตฌ์ฑ์์๋ค์ ์ดํด๋ณด๋ ๊ฒ ์ด์์ ๋๋ค. ๊ทธ๋ฆฌ๊ณ ์ด๋ ๊ตฌ์ฑ์์๋ค๊ณผ ๋น์ฉ ํ๋ฌ์ค ๋ชจ๋ธ์ ์ํฉ์ด ์๋๋๋ค. |
So, because of that and because of the strong performance, the strong efficiency, and the best TCO that enables us a full TCO incorporating all of the software that we provide both inside of the systems and support them throughout their lifetime. Tim Arcuri Got it. Can we talk about your Networking business? I get a lot of questions, it was down last quarter, it was expected to be up. And some people think that โWell, as you transition from InfiniBand to Ethernet that your position in networking is a little less strong.โ So can you just talk about the Networking business and should we expect it to grow alongside of the compute business? | ๊ทธ๋์ ๊ทธ๋ฐ ์ด์ ๋ก, ๊ทธ๋ฆฌ๊ณ ๊ฐ๋ ฅํ ์ฑ๋ฅ, ๋ฐ์ด๋ ํจ์จ์ฑ, ๊ทธ๋ฆฌ๊ณ ์์คํ
๋ด๋ถ์ ์ ๊ณตํ๋ ๋ชจ๋ ์ํํธ์จ์ด์ ์ ์ฒด ์๋ช
๊ธฐ๊ฐ ๋์์ ์ง์์ ํฌํจํ ์์ ํ TCO๋ฅผ ๊ฐ๋ฅํ๊ฒ ํ๋ ์ต๊ณ ์ TCO ๋๋ฌธ์
๋๋ค. Tim Arcuri ์๊ฒ ์ต๋๋ค. ๋คํธ์ํน ์ฌ์ ์ ๋ํด ์ด์ผ๊ธฐํด๋ณผ ์ ์์๊น์? ๋ง์ ์ง๋ฌธ์ ๋ฐ๊ณ ์๋๋ฐ, ์ง๋ ๋ถ๊ธฐ์ ๊ฐ์ํ๊ณ ์ฆ๊ฐํ ๊ฒ์ผ๋ก ์์๋์์ต๋๋ค. ๊ทธ๋ฆฌ๊ณ ์ผ๋ถ ์ฌ๋๋ค์ "InfiniBand์์ Ethernet์ผ๋ก ์ ํํ๋ฉด์ ๋คํธ์ํน์์์ ๊ทํ์ ํฌ์ง์ ์ด ๋ค์ ์ฝํด์ง ๊ฒ ์๋๋"๊ณ ์๊ฐํฉ๋๋ค. ๊ทธ๋์ ๋คํธ์ํน ์ฌ์ ์ ๋ํด ๋ง์ํด ์ฃผ์๊ณ , ์ปดํจํธ ์ฌ์ ๊ณผ ํจ๊ป ์ฑ์ฅํ ๊ฒ์ผ๋ก ์์ํด์ผ ํ๋์ง ์๋ ค์ฃผ์ค ์ ์๋์? |
Colette Kress Our Networking business is one of the most important additions that we added when we went to a datacenter scale. The ability to think through, not just the time where the data, the work is being done at a data processing or the use of the compute and/or the GPU is essential to think through the Networkingโs position inside of that datacenter. So we have two different offerings. We have InfiniBand and InfiniBand had tremendous success with many of the largest supercomputers in the world for decades. And that has been very important in terms of the size of data, the size of speed of data going through. | ์ฝ๋ ํฌ๋ ์ค ์ฐ๋ฆฌ์ ๋คํธ์ํน ์ฌ์ ์ ๋ฐ์ดํฐ์ผํฐ ๊ท๋ชจ๋ก ํ์ฅํ ๋ ์ถ๊ฐํ ๊ฐ์ฅ ์ค์ํ ์ฌ์ ์ค ํ๋์ ๋๋ค. ๋ฐ์ดํฐ ์ฒ๋ฆฌ๋ ์ปดํจํธ ๋ฐ/๋๋ GPU ์ฌ์ฉ์ด ์ด๋ฃจ์ด์ง๋ ์์ ๋ฟ๋ง ์๋๋ผ, ๋ฐ์ดํฐ์ผํฐ ๋ด์์ ๋คํธ์ํน์ ์์น๋ฅผ ์ข ํฉ์ ์ผ๋ก ๊ณ ๋ คํ๋ ๋ฅ๋ ฅ์ด ํ์์ ์ ๋๋ค. ์ ํฌ๋ ๋ ๊ฐ์ง ๋ค๋ฅธ ์ ํ์ ์ ๊ณตํ๊ณ ์์ต๋๋ค. ์ธํผ๋๋ฐด๋(InfiniBand)๊ฐ ์์ผ๋ฉฐ, ์ธํผ๋๋ฐด๋๋ ์์ญ ๋ ๊ฐ ์ธ๊ณ ์ต๋ ์ํผ์ปดํจํฐ๋ค๊ณผ ํจ๊ป ์์ฒญ๋ ์ฑ๊ณต์ ๊ฑฐ๋์์ต๋๋ค. ๊ทธ๋ฆฌ๊ณ ์ด๋ ๋ฐ์ดํฐ์ ๊ท๋ชจ, ๋ฐ์ดํฐ ์ ์ก ์๋ ์ธก๋ฉด์์ ๋งค์ฐ ์ค์ํ ์ญํ ์ ํด์์ต๋๋ค. |
It had different views in terms of how to deal with the traffic that will be there. Ethernet, a great configuration that is the standard for many of the enterprises. But Ethernet was not built for AI. Ethernet was just built for the networking inside of datacenters. So we are taking some of the best of the breeds of what you see in terms of inter InfiniBand and creating Ethernet for AI. That allows customers now, both the choice between those. We can be full end-to-end systems with InfiniBand and now you have your choice in terms of what we do with Ethernet. Both of these things are a growth option for us. In terms of this last quarter, we had some timing issues. | ํธ๋ํฝ ์ฒ๋ฆฌ ๋ฐฉ์์ ๋ํด์๋ ์๋ก ๋ค๋ฅธ ๊ด์ ๋ค์ด ์์์ต๋๋ค. ์ด๋๋ท์ ๋ง์ ๊ธฐ์ ๋ค์ ํ์ค์ธ ํ๋ฅญํ ๊ตฌ์ฑ์ด์ง๋ง, ์ด๋๋ท์ AI๋ฅผ ์ํด ๊ตฌ์ถ๋ ๊ฒ์ด ์๋๋๋ค. ์ด๋๋ท์ ๋จ์ง ๋ฐ์ดํฐ์ผํฐ ๋ด๋ถ์ ๋คํธ์ํน์ ์ํด ๊ตฌ์ถ๋ ๊ฒ์ ๋๋ค. ๊ทธ๋์ ์ ํฌ๋ ์ธํผ๋๋ฐด๋์์ ๋ณผ ์ ์๋ ์ต๊ณ ์ ๊ธฐ์ ๋ค์ ์ผ๋ถ ํ์ฉํ์ฌ AI์ฉ ์ด๋๋ท์ ๋ง๋ค๊ณ ์์ต๋๋ค. ์ด๋ฅผ ํตํด ๊ณ ๊ฐ๋ค์ ์ด์ ๋ ๊ฐ์ง ์ ํ๊ถ์ ๊ฐ๊ฒ ๋ฉ๋๋ค. ์ ํฌ๋ ์ธํผ๋๋ฐด๋๋ก ์์ ํ ์๋ํฌ์๋ ์์คํ ์ ์ ๊ณตํ ์ ์๊ณ , ์ด์ ์ด๋๋ท์ผ๋ก๋ ์ ํ์ ์ฌ์ง๋ฅผ ๋๋ฆด ์ ์์ต๋๋ค. ์ด ๋ ๊ฐ์ง ๋ชจ๋ ์ ํฌ์๊ฒ๋ ์ฑ์ฅ ์ต์ ์ ๋๋ค. ์ง๋ ๋ถ๊ธฐ์ ๊ฒฝ์ฐ, ์ ํฌ๋ ์ผ๋ถ ํ์ด๋ฐ ์ด์๊ฐ ์์์ต๋๋ค. |
