본문 바로가기

마이크로소프트 -엔비디아-AMD-인텔

"Nvidia의 긴급 주문 폭증으로 TSMC의 5나노미터 제조 공정의 가동률 급증(2023.05.26)

22023.05.26

Nvidia rush orders lifting TSMC 5nm fab utilization (digitimes.com)

Nvidia의 긴급 주문 폭증으로 TSMC의 5나노미터 제조 공정의 가동률 급증

 

An influx of Nvidia orders requiring super hot runs (SHR) have shored up capacity utilization rates for TSMC's 5nm process platform to almost full, according to industry sources.

Nvidia의 급작스러운 주문으로 TSMC의 5나노미터 제조 공정의 가동률이 거의 가득 차게 되었다고 

업계 소식통들은 전했습니다.

 

Nvidia is placing rush orders with TSMC for its H100 and A100, as well as the H800 and A800 designed specifically for China, the sources said. TSMC's SHR for Nvidia's AI GPUs will last throughout the year, the sources added.

소식통에 따르면 Nvidia는 H100 및 A100과 중국을 위해 특별히 설계된 H800 및 A800을 위해 TSMC에 긴급 주문을 내고 있다고 합니다. 업계 소식통은 Nvidia의 인공지능 그래픽 처리 장치에 대한 TSMC의 SHR 작업은 올해 내내 계속될 것이라고 덧붙였습니다.

 

 

The Nvidia orders have already lifted the utilization rates of TSMC's 7/6nm process family, which had previously been falling fast amid weakening demand, and the foundry house's utilization rates for the 5/4 processes are also rising fast, reaching almost full capacity, the sources said.

 

소식통에 따르면 Nvidia의 주문으로 인해 TSMC의 7/6나노미터 공정 패밀리의 이용률이 이전에 약화된 수요 속에서 

급속히 하락하던 상황에서 이미 회복되었으며, 5/4나노미터 공정에 대한 TSMC의 이용률도 급증하며

완전 가동률에 도달하게 되었다고 합니다.

 

Announcing the company's results for the first quarter of financial 2024 (ended April 30, 2023), Nvidia CEO Jensen Huang noted that the computer industry is undergoing two transitions at the same time, namely accelerated computing and generative AI.

 

Nvidia CEO 제이슨 황은 2024회계년도 1분기(2023년 4월 30일 종료)의 회사 실적 발표에서 컴퓨터 산업이 

가속화된 컴퓨팅과 생성 AI라는 두 가지 전환을 동시에 겪고 있다고 언급했습니다.

 

"A trillion dollars of installed global data center infrastructure will transition from general purpose to accelerated computing as companies race to apply generative AI into every product, service and business process," said Huang, as cited by a company press release.

 

"Our entire data center family of products -- H100, Grace CPU, Grace Hopper Superchip, NVLink, Quantum 400 InfiniBand and BlueField-3 DPU -- is in production.

We are significantly increasing our supply to meet surging demand for them."

 

"Nvidia CEO 제이슨 황은 회사 보도자료에 인용된 내용으로 "세계적으로 설치된 1조 달러 규모의 데이터 센터 인프라는 

기업들이 생성 AI를 모든 제품, 서비스 및 비즈니스 프로세스에 적용하기 위해 일반용도에서 가속화된 컴퓨팅으로 전환될 것입니다"라고 말했습니다.

 

 "우리의 전체 데이터 센터 제품인 H100, Grace CPU, Grace Hopper Superchip, NVLink, Quantum 400 InfiniBand 및 BlueField-3 DPU는 현재 생산 중에 있습니다. 우리는 수요 급증에 대응하기 위해 공급량을 크게 늘리고 있습니다."

 

Nvidia is riding the AI wave triggered the arrival of ChatGPT. The company has an overwhelming presence in the AI GPU market. In the wake of the US chip sanctions against China, Nvidia quickly introduced scaled down versions of its H100 and A100 chips for the country -- the H800 and A800 -- to circumvent the ban. Demand from major Chinese firms, such as Baidu, for the cut-down versions has been strong, the sources said.