But now, what you will see in terms of the continuation of our Networking will definitely grow. With our designs in terms of Networking with our compute, we have some of the strongest clusters that are being built and also using our Networking. That connection that we have done has been a very important part of the work that we've done since the acquisition of Mellanox. Folks do represent and understand our use of networking and how that can help their overall system as a whole. Tim Arcuri Great. Can we talk about scaling of these large language models? There's been some articles written that Google and OpenAI are having a hard time to get better results out of these larger models. | ํ์ง๋ง ์ด์ ๋คํธ์ํน ์ฌ์
์ ์ง์์ ์ธ ์ฑ์ฅ์ ํ์คํ ๋ณด์ค ์ ์์ ๊ฒ์
๋๋ค. ์ปดํจํ
๊ณผ ๋คํธ์ํน ์ค๊ณ ์ธก๋ฉด์์, ์ ํฌ๋ ๋คํธ์ํน์ ํ์ฉํ์ฌ ๊ตฌ์ถ๋๊ณ ์๋ ๊ฐ์ฅ ๊ฐ๋ ฅํ ํด๋ฌ์คํฐ๋ค ์ค ์ผ๋ถ๋ฅผ ๋ณด์ ํ๊ณ ์์ต๋๋ค. ์ ํฌ๊ฐ ๊ตฌ์ถํ ์ด๋ฌํ ์ฐ๊ฒฐ์ฑ์ ๋ฉ๋ผ๋
น์ค ์ธ์ ์ดํ ์งํํด์จ ์์
์ ๋งค์ฐ ์ค์ํ ๋ถ๋ถ์ด์์ต๋๋ค. ๊ณ ๊ฐ๋ค์ ์ ํฌ์ ๋คํธ์ํน ํ์ฉ๋ฒ๊ณผ ์ด๊ฒ์ด ์ ์ฒด ์์คํ
์ ์ด๋ป๊ฒ ๋์์ด ๋ ์ ์๋์ง๋ฅผ ์ดํดํ๊ณ ์ธ์ ํ๊ณ ์์ต๋๋ค. **ํ ์์ฟ ๋ฆฌ(Tim Arcuri)** ์ข์ต๋๋ค. ์ด๋ฌํ ๋๊ท๋ชจ ์ธ์ด ๋ชจ๋ธ์ ์ค์ผ์ผ๋ง์ ๋ํด ์ด์ผ๊ธฐํด๋ณผ ์ ์์๊น์? ๊ตฌ๊ธ๊ณผ OpenAI๊ฐ ์ด๋ฌํ ๋ํ ๋ชจ๋ธ์์ ๋ ๋์ ๊ฒฐ๊ณผ๋ฅผ ์ป๋ ๋ฐ ์ด๋ ค์์ ๊ฒช๊ณ ์๋ค๋ ๊ธฐ์ฌ๋ค์ด ๋์ค๊ณ ์๋๋ฐ์. |
But on the other hand, you had Meta on their earnings call and you had others like Anthropic saying that the scaling is of live and well. So can you just speak to that? I know that there's some nuances in post-training and of course, there's the test time difference with the โ some of these new models from OpenAI. Is the scaling question something that investors should be thinking about? Colette Kress When we look at what we are seeing in the size of clusters that are being built and the work that many of our customers are looking to do, the scaling laws that we see particularly in terms of training, they're still here. | ํ์ง๋ง ํํธ์ผ๋ก๋ Meta๊ฐ ์ค์ ๋ฐํ์์, ๊ทธ๋ฆฌ๊ณ Anthropic๊ณผ ๊ฐ์ ๋ค๋ฅธ ํ์ฌ๋ค์ด ์ค์ผ์ผ๋ง์ด ์ฌ์ ํ ์ด์์๊ณ ์ ์๋ํ๊ณ ์๋ค๊ณ ๋งํ์ต๋๋ค. ์ด์ ๋ํด ๋ง์ํด ์ฃผ์ค ์ ์๋์? ํฌ์คํธ ํธ๋ ์ด๋์์ ์ผ๋ถ ๋์์ค๊ฐ ์๊ณ , ๋ฌผ๋ก OpenAI์ ์ผ๋ถ ์๋ก์ด ๋ชจ๋ธ๋ค๊ณผ ๊ด๋ จ๋ ํ
์คํธ ํ์ ์ฐจ์ด๋ ์๋ค๋ ๊ฒ์ ์๊ณ ์์ต๋๋ค. ์ค์ผ์ผ๋ง ๋ฌธ์ ๊ฐ ํฌ์์๋ค์ด ๊ณ ๋ คํด์ผ ํ ์ฌ์์ธ๊ฐ์? ์ฝ๋ ํฌ๋ ์ค ์ฐ๋ฆฌ๊ฐ ๊ตฌ์ถ๋๊ณ ์๋ ํด๋ฌ์คํฐ์ ๊ท๋ชจ์ ๋ง์ ๊ณ ๊ฐ๋ค์ด ์ํํ๊ณ ์ ํ๋ ์์ ์ ๋ณด๋ฉด, ํนํ ํธ๋ ์ด๋ ์ธก๋ฉด์์ ์ฐ๋ฆฌ๊ฐ ๋ณด๊ณ ์๋ ์ค์ผ์ผ๋ง ๋ฒ์น๋ค์ ์ฌ์ ํ ์ ํจํฉ๋๋ค. |
I think you will see more and more larger models and complex models in this next-generation of Blackwell. What that means is, there is this phase of post-training that's coming back with reinforcement learning that are truly looking for the human piece of it and also using in terms of synthetic data to fine-tune that models. Another way of saying that is trainingโs never done, and there's a lot of work that continues. But there's also been new skilling laws that is focused on the inferencing phase. If you recall, we are the largest inferencing