 

ChatGPT의 등장으로 인해 Nvidia는 AI 파동에 탑승하고 있습니다. 이 회사는 AI GPU 시장에서 압도적인 입지를 가지고 있습니다. 중국에 대한 미국의 칩 제재 이후, Nvidia는 신속하게 H100과 A100 칩의 축소판인 H800과 A800을 중국에서의 금지 조치를 피하기 위해 도입했습니다. 주요 중국 기업인 바이두(Baidu)를 포함한 중국 기업들로부터 축소판에 대한 수요가 강력하다고 소식통들은 전했습니다.

 

Nvidia has reported impressive financial results for its first quarter of fiscal 2024, with revenue rising 19% sequentially to reach US$7.19 billion. It has also provided a rosy outlook for the second quarter.

 

Nvidia는 2024회계년도 1분기에 대한 인상적인 재무결과를 발표했습니다. 매출액은 전분기대비 19% 증가하여

71.9억 달러에 이르렀습니다. 또한, 2분기에 대해 밝은 전망을 제시했습니다.

 

Even Intel CEO Pat Gelsinger has admitted that Nvidia is in a strong position to capture the AI opportunities.

Nvidia's strong performance is benefiting its chipmaking partner, TSMC. TSMC previously was only expecting significant amounts of orders coming from Apple supporting the vendor's launch of new devices, and it was not expecting the influx of Nvidia's rush orders, the sources said.

 

인텔(Intel)의 CEO인 팻 겔싱거도 Nvidia가 AI 기회를 포착하기에 강력한 입지에 있다고 인정했습니다.
Nvidia의 강력한 성과는 칩 제조 파트너인 TSMC에도 혜택을 주고 있습니다.

TSMC는 이전에는 애플(Apple)의 신제품 출시를 지원하기 위한 주문이 주로 예상되었지만,

Nvidia의 긴급 주문 폭증을 예상하지 못했습니다. 이는 소식통들에 의해 전해졌습니다.

 

At its investors conference in April 2023, TSMC said its first-quarter 2023 business was impacted by weakening macroeconomic conditions and softening end market demand, which led customers to adjust their demand accordingly. It said the company expected business would continue to be impacted by customers' further inventory adjustment in the second quarter.

 

2023년 4월에 개최된 TSMC의 투자자 회의에서, TSMC는 2023년 1분기 사업이 경기 약세와 

최종 소비자 시장 수요의 약화로 영향을 받았으며, 이로 인해 고객들이 수요를 조정하게 되었다고 밝혔습니다. 

또한, 회사는 2분기에 고객들의 재고 조정으로 인해 사업이 계속해서 영향을 받을 것으로 예상했습니다.

 

Market observers said TSMC may revise its guidance upward for the entire 2023 at its July investors conference.

시장 전문가들은 TSMC가 2023년 전체에 대한 전망을 7월 투자자 회의에서 상향 조정할 수도 있다고 말했습니다.

 

---------------------------------------

 

Nvidia CEO highlights accelerated computing and AI’s role in chip manufacturing at ITF World 2023 | VentureBeat

 

In his keynote address at the ITF World 2023 semiconductor conference, Jensen Huang, founder and CEO of Nvidia, emphasized the profound impact of accelerated computing and artificial intelligence (AI) in the chip manufacturing industry. In his video presentation, Huang provided a comprehensive overview of the latest computing advancements propelling industries worldwide.

 

Nvidia의 창립자이자 CEO인 젠슨 황은 ITF World 2023 반도체 컨퍼런스에서 기조 연설을 통해 가속화된 컴퓨팅과 인공지능(AI)이 반도체 제조 산업에 미치는 깊은 영향을 강조했습니다. 황은 비디오 프레젠테이션에서 전 세계 산업을 주도하는 최신 컴퓨팅 기술의 종합적인 개요를 제공했습니다.

 

Huang highlighted the potential of Nvidia’s accelerated computing and AI solutions in chipmaking. He emphasized their intersection with semiconductor manufacturing. He also stressed the need for a new approach to meet the rising demand for computing power while addressing concerns regarding net-zero goals.

 

황은 Nvidia의 가속 컴퓨팅과 AI 솔루션의 반도체 제조 분야에서의 잠재력을 강조했습니다.