provider that exist today. We do the most inferencing versus any other different types of configuration, why? it's very hard. | ๋ค์ ์ธ๋ Blackwell์์๋ ์ ์ ๋ ํฌ๊ณ ๋ณต์กํ ๋ชจ๋ธ๋ค์ ๋ณด๊ฒ ๋ ๊ฒ์ด๋ผ๊ณ ์๊ฐํฉ๋๋ค. ์ด๊ฒ์ด ์๋ฏธํ๋ ๋ฐ๋, ๊ฐํํ์ต์ ํตํ ์ฌํ ํ๋ จ(post-training) ๋จ๊ณ๊ฐ ๋ฑ์ฅํ๊ณ ์์ผ๋ฉฐ, ์ด๋ ์ง์ ์ผ๋ก ์ธ๊ฐ์ ์์๋ฅผ ์ฐพ๊ณ ์๊ณ ๋ํ ํด๋น ๋ชจ๋ธ๋ค์ ๋ฏธ์ธ ์กฐ์ ํ๊ธฐ ์ํด ํฉ์ฑ ๋ฐ์ดํฐ๋ฅผ ํ์ฉํ๊ณ ์๋ค๋ ๊ฒ์ ๋๋ค. ๋ค์ ๋งํด์, ํ๋ จ์ ๊ฒฐ์ฝ ๋๋์ง ์์ผ๋ฉฐ ๊ณ์ํด์ ๋ง์ ์์ ์ด ์ด์ด์ง๋๋ค. ํ์ง๋ง ์ถ๋ก (inferencing) ๋จ๊ณ์ ์ด์ ์ ๋ง์ถ ์๋ก์ด ์ค์ผ์ผ๋ง ๋ฒ์น๋ค๋ ๋ฑ์ฅํ์ต๋๋ค. ๊ธฐ์ตํ์๊ฒ ์ง๋ง, ์ฐ๋ฆฌ๋ ํ์ฌ ์กด์ฌํ๋ ๊ฐ์ฅ ํฐ ์ถ๋ก ์ ๊ณต์ ์ฒด์ ๋๋ค. ์ฐ๋ฆฌ๋ ๋ค๋ฅธ ์ด๋ค ์ ํ์ ๊ตฌ์ฑ๋ณด๋ค๋ ๊ฐ์ฅ ๋ง์ ์ถ๋ก ์ ์ํํฉ๋๋ค. ์๋ํ๋ฉด ๋งค์ฐ ์ด๋ ต๊ธฐ ๋๋ฌธ์ ๋๋ค. |
And what we are seeing from a scaling, is an important part from the onset of Generative AI to now, more are looking in terms of the focus in terms of reasoning for deep thinking and taking the time to do that. That is still going to now require additional amount of compute and a compute that is able to do at the least amount of latency for that time that you will spend in terms of the reasoning factor of it. So we still see those scaling laws still being important and more and more new laws will be probably formed over the next decades. Tim Arcuri You talk about a new and emerging piece of the demand picture, which is more in government-backed projects, Sovereign as we call them. | ๊ทธ๋ฆฌ๊ณ ํ์ฅ์ฑ ์ธก๋ฉด์์ ์ฐ๋ฆฌ๊ฐ ๋ณด๊ณ ์๋ ๊ฒ์, ์์ฑํ AI ์ด๊ธฐ๋ถํฐ ์ง๊ธ๊น์ง ์ค์ํ ๋ถ๋ถ์ผ๋ก, ๋ ๋ง์ ๊ธฐ์
๋ค์ด ์ฌ์ธต์ ์ฌ๊ณ ๋ฅผ ์ํ ์ถ๋ก ๊ณผ ๊ทธ์ ํ์ํ ์๊ฐ ํฌ์์ ์ด์ ์ ๋ง์ถ๊ณ ์๋ค๋ ์ ์
๋๋ค. ์ด๋ ์ฌ์ ํ ์ถ๊ฐ์ ์ธ ์ปดํจํ
ํ์๋ฅผ ํ์๋ก ํ๋ฉฐ, ์ถ๋ก ๊ณผ์ ์์ ์์๋๋ ์๊ฐ ๋์ ์ต์ํ์ ์ง์ฐ์๊ฐ์ผ๋ก ์ฒ๋ฆฌํ ์ ์๋ ์ปดํจํ
๋ฅ๋ ฅ์ด ์๊ตฌ๋ฉ๋๋ค. ๋ฐ๋ผ์ ์ฐ๋ฆฌ๋ ์ด๋ฌํ ํ์ฅ ๋ฒ์น๋ค์ด ์ฌ์ ํ ์ค์ํ๋ฉฐ, ํฅํ ์์ญ ๋
๋์ ๋ ๋ง์ ์๋ก์ด ๋ฒ์น๋ค์ด ํ์ฑ๋ ๊ฒ์ผ๋ก ๋ณด๊ณ ์์ต๋๋ค. ํ ์๋ฅด์ฟ ๋ฆฌ(Tim Arcuri): ์์ ์ ๋ง์์ ์๋กญ๊ฒ ๋ถ์ํ๋ ๋ถ๋ถ์ ๋ํด ๋ง์ํด ์ฃผ์ จ๋๋ฐ, ์ด๋ ์ ๋ถ ์ง์ ํ๋ก์ ํธ, ์ฆ ์ฐ๋ฆฌ๊ฐ ์๋ฒ๋ฆฐ(Sovereign) ํ๋ก์ ํธ๋ผ๊ณ ๋ถ๋ฅด๋ ๋ถ์ผ์์ ๋ ๋ง์ด ๋ํ๋๊ณ ์์ต๋๋ค. |
You said that you're going to do double-digit billion dollars this year in that - for those projects as a whole. Can you talk about some of the examples of where that demand is coming from? And sort of how to think about how big it could be? I kind of think of it as well maybe some of these larger projects in the Middle East could eventually be as large as a US BSP. So this could be a very large piece of demand and I am wondering how you sort of think about that you could get? Colette Kress So Sovereign AI has been a very interesting part of what we've seen in terms of Generative AI. | ์ฝ๋ ํธ ํฌ๋ ์ค ์๋ฒ๋ฆฐ AI๋ ์์ฑํ AI์ ๊ด๋ จํ์ฌ ์ฐ๋ฆฌ๊ฐ ๋ชฉ๊ฒฉํ ๋งค์ฐ ํฅ๋ฏธ๋ก์ด ๋ถ๋ถ์ด์์ต๋๋ค. --- **์ง๋ฌธ (๋ฒ์ญ):** ์ฌํด ํด๋น ํ๋ก์ ํธ๋ค ์ ์ฒด์ ์ผ๋ก ์๋ฐฑ์ต ๋ฌ๋ฌ ๊ท๋ชจ์ ๋งค์ถ์ ๋ฌ์ฑํ ๊ฒ์ด๋ผ๊ณ ๋ง์ํ์ จ๋๋ฐ, ๊ทธ๋ฌํ ์์๊ฐ ์ด๋์ ๋์ค๊ณ ์๋์ง ๋ช ๊ฐ์ง ์ฌ๋ก๋ฅผ ๋ง์ํด ์ฃผ์ค ์ ์๋์? ๊ทธ๋ฆฌ๊ณ ๊ทธ ๊ท๋ชจ๊ฐ ์ผ๋ง๋ ํด ์ ์๋์ง์ ๋ํด ์ด๋ป๊ฒ ์๊ฐํด์ผ ํ ๊น์? ์ค๋์ ์ผ๋ถ ๋๊ท๋ชจ ํ๋ก์ ํธ๋ค์ด ๊ฒฐ๊ตญ ๋ฏธ๊ตญ BSP๋งํผ ํด ์๋ ์๋ค๊ณ ์๊ฐํ๋๋ฐ, ์ด๋ ๋งค์ฐ ํฐ ์์๊ฐ ๋ ์ ์๊ณ , ๊ทํ๊ป์ ์ด๋ฅผ ์ด๋ป๊ฒ ์๊ฐํ๊ณ ๊ณ์ ์ง ๊ถ๊ธํฉ๋๋ค. |
Very simply said, what they saw here that we had in the US, every country, every country that has a duty to see when I want that too, okay? And they want a model, a foundational model in their own language, in their own culture to support their nation. As they see the importance of what AI will be in the next decades going forward. So, the amount of different countries that we are working with or even working in certain regions as you focused, is a very large part of our work that we are doing, spending around globally of AI. It is not just a West Coast of the US. It is really taking part around the world. Not all of it and even only parts of it are a government funding. | ์์ฃผ ๊ฐ๋จํ ๋ง์๋๋ฆฌ๋ฉด, ๋ฏธ๊ตญ์์ ์ฐ๋ฆฌ๊ฐ ๋ณด์ ํ ๊ฒ์ ๋ค๋ฅธ ๋๋ผ๋ค์ด ๋ณด๊ณ ์๊ณ , ๋ชจ๋ ๊ตญ๊ฐ, ์๋ฌด๋ฅผ ๊ฐ์ง ๋ชจ๋ ๊ตญ๊ฐ๋ค์ด ๊ทธ๊ฒ์ ์ํ๊ณ ์์ต๋๋ค. ๊ทธ๋ค๋ ์๊ตญ์ ์ธ์ด์ ๋ฌธํ๋ก ๋, ์๊ตญ์ ์ง์ํ ์ ์๋ ๋ชจ๋ธ, ์ฆ ํ์ด๋ฐ์ด์
๋ชจ๋ธ์ ์ํ๊ณ ์์ต๋๋ค. ํฅํ ์์ญ ๋
๊ฐ AI๊ฐ ์ผ๋ง๋ ์ค์ํ ์ง๋ฅผ ๊ทธ๋ค์ด ์ธ์ํ๊ณ ์๊ธฐ ๋๋ฌธ์
๋๋ค. ๋ฐ๋ผ์ ์ฐ๋ฆฌ๊ฐ ํ๋ ฅํ๊ณ ์๊ฑฐ๋ ํน์ ์ง์ญ์์ ์์ ํ๊ณ ์๋ ๋ค์ํ ๊ตญ๊ฐ๋ค์ ์๋, ์ฐ๋ฆฌ๊ฐ ์ ์ธ๊ณ์ ์ผ๋ก AI ํ์ฐ์ ์ํด ์ํํ๊ณ ์๋ ์ ๋ฌด์ ๋งค์ฐ ํฐ ๋ถ๋ถ์ ์ฐจ์งํฉ๋๋ค. ์ด๋ ๋จ์ํ ๋ฏธ๊ตญ ์๋ถ ์ง์ญ์๋ง ๊ตญํ๋ ๊ฒ์ด ์๋๋๋ค. ์ค์ ๋ก ์ ์ธ๊ณ์ ์ผ๋ก ์ฐธ์ฌํ๊ณ ์์ต๋๋ค. ๋ชจ๋ ๊ฒ์ด ์ ๋ถ ์๊ธ ์ง์์ ์๋๋ฉฐ, ์ผ๋ถ๋ง์ด ์ ๋ถ ์๊ธ ์ง์์ ๋ฐ๊ณ ์์ต๋๋ค. |
Many of them are looking at very large companies that will start a new type of regional CSPs that they're able to support accelerated computing and may have a set of tenets. And weโll likely have a foundational model with them in order to support the enterprises. You've seen our talk in terms of what is happening in Japan. There is an area where SoftBank is very interested in building a very, very sizable model. You see, in terms of India, and many of the CSPs there working in terms of what they will incorporate as well. That moves all the way to the Europe. It moves to the Nordics and itโs also very important part in the Asia-pacific area. | ์ด๋ค ์ค ๋ค์๋ ๊ฐ์ ์ปดํจํ ์ ์ง์ํ ์ ์๋ ์๋ก์ด ์ ํ์ ์ง์ญ CSP(ํด๋ผ์ฐ๋ ์๋น์ค ์ ๊ณต์ ์ฒด)๋ฅผ ์์ํ ๋งค์ฐ ๋๊ท๋ชจ ๊ธฐ์ ๋ค์ ๊ฒํ ํ๊ณ ์์ผ๋ฉฐ, ์ผ๋ จ์ ์์น์ ๊ฐ์ง๊ณ ์์ ๊ฒ์ ๋๋ค. ๊ทธ๋ฆฌ๊ณ ๊ธฐ์ ๋ค์ ์ง์ํ๊ธฐ ์ํด ์ด๋ค๊ณผ ํจ๊ป ํ์ด๋ฐ์ด์ ๋ชจ๋ธ์ ๋ณด์ ํ๊ฒ ๋ ๊ฐ๋ฅ์ฑ์ด ๋์ต๋๋ค. ์ผ๋ณธ์์ ์ผ์ด๋๊ณ ์๋ ์ํฉ์ ๋ํ ์ ํฌ์ ๋ ผ์๋ฅผ ๋ณด์ จ์ ๊ฒ์ ๋๋ค. ์ํํธ๋ฑ ํฌ๊ฐ ๋งค์ฐ ๋๊ท๋ชจ ๋ชจ๋ธ ๊ตฌ์ถ์ ํฐ ๊ด์ฌ์ ๋ณด์ด๊ณ ์๋ ์์ญ์ด ์์ต๋๋ค. ์ธ๋์ ๊ฒฝ์ฐ์๋ ๋ณด์๋ฉด, ๊ทธ๊ณณ์ ๋ง์ CSP๋ค์ด ๋ฌด์์ ํตํฉํ ์ง์ ๋ํด ์์ ํ๊ณ ์์ต๋๋ค. ์ด๋ ์ ๋ฝ ์ ์ฒด๋ก ํ์ฐ๋๊ณ ์์ต๋๋ค. ๋ถ์ ๋ฝ ์ง์ญ์ผ๋ก๋ ํ์ฐ๋๊ณ ์์ผ๋ฉฐ, ์์์-ํํ์ ์ง์ญ์์๋ ๋งค์ฐ ์ค์ํ ๋ถ๋ถ์ ๋๋ค. |
So this will continue some parts of it that you want to think about is the next-generation of what we saw in terms of supercomputing in each of those countries. You will see from GDP what they need to do for AI and the Sovereign AI. Tim Arcuri Great. Can you talk about your Software business for just a minute or so? You had said that it's going to be crossing over the $2 billion a year runrate exiting this year. If I sort of do some back of the envelope of math and I try to figure out how many of your GPUs are you directly monetizing on for your modification of those. | ์ด๋ ๊ฐ ๊ตญ๊ฐ์์ ์ํผ์ปดํจํ
์ธก๋ฉด์์ ์ฐ๋ฆฌ๊ฐ ๋ณธ ์ฐจ์ธ๋ ๊ธฐ์ ์ ์ผ๋ถ๊ฐ ๊ณ์๋ ๊ฒ์
๋๋ค. GDP ๊ด์ ์์ ๊ทธ๋ค์ด AI์ ์๋ฒ๋ฆฐ AI๋ฅผ ์ํด ๋ฌด์์ ํด์ผ ํ๋์ง ๋ณด์ค ์ ์์ ๊ฒ์
๋๋ค. Tim Arcuri ์ข์ต๋๋ค. ์ํํธ์จ์ด ์ฌ์ ์ ๋ํด ์ ๊น ๋ง์ํด ์ฃผ์ค ์ ์๋์? ์ฌํด ๋ง ๊ธฐ์ค์ผ๋ก ์ฐ๊ฐ 20์ต ๋ฌ๋ฌ ๋ฐ๋ ์ดํธ๋ฅผ ๋์ด์ค ๊ฒ์ด๋ผ๊ณ ๋ง์ํ์ จ๋๋ฐ์. ์ ๊ฐ ๋๋ต์ ์ผ๋ก ๊ณ์ฐํด๋ณด๊ณ ๊ทํ์ GPU ์ค ์ผ๋ง๋ ๋ง์ ์๋ฅผ ์ง์ ์์ตํํ๊ณ ๊ณ์ ์ง ํ์ ํด๋ณด๋ ค๊ณ ํฉ๋๋ค. |