그는 이들이 반도체 제조와의 교차점을 가지고 있음을 강조했습니다. 또한, 컴퓨팅 파워에 대한 수요 상승에 대응하면서

순수 영향 목표에 대한 우려를 다루기 위해 새로운 접근 방식이 필요하다고 강조했습니다.

 

“We are experiencing two simultaneous platform transitions — accelerated computing and generative AI,” Huang said. “I am thrilled to see Nvidia accelerated computing and AI in service of the world’s chipmaking industry.”

As an example of how AI and accelerated computing are transforming the technology industry, Huang explained that to achieve advanced chip manufacturing, over 1,000 precise steps must be executed to create features the size of a biomolecule, with each step executed perfectly to ensure functional output.

 

황은 "우리는 가속화된 컴퓨팅과 생성 AI라는 두 가지 동시적인 플랫폼 전환이 진행되고 있습니다,"라고 말했습니다. "나는 Nvidia의 가속화된 컴퓨팅과 AI가 세계의 반도체 제조 산업에 기여하는 것을 보는 것에 감격합니다."

인공지능과 가속화된 컴퓨팅이 기술 산업을 변화시키는 예로, 황은 고급 반도체 제조를 위해서는 생물 분자 크기의 특징을 생성하기 위해 1,000개 이상의 정밀한 단계가 실행되어야 하며, 각 단계가 완벽하게 실행되어 기능적인 결과물이 보장되어야 한다고 설명했습니다.

 

“Sophisticated computational sciences are performed at every stage to compute the features to be patterned and to do defect detection for in-line process control,” Huang said. “Chip manufacturing is an ideal application for Nvidia accelerated and AI computing.”

 

황은 "패턴화될 특징을 계산하고 인라인 공정 제어를 위한 결함 검출을 수행하기 위해 모든 단계에서 정교한 계산 과학이 수행됩니다,"라고 말했습니다. "반도체 제조는 Nvidia의 가속화된 컴퓨팅과 AI 컴퓨팅에 이상적인 응용 분야입니다."

 

Furthermore, Huang recognized that while the exponential performance growth of central processing units (CPUs) had primarily driven the technology industry for almost four decades, CPU design has reached a state of maturity, resulting in a slowdown in the rate at which semiconductors enhance their power and efficiency.

 

또한, 황은 중앙 처리 장치(CPU)의 폭발적인 성능 성장이 거의 40년 동안 기술 산업을 주도해왔지만, CPU 설계가 성숙 상태에 도달하여 반도체가 전력과 효율성을 향상시키는 속도가 저하되고 있다는 사실을 인정했습니다.

 

Huang noted Nvidia’s pioneering efforts in accelerated computing, a groundbreaking approach that combines the parallel processing capabilities of graphics processing units (GPUs) with CPUs. Nvidia, he said, is well suited to tackle today’s computational science challenges, and his company’s accelerated computing is fueling the AI revolution.

 

황은 Nvidia의 가속화된 컴퓨팅에 대한 개척적인 노력을 언급했습니다. 

이는 그래픽 처리 장치(GPU)의 병렬 처리 능력과 CPU를 결합한 혁신적인 접근 방식입니다. 

그는 Nvidia가 오늘날의 계산 과학적인 도전에 잘 대처할 수 있으며, 

그의 회사의 가속화된 컴퓨팅이 AI 혁명을 촉진하고 있다고 말했습니다.

 

 

He cited several examples of how Nvidia GPUs are increasingly crucial in chip manufacturing. Companies like D2S, IMS Nanofabrication and NuFlare are using Nvidia GPUs to accelerate pattern rendering and mask process correction in the creation of photomasks — stencils used to transfer patterns onto wafers using electron beams.

 

그는 Nvidia GPU가 반도체 제조에서 점점 중요해지고 있는 몇 가지 사례를 언급했습다.

D2S, IMS Nanofabrication, NuFlare와 같은 기업들은 Nvidia GPU를 사용하여 광학 마스크를 만들 때 패턴 렌더링과

마스크 프로세스 보정을 가속화하고, 전자 빔을 사용하여 웨이퍼에 패턴을 전달하는 데 사용되는 스텐실을 생성합니다.