I get something like 10% attach rate, where you're roughly directly licensing 10% of your So can you talk about sort of how you think about the attach rate, and how and how successful you are and directly licensing software for your GPUs? Colette Kress Yeah. Our software platform is so essential to many of our enterprises and many of our regional CSPs, our future AI factories, our AI foundations work that we're doing. Why is that the case? | ์ ๋ ๋๋ต 10% ์ ๋์ ๋ถ์ฐฉ๋ฅ (attach rate)์ ๋ณด๊ณ ์๋๋ฐ, ์ด๋ GPU์ ๋ํด ์ง์ ๋ผ์ด์ ์ฑํ๋ ๋น์จ์ด ๋๋ต 10% ์ ๋๋ผ๋ ์๋ฏธ์
๋๋ค. ๋ถ์ฐฉ๋ฅ ์ ๋ํด ์ด๋ป๊ฒ ์๊ฐํ์๋์ง, ๊ทธ๋ฆฌ๊ณ GPU์ฉ ์ํํธ์จ์ด ์ง์ ๋ผ์ด์ ์ฑ์์ ์ผ๋ง๋ ์ฑ๊ณต์ ์ธ์ง์ ๋ํด ๋ง์ํด ์ฃผ์ค ์ ์๋์? ์ฝ๋ ํธ ํฌ๋ ์ค(Colette Kress): ๋ค. ์ ํฌ ์ํํธ์จ์ด ํ๋ซํผ์ ๋ง์ ๊ธฐ์ ๋ค๊ณผ ์ง์ญ CSP(ํด๋ผ์ฐ๋ ์๋น์ค ์ ๊ณต์ ์ฒด)๋ค, ๋ฏธ๋์ AI ํฉํ ๋ฆฌ๋ค, ๊ทธ๋ฆฌ๊ณ ์ ํฌ๊ฐ ์งํํ๊ณ ์๋ AI ํ์ด๋ฐ์ด์ ์์ ์ ๋งค์ฐ ํ์์ ์ ๋๋ค. ์ ๊ทธ๋ฐ์ง ๋ง์๋๋ฆฌ๊ฒ ์ต๋๋ค. |
What you have is a situation where that first steps of understanding how to move and get started on AI, we have thousands of different applications, as well as CUDA libraries and CUDA work that we have done for each industry, as well as each major workload within those industries. Your enterprises have to have that piece, not only for the work that they need to do to do on their own, but their work that they need to do to support the infrastructure in that datacenter. So, we are building that for them and in many cases, your enterprise customers will have a very strong attachment to that software as they will need that for the work that they're doing. | ์ฌ๋ฌ๋ถ์ด ์ง๋ฉดํ ์ํฉ์ AI๋ฅผ ์ด๋ป๊ฒ ํ์ฉํ๊ณ ์์ํ ์ง์ ๋ํ ์ฒซ ๋ฒ์งธ ๋จ๊ณ๋ฅผ ์ดํดํ๋ ๊ฒ์ธ๋ฐ, ์ ํฌ๋ ์์ฒ ๊ฐ์ ๋ค์ํ ์ ํ๋ฆฌ์ผ์ด์ ๊ณผ CUDA ๋ผ์ด๋ธ๋ฌ๋ฆฌ, ๊ทธ๋ฆฌ๊ณ ๊ฐ ์ฐ์ ๋ณ๋ก, ๋ํ ํด๋น ์ฐ์ ๋ด ์ฃผ์ ์ํฌ๋ก๋๋ณ๋ก ์ ํฌ๊ฐ ๊ฐ๋ฐํ CUDA ์์ ๋ค์ ๋ณด์ ํ๊ณ ์์ต๋๋ค. ๊ธฐ์ ๊ณ ๊ฐ๋ค์ ์์ฒด์ ์ผ๋ก ์ํํด์ผ ํ๋ ์์ ๋ฟ๋ง ์๋๋ผ ๋ฐ์ดํฐ์ผํฐ ๋ด ์ธํ๋ผ๋ฅผ ์ง์ํ๊ธฐ ์ํด ํ์ํ ์์ ์ ์ํด์๋ ์ด๋ฌํ ์์๋ค์ ๋ฐ๋์ ๊ฐ์ถฐ์ผ ํฉ๋๋ค. ๋ฐ๋ผ์ ์ ํฌ๋ ์ด๋ฅผ ์ํด ๊ตฌ์ถํ๊ณ ์์ผ๋ฉฐ, ๋ง์ ๊ฒฝ์ฐ ๊ธฐ์ ๊ณ ๊ฐ๋ค์ ์์ ๋ค์ด ์ํํ๋ ์์ ์ ํ์ํ๊ธฐ ๋๋ฌธ์ ํด๋น ์ํํธ์จ์ด์ ๋ํด ๋งค์ฐ ๊ฐํ ์์กด์ฑ์ ๊ฐ๊ฒ ๋ ๊ฒ์ ๋๋ค. |
Those that have very, very large, software teams and as I have been self-building for several decades is different. But as you can imagine the world, can't go back to building all of those software engineers. And so we have spent a very quality amount of time helping a lot of enterprises. It's working with the enterprises from the onset of them choosing what type of compute they're doing to the delivery to helping them with their models and setting up all of their different apps and all of the overall inferencing where there are the entire way. So, it's more than just the actual software in there. That software also comes with true support and services from the company. | ๋งค์ฐ ๋๊ท๋ชจ์ ์ํํธ์จ์ด ํ์ ๋ณด์ ํ๊ณ ์์ผ๋ฉฐ ์์ญ ๋ ๊ฐ ์์ฒด ๊ตฌ์ถ์ ํด์จ ๊ธฐ์ ๋ค์ ๊ฒฝ์ฐ๋ ๋ค๋ฆ ๋๋ค. ํ์ง๋ง ์์๋ค์ํผ ์ ์ธ๊ณ๊ฐ ๋ชจ๋ ์ํํธ์จ์ด ์์ง๋์ด๋ฅผ ๋ค์ ๊ตฌ์ถํ๋ ๋ฐฉํฅ์ผ๋ก ๋์๊ฐ ์๋ ์์ต๋๋ค. ๋ฐ๋ผ์ ์ ํฌ๋ ๋ง์ ๊ธฐ์ ๋ค์ ์ง์ํ๋ ๋ฐ ์๋นํ ์๊ฐ์ ํฌ์ํด์์ต๋๋ค. ๊ธฐ์ ๋ค์ด ์ด๋ค ์ ํ์ ์ปดํจํ ์ ์ ํํ ์ง ๊ฒฐ์ ํ๋ ์ด๊ธฐ ๋จ๊ณ๋ถํฐ ๋ฐฐํฌ, ๋ชจ๋ธ ์ง์, ๋ค์ํ ์ ํ๋ฆฌ์ผ์ด์ ์ค์ , ๊ทธ๋ฆฌ๊ณ ์ ์ฒด์ ์ธ ์ถ๋ก ๊ณผ์ ์ ์ด๋ฅด๊ธฐ๊น์ง ์ ๊ณผ์ ์ ๊ฑธ์ณ ๊ธฐ์ ๋ค๊ณผ ํ๋ ฅํ๊ณ ์์ต๋๋ค. ๋ฐ๋ผ์ ์ด๋ ๋จ์ํ ์ํํธ์จ์ด ๊ทธ ์์ฒด๋ฅผ ๋์ด์๋ ๊ฒ์ ๋๋ค. ์ด ์ํํธ์จ์ด์๋ ํ์ฌ์ ์ง์ ํ ์ง์๊ณผ ์๋น์ค๋ ํจ๊ป ์ ๊ณต๋ฉ๋๋ค. |
Tim Arcuri So we looked just the enterprise market, your attach rate could be actually quite a bit higher than that? Colette Kress Absolutely. Thatโs correct. Tim Arcuri Got it. Great. We all talk about datacentre, but inference at the Edge is going to become a much bigger theme. So can you talk about your position at the Edge? You have a large installed base in PC. That should play pretty well for you and you have Omniverse. Robotics is a huge theme. So can you talk about some of your offerings and how we should think of you as a player at the Edge? Colette Kress Yes, Edge Computing, Edge AI, very similar will likely go hand-in-hand and be there. | Tim Arcuri ๊ทธ๋์ ๊ธฐ์ ์์ฅ๋ง ๋ณด๋ฉด, ์ค์ ๋ก ๊ทํ์ ์ดํ์น ๋ ์ดํธ(attach rate)๊ฐ ๊ทธ๋ณด๋ค ํจ์ฌ ๋์ ์ ์๋ค๋ ๋ง์์ด์ ๊ฐ์? Colette Kress ๋ง์ต๋๋ค. ๊ทธ๋ ์ต๋๋ค. Tim Arcuri ์๊ฒ ์ต๋๋ค. ํ๋ฅญํฉ๋๋ค. ์ฐ๋ฆฌ ๋ชจ๋ ๋ฐ์ดํฐ์ผํฐ์ ๋ํด ์ด์ผ๊ธฐํ์ง๋ง, ์ฃ์ง์์์ ์ถ๋ก (inference)์ด ํจ์ฌ ๋ ํฐ ํ ๋ง๊ฐ ๋ ๊ฒ์ ๋๋ค. ๊ทธ๋์ ์ฃ์ง์์์ ๊ทํ์ ํฌ์ง์ ์ ๋ํด ๋ง์ํด ์ฃผ์ค ์ ์๋์? PC์์ ๋๊ท๋ชจ ์ค์น ๊ธฐ๋ฐ์ ๋ณด์ ํ๊ณ ๊ณ์์ฃ . ์ด๋ ๊ทํ์๊ฒ ๊ฝค ์ ๋ฆฌํ๊ฒ ์์ฉํ ๊ฒ์ด๊ณ , ์ด๋๋ฒ์ค(Omniverse)๋ ์์ต๋๋ค. ๋ก๋ณดํฑ์ค๋ ๊ฑฐ๋ํ ํ ๋ง์ ๋๋ค. ๊ทธ๋์ ๊ทํ์ ์ผ๋ถ ์ ํ๋ค๊ณผ ์ฃ์ง ํ๋ ์ด์ด๋ก์ ๊ทํ๋ฅผ ์ด๋ป๊ฒ ์๊ฐํด์ผ ํ๋์ง์ ๋ํด ๋ง์ํด ์ฃผ์ค ์ ์๋์? Colette Kress ๋ค, ์ฃ์ง ์ปดํจํ , ์ฃ์ง AI๋ ๋งค์ฐ ์ ์ฌํ๋ฉฐ ํจ๊ป ๋ฐ์ ํ๊ณ ๊ทธ๊ณณ์ ์กด์ฌํ ๊ฐ๋ฅ์ฑ์ด ๋์ต๋๋ค. |
What that means is, you will have factories. You will see folks in their datacenter collecting data and providing that data to go into many of the overall Edge appliances, Edge appliances that you may think, the cars. The cars that are autonomous-driven. The next phase will likely be in terms of the robotics. A very, very big industry where the data and the learning is happening back in the datacenter and inside of those different devices includes our capabilities, to support them. It's an important industry. We do know the - I would say the datacenter piece of that is a very large market and very important for what we will see going forward. | ์ด๊ฒ์ด ์๋ฏธํ๋ ๋ฐ๋, ์ฌ๋ฌ๋ถ์ด ๊ณต์ฅ๋ค์ ๋ณด๊ฒ ๋ ๊ฒ์ด๋ผ๋ ์ ์ ๋๋ค. ๋ฐ์ดํฐ์ผํฐ์์ ๋ฐ์ดํฐ๋ฅผ ์์งํ๊ณ ๊ทธ ๋ฐ์ดํฐ๋ฅผ ๋ค์ํ ์ ์ฒด ์ฃ์ง ์ดํ๋ผ์ด์ธ์ค๋ก ์ ๊ณตํ๋ ์ฌ๋๋ค์ ๋ณด๊ฒ ๋ ๊ฒ์ ๋๋ค. ์ฌ๋ฌ๋ถ์ด ์๊ฐํ ์ ์๋ ์ฃ์ง ์ดํ๋ผ์ด์ธ์ค, ์ฆ ์๋์ฐจ๋ค ๋ง์ ๋๋ค. ์์จ์ฃผํ ์๋์ฐจ๋ค ๋ง์ด์ฃ . ๋ค์ ๋จ๊ณ๋ ๋ก๋ณดํฑ์ค ์ธก๋ฉด์์ ๋ํ๋ ๊ฐ๋ฅ์ฑ์ด ๋์ต๋๋ค. ๋ฐ์ดํฐ์ ํ์ต์ด ๋ฐ์ดํฐ์ผํฐ์์ ๊ทธ๋ฆฌ๊ณ ์ด๋ฌํ ๋ค์ํ ๋๋ฐ์ด์ค ๋ด๋ถ์์ ์ผ์ด๋๋ ๋งค์ฐ, ๋งค์ฐ ํฐ ์ฐ์ ์ด๋ฉฐ, ์ด๋ค์ ์ง์ํ๊ธฐ ์ํ ์ฐ๋ฆฌ์ ์ญ๋์ด ํฌํจ๋์ด ์์ต๋๋ค. ์ด๋ ์ค์ํ ์ฐ์ ์ ๋๋ค. ์ฐ๋ฆฌ๋ - ์ด ์ค์์ ๋ฐ์ดํฐ์ผํฐ ๋ถ๋ถ์ด ๋งค์ฐ ํฐ ์์ฅ์ด๋ฉฐ ์์ผ๋ก ์ฐ๋ฆฌ๊ฐ ๋ณด๊ฒ ๋ ๊ฒ์ ์์ด ๋งค์ฐ ์ค์ํ๋ค๋ ์ ์ ์๊ณ ์์ต๋๋ค. |
That incorporates even a new set of software that we're doing. As, you know, we're doing the software for autonomous vehicles that will come to market later in this next calendar year. Additionally, when you think about the work that we can do inside of the robotics and also from that software and the work that we can do with many of the factories with Omniverse and the overall layout of how that will work. So, these are very strong areas of focus even outside of just a standard datacentre. But yes, Edge Computing will be an important piece too. Tim Arcuri Great. Can we talk just for a minute about inference versus training? You have been saying that inference is about 40% of your revenue. | ์ด๋ ์ฐ๋ฆฌ๊ฐ ๊ฐ๋ฐํ๊ณ ์๋ ์๋ก์ด ์ํํธ์จ์ด ์ธํธ๊น์ง ํฌํจํ๋ ๊ฒ์