 

Meanwhile, semiconductor manufacturers KLAApplied Materials and Hitachi High-Tech are incorporating Nvidia GPUs into their e-beam and optical wafer inspection and review systems.

 

“We have already accelerated the processing by 50 times,” Huang said. “Tens of thousands of CPU servers can be replaced by a few hundred Nvidia DGX systems, reducing power and cost by an order of magnitude.”

In March, Nvidia launched cuLitho, a software library that offers optimized tools and algorithms for computational lithography, accelerated by GPUs.

 

한편, KLA, Applied Materials, Hitachi High-Tech와 같은 반도체 제조업체들은 Nvidia GPU를 전자 빔 및 광학 웨이퍼 검사 및 검토 시스템에 통합하고 있습니다.

황은 "우리는 이미 처리 속도를 50배 가속화했습니다,"라고 말했습니다. "수십만 개의 CPU 서버를 몇 백 개의 Nvidia DGX 시스템으로 대체함으로써 전력과 비용을 몇배 이상 절감할 수 있습니다."

Nvidia는 cuLitho라는 소프트웨어 라이브러리를 출시하여 GPU를 활용한 계산 리소그라피에 

최적화된 도구와 알고리즘을 제공했습니다.

 

 

Huang stressed the far-reaching influence of AI and accelerated computing, extending beyond chip manufacturing to permeate the entire technology industry. He acknowledged that the concurrent shifts in accelerated computing and generative AI are shaping the future of the technological landscape.

 

Looking to the future, Huang referred to the next wave of AI as “embodied AI” — intelligent systems capable of understanding, reasoning and interacting with the physical world. He cited robotics, autonomous vehicles,

and chatbots with heightened physical world comprehension as examples of this technology.

 

미래를 바라볼 때, 황은 "구체적 AI(Embodied AI)"라는 다음 AI 파동을 언급했습니다. 

이는 물리적인 세계를 이해하고 추론하며 상호작용할 수 있는 지능 시스템을 의미합니다. 

그는 이 기술의 예로 로봇, 자율주행 차량 및 물리적 세계 이해 능력이 강화된 챗봇을 들었습니다.

 

To demonstrate advancements in embodied AI, Huang unveiled Nvidia VIMA, a multimodal embodied AI system capable of carrying out intricate tasks guided by visual text prompts. Through acquiring concepts, comprehending boundaries and even emulating physics, VIMA signifies a notable progression in AI capabilities.

 

Huang also revealed Nvidia’s Earth-2 project, designed to develop a digital replica of the planet. Earth-2 will forecast weather patterns, provide long-range climate predictions and ultimately contribute to the search for affordable and environmentally friendly energy solutions

.

This endeavor employs FourCastNet, a physics-AI model that rapidly simulates global weather patterns. These systems hold great potential for addressing pressing issues of our era, including the demand for sustainable energy solutions.\

 

“The reactor plasma physics-AI runs on Nvidia AI, and its digital twin runs in Nvidia Omniverse,” Huang said. “Such systems hold promise for further advancements in the semiconductor industry. I look forward to physics-AI, robotics, and Omniverse-based digital twins helping to advance the future of chip manufacturing.”

 

------------------------------------------

Nvidia CEO Jensen Huang says 2023 is a ‘perfect year to graduate’ thanks to A.I. (yahoo.com)

 

In 2023, he continued, graduates will join the workforce just as A.I. has “reinvented computing from the ground up. In every way this is the rebirth of the computer industry.”

“A.I. has opened immense opportunities,” the billionaire CEO continued. “Agile companies will take advantage of A.I. and boost their position,” while firms that don’t will perish. Graduates must “take advantage of A.I. and do amazing things with an A.I. copilot by your side,” he added.

 

"2023년에는 그는 덧붙였다, 졸업생들은 인공지능이 '기반부터 컴퓨팅을 혁신하였다. 

모든 면에서 이것은 컴퓨터 산업의 부활이다'라며 새롭게 직장에 합류할 것이다."