๋๋ค. ์์๋ค์ํผ, ์ฐ๋ฆฌ๋ ๋ค์ ์ฐ๋ ํ๋ฐ์ ์์ฅ์ ์ถ์๋ ์์จ์ฃผํ์ฐจ์ฉ ์ํํธ์จ์ด๋ฅผ ๊ฐ๋ฐํ๊ณ ์์ต๋๋ค. ๋ํ ๋ก๋ณดํฑ์ค ๋ถ์ผ์์ ํ ์ ์๋ ์์
๊ณผ ํด๋น ์ํํธ์จ์ด๋ก๋ถํฐ, ๊ทธ๋ฆฌ๊ณ Omniverse๋ฅผ ํตํด ๋ง์ ๊ณต์ฅ๋ค๊ณผ ํจ๊ป ํ ์ ์๋ ์์
๋ฐ ์ด๊ฒ์ด ์ด๋ป๊ฒ ์๋ํ ์ง์ ๋ํ ์ ๋ฐ์ ์ธ ๋ ์ด์์์ ์๊ฐํด๋ณด๋ฉด ๋ง์
๋๋ค. ๋ฐ๋ผ์ ์ด๋ฌํ ์์ญ๋ค์ ์ผ๋ฐ์ ์ธ ๋ฐ์ดํฐ์ผํฐ๋ฅผ ๋์ด์๋ ๋งค์ฐ ๊ฐ๋ ฅํ ์ง์ค ๋ถ์ผ์
๋๋ค. ํ์ง๋ง ๋ค, ์ฃ์ง ์ปดํจํ
๋ ์ค์ํ ๋ถ๋ถ์ด ๋ ๊ฒ์
๋๋ค. Tim Arcuri ์ข์ต๋๋ค. ์ถ๋ก ๋ ํ๋ จ์ ๋ํด ์ ์ ์ด์ผ๊ธฐํด๋ณผ ์ ์์๊น์? ์ถ๋ก ์ด ๋งค์ถ์ ์ฝ 40%๋ฅผ ์ฐจ์งํ๋ค๊ณ ๋ง์ํ์ จ๋๋ฐ์. |
Can you talk about how you see that evolving? Colette Kress It is about 40%, when we had communicated that we start thinking through what we are seeing the use cases for what they're doing. We see a lot of time spent in terms of the inferencing. And this is even before you were seeing a lot of the Generative AI applications that are still in the worst to be put out there. So, with the recommend - recommender engines, that is a very significant part of the inferencing today. So our growth of that 40% will likely be seen as we move forward. But as I discussed earlier, we are still the largest in terms of inferencing. | ์ฝ๋ ํฌ๋ ์ค ์ฝ 40%์ ๋๋ค. ์ ํฌ๊ฐ ๊ณ ๊ฐ๋ค์ ์ฌ์ฉ ์ฌ๋ก๋ฅผ ์ดํด๋ณด๋ฉฐ ๋ถ์ํ ๊ฒฐ๊ณผ๋ฅผ ๋ง์๋๋ฆฐ ๋ฐ ์๋๋ฐ, ์ถ๋ก (inferencing)์ ์๋นํ ์๊ฐ์ด ์์๋๊ณ ์์์ ํ์ธํ์ต๋๋ค. ์ด๋ ์์ง ์ถ์๋์ง ์์ ๋ง์ ์์ฑํ AI ์ ํ๋ฆฌ์ผ์ด์ ๋ค์ด ๋์ค๊ธฐ ์ ์ ์ํฉ์ ๋๋ค. ๋ฐ๋ผ์ ์ถ์ฒ ์์ง(recommender engines)์ด ํ์ฌ ์ถ๋ก ์์ ์ ๋งค์ฐ ์ค์ํ ๋ถ๋ถ์ ์ฐจ์งํ๊ณ ์์ต๋๋ค. ์์ผ๋ก ๋์๊ฐ๋ฉด์ ์ด 40%์ ์ฑ์ฅ์ ์ง์์ ์ผ๋ก ๋ณผ ์ ์์ ๊ฒ์ผ๋ก ์์๋ฉ๋๋ค. ํ์ง๋ง ์์ ๋ง์๋๋ฆฐ ๋ฐ์ ๊ฐ์ด, ์ ํฌ๋ ์ฌ์ ํ ์ถ๋ก ๋ถ์ผ์์ ๊ฐ์ฅ ํฐ ์ ์ ์จ์ ์ฐจ์งํ๊ณ ์์ต๋๋ค. |
And when we think about the Blackwell architecture, particularly the GB200, NVL that is an important configuration that is 30X Improvement in terms of inferencing performance from our current generation. That is such an important piece for many of the customers. They will likely use at the very onset building, what they need for their foundational model. But that important part of inferencing going forward with Blackwell is - has been very well received by many of our customers. Tim Arcuri Great. And then, you talked about some constraints on Blackwell, and it sounds like they're going to go away maybe mid-โ25 that they begin to ease. Can you talk about what some of those are? | ๊ทธ๋ฆฌ๊ณ Blackwell ์ํคํ
์ฒ, ํนํ GB200, NVL์ ๋ํด ์๊ฐํด๋ณด๋ฉด, ์ด๋ ํ์ฌ ์ธ๋ ๋๋น ์ถ๋ก ์ฑ๋ฅ์์ 30๋ฐฐ ๊ฐ์ ์ ๋ณด์ฌ์ฃผ๋ ์ค์ํ ๊ตฌ์ฑ์
๋๋ค. ์ด๋ ๋ง์ ๊ณ ๊ฐ๋ค์๊ฒ ๋งค์ฐ ์ค์ํ ์์์
๋๋ค. ๊ณ ๊ฐ๋ค์ ์ด๊ธฐ์ ํ์ด๋ฐ์ด์
๋ชจ๋ธ์ ํ์ํ ๊ฒ์ ๊ตฌ์ถํ๋ ๋ฐ ์ด๋ฅผ ํ์ฉํ ๊ฐ๋ฅ์ฑ์ด ๋์ต๋๋ค. ํ์ง๋ง ํฅํ Blackwell๊ณผ ํจ๊ปํ๋ ์ถ๋ก ์ ์ค์ํ ๋ถ๋ถ์ ๋ง์ ๊ณ ๊ฐ๋ค๋ก๋ถํฐ ๋งค์ฐ ์ข์ ๋ฐ์์ ์ป๊ณ ์์ต๋๋ค. Tim Arcuri ์ข์ต๋๋ค. ๊ทธ๋ฆฌ๊ณ Blackwell์ ์ผ๋ถ ์ ์ฝ์ฌํญ์ ๋ํด ๋ง์ํ์ จ๋๋ฐ, 25๋ ์ค๋ฐ๊ฒฝ์ ์ํ๋๊ธฐ ์์ํ ๊ฒ ๊ฐ๋ค๊ณ ๋ค๋ ธ์ต๋๋ค. ๊ทธ๋ฌํ ์ ์ฝ์ฌํญ๋ค์ด ๋ฌด์์ธ์ง ๋ง์ํด ์ฃผ์ค ์ ์๋์? |