"억만장자 CEO는 계속해서 '인공지능은 엄청난 기회를 열어주었다. 유연한 기업들은 인공지능을 활용하여 

자신들의 위치를 강화할 것이며, 그렇지 않은 기업들은 멸망할 것이다'라고 말했다

그는 졸업생들에게 '인공지능을 활용하고 인공지능 동반 비행사로서 놀라운 일들을 이루어야 한다'고 덧붙였다.

 

A.I. will create “new jobs that didn’t exist before,” such as prompt engineering, A.I. factory operations, and A.I. safety engineers, he said. He acknowledged that some jobs will be lost to automation, but added that while people fret about losing their job to A.I., they might actually lose it to someone who’s become more expert at using A.I.

 

IBM CEO Arvind Krishna points to declining populations to calm fears about A.I. taking jobs

Employees doing routine tasks that A.I. can perform is “not an option” given population declines, argues the IBM chief.

fortune.com

 

"인공지능은 "이전에 존재하지 않았던 새로운 직업들"을 만들어낼 것이다. 

예를 들어, 신속한 엔지니어링, 인공지능 공장 운영, 그리고 인공지능 안전 엔지니어 등이 그러한 직업들이다"라고 

그는 말했다. 그는 일부 직업이 자동화로 인해 사라질 것임을 인정했지만, 사람들이 자신의 일을 인공지능에게서 

잃을 것을 걱정할 때, 실제로는 더 뛰어난 인공지능 사용자에게서 일자리를 잃을 수도 있다"고 덧붙였다.

 

Huang urged graduates to “have the humility to confront failure and admit a mistake,” telling them they would “endure pain and suffering.” He described “humiliating and embarrassing” failures he suffered in Nvidia’s early days. For example, Nvidia realized at one point that its “technically poor” architecture was the “wrong strategy,” but it had already won a contract with Sega to help make a new gaming console using it.

 

황 CEO는 졸업생들에게 "실패를 직면하고 실수를 인정하는 겸손함을 가져야 한다"며, 그들은 "고통과 고통을

인내할 것"을 촉구했습니다. 그는 Nvidia 초창기에 겪은 "부끄럽고 수치스러운" 실패들을 설명했습니다.

예를 들어, Nvidia는 한 때 기술적으로 미흡한 아키텍처가 "잘못된 전략"이었음을 깨달았지만,

이미 Sega와의 계약을 획득하여 새로운 게임 콘솔을 만드는 데 도움을 주어야 했습니다.

 

“We had to stop, but I needed Sega to pay us in whole or Nvidia would be out of business. I was embarrassed to ask,” he explained, but to his amazement Sega’s CEO agreed, giving Nvidia “six months to survive.”

Nvidia did survive, becoming a key player first in the gaming industry and now in artificial intelligence.

With the A.I. era here, Huang urged graduates to “run after it like we did. Run, don’t walk.”

 

“우리는 중단해야 했지만, Sega가 우리에게 전액을 지불하지 않으면 Nvidia는 사업을 중단해야 했습니다. 제가 부탁하는 것이 부끄러웠습니다," 그는 설명했습니다. 그러나 놀라운 일로 Sega의 CEO가 동의하여 Nvidia가 "생존하기 위해 여섯 개월의 기간"을 얻게 되었습니다."

Nvidia는 생존하며, 먼저 게임 산업에서 주요한 참가자가 되었고 이제는 인공지능 분야에서도 중요한 위치를 차지하고 있습니다.

A.I. 시대가 왔으므로, 황 CEO는 졸업생들에게 "우리가 한 것처럼 따라가라. 걷지 말고 달리라"고 촉구했습니다.

 

-------------------------------------

2023.05.29

AI로 폭발한 GPU 수요 "美 코로나 때 화장지만큼 구하기 힘들어" | 연합뉴스 (yna.co.kr)

 

AI로 폭발한 GPU 수요 "美 코로나 때 화장지만큼 구하기 힘들어" | 연합뉴스

(서울=연합뉴스) 이도연 기자 = 전 세계적으로 인공지능(AI) 열풍이 불면서 이에 필요한 반도체 수요도 급증했지만, 공급이 수요를 따라가지 못...

www.yna.co.kr

 

엔비디아가 전 세계 공급량 90% 차지…"주문해도 6개월 기다려야"