And maybe is that right to assume that they do sort of begin to go away in the middle of โ25 Colette Kress When we think through the building of Blackwell and the designing and working with our customers on the configuration, the demand Cane's Fast and Furious in terms of what demand is exceptional. And we are working with a tremendous set of partners. We talk with our suppliers each and every day to help them. So right now, yes, we need to scale to build enough Blackwell for what we see in the demand in front of us. And we are probably going to be supply constrained pretty much through the first parts of the new fiscal year. When you say in terms of where there is constraint? | ๊ทธ๋ฆฌ๊ณ ํน์ ๊ทธ๋ฐ ์ ์ฝ๋ค์ด 25๋
์ค๋ฐ์ฏค๋ถํฐ ํด์๋๊ธฐ ์์ํ๋ค๊ณ ๊ฐ์ ํ๋ ๊ฒ์ด ๋ง์๊น์? **Colette Kress** Blackwell ๊ตฌ์ถ๊ณผ ์ค๊ณ, ๊ทธ๋ฆฌ๊ณ ๊ณ ๊ฐ๋ค๊ณผ์ ๊ตฌ์ฑ ์์ ์ ์งํํ๋ฉด์ ์๊ฐํด๋ณด๋ฉด, ์์๊ฐ ์ ๋ง ๋น ๋ฅด๊ณ ๊ฒฉ๋ ฌํฉ๋๋ค. ์์๊ฐ ์ ๋ง ์์ธ์ ์ธ ์ํฉ์ ๋๋ค. ๊ทธ๋ฆฌ๊ณ ์ ํฌ๋ ์์ฒญ๋ ํํธ๋๋ค๊ณผ ํจ๊ป ์์ ํ๊ณ ์์ต๋๋ค. ์ ํฌ๋ ๊ณต๊ธ์ ์ฒด๋ค๊ณผ ๋งค์ผ ๋ํํ๋ฉฐ ๊ทธ๋ค์ ์ง์ํ๊ณ ์์ต๋๋ค. ๋ฐ๋ผ์ ํ์ฌ๋ก์๋, ๋ค, ์ ํฌ ์์ ๋ณด์ด๋ ์์์ ๋ง์ถฐ ์ถฉ๋ถํ Blackwell์ ๊ตฌ์ถํ๊ธฐ ์ํด ๊ท๋ชจ๋ฅผ ํ๋ํด์ผ ํฉ๋๋ค. ๊ทธ๋ฆฌ๊ณ ์ ํฌ๋ ์๋ง๋ ์ ํ๊ณ์ฐ๋ ์ด๋ฐ ๋ด๋ด ๊ณต๊ธ ์ ์ฝ์ ๋ฐ์ ๊ฒ์ผ๋ก ๋ณด์ ๋๋ค. ์ด๋์ ์ ์ฝ์ด ์๋์ง์ ๋ํด ๋ง์ํ์๋ ๊ฑด๊ฐ์? |
It depends in terms of the configurations, but some of the challenges that you have are working again in terms of the co-op space. So the work that you need to do in terms of the different configurations and all of the work that we do in terms of the Networking those switching to get that right. Depending on those configurations that can be supply constrained. But we will have right out of the gate in terms of this quarter, we are on track to ship Blackwell. Blackwell is doing just fine and we're very excited to bring multiple configurations to our customers this quarter. Tim Arcuri Well, I think it's going to be an amazing year next year for sure. So, anyway, thank you, Colette. | ๊ตฌ์ฑ์ ๋ฐ๋ผ ๋ค๋ฅด์ง๋ง, ๋ช ๊ฐ์ง ๊ณผ์ ๋ค์ด ์์ต๋๋ค. ํนํ co-op ๊ณต๊ฐ์์ ์์
ํ ๋ ๋ฐ์ํ๋ ๋ฌธ์ ๋ค์ด์ฃ . ๋ค์ํ ๊ตฌ์ฑ์ ๋ํ ์์
๊ณผ ๋คํธ์ํน, ์ค์์นญ์ ์ฌ๋ฐ๋ฅด๊ฒ ๊ตฌํํ๊ธฐ ์ํ ๋ชจ๋ ์์
๋ค์ด ํ์ํฉ๋๋ค. ์ด๋ฌํ ๊ตฌ์ฑ์ ๋ฐ๋ผ ๊ณต๊ธ ์ ์ฝ์ด ์์ ์ ์์ต๋๋ค. ํ์ง๋ง ์ด๋ฒ ๋ถ๊ธฐ๋ถํฐ ๋ฐ๋ก Blackwell ์ถํ ๊ณํ๋๋ก ์งํํ๊ณ ์์ต๋๋ค. Blackwell์ ์์กฐ๋กญ๊ฒ ์งํ๋๊ณ ์์ผ๋ฉฐ, ์ด๋ฒ ๋ถ๊ธฐ์ ๊ณ ๊ฐ๋ค์๊ฒ ๋ค์ํ ๊ตฌ์ฑ์ ์ ๊ณตํ ์ ์๊ฒ ๋์ด ๋งค์ฐ ๊ธฐ๋ํ๊ณ ์์ต๋๋ค. Tim Arcuri: ๋ด๋ ์ ๋ถ๋ช ํ ๋๋ผ์ด ํ ํด๊ฐ ๋ ๊ฒ ๊ฐ์ต๋๋ค. ์ด์จ๋ , Colette, ๊ฐ์ฌํฉ๋๋ค. |
Believe you are appreciated. Colette Kress Yes. Okay. Bye. | ๊ฐ์ฌํฉ๋๋ค. ์ฝ๋ ํธ ํฌ๋ ์ค ๋ค. ์๊ฒ ์ต๋๋ค. ์๋ ํ ๊ณ์ธ์. |
โข ์ฃผ์ ์ค์ /์ ๋ง:
- ๋ธ๋์ฐ(Blackwell) ์ํคํ
์ฒ๊ฐ 2024๋
1๋ถ๊ธฐ ์ถ์ ์์ ์ด๋ฉฐ, ๊ณต๊ธ ์ ์ฝ์ 2025๋
์ค๋ฐ๊น์ง ์ง์๋ ์ ๋ง
- ๊ทธ๋ก์ค๋ง์ง์ ๋ธ๋์ฐ ์ด๊ธฐ ์ถํ๋ก 71-72.5%๋๋ก ํ๋ฝํ๋ค๊ฐ ์ฐ๋ง์๋ ์ค๋ฐ 70%๋๋ก ํ๋ณต ์์
- ์ํํธ์จ์ด ์ฌ์
์ฐ๊ฐ ๋งค์ถ์ด 20์ต ๋ฌ๋ฌ ๊ท๋ชจ๋ก ์ฑ์ฅ ์ค
โข ์์ ๋ํฅ:
- ๋ฐ์ดํฐ์ผํฐ/AI ์์๋ ์ฌ์ ํ ๊ฐ์ธ์ด๋ฉฐ ๊ณต๊ธ ๋ถ์กฑ ์ํ ์ง์
- ์ถ๋ก (Inference) ๋ถ๋ฌธ์ด ์ ์ฒด ๋งค์ถ์ ์ฝ 40% ์ฐจ์งํ๋ฉฐ ์ฑ์ฅ์ธ
- ๊ธ๋ก๋ฒ ์ ๋ถ/๊ธฐ์
๋ค์ ์๊ตญ AI ๊ตฌ์ถ ์์๊ฐ ์๋ก์ด ์ฑ์ฅ ๋๋ ฅ์ผ๋ก ๋ถ์
โข ๋ฆฌ์คํฌ/๊ธฐํ ์์ธ:
- ๋ธ๋์ฐ ์ด๊ธฐ ์์ฐ์ ๋ฐ๋ฅธ ๋ง์ง ์๋ฐ์ด ์์ผ๋ ์ฅ๊ธฐ์ ์ผ๋ก๋ ํ๋ณต ์ ๋ง
- ์ฃ์ง ์ปดํจํ
, ๋ก๋ณดํฑ์ค ๋ฑ ์ ๊ท ์์ฅ ์ง์ถ๋ก ์ฑ์ฅ ๊ธฐํ ํ๋
- ํ๊ธ ๋ณด์ ๊ณ ๊ธ์ฆ์ ๋ฐ๋ฅธ ํจ์จ์ ์๋ณธ ๋ฐฐ๋ถ์ด ๊ณผ์ ๋ก ๋๋
Original | Translation |
---|---|
Question-and-Answer Session | ์ง์์๋ต ์ธ์ |
I don't see any transcript provided to summarize. Please share the earnings call transcript you'd like me to analyze, and I'll provide a clear 3-5 bullet point summary in Korean focusing on:
- ์ฃผ์ ์ฌ๋ฌด ์งํ
- ๊ฐ์ด๋์ค ์
๋ฐ์ดํธ
- ๊ฒฝ์์ง์ ํค
- ์ ์ฌ์ ๋ฆฌ์คํฌ์ ๊ฐ์