Chip, because Tsmc and Asml don’t envy Nvidia
Nvidia announces a series of supercomputers for artificial intelligence and the stock skyrockets: it exceeds the trillion mark, then settles just below. And the Tsmc and Asml enjoy the rival's successes. All the details
These were days full of important announcements for NVIDIA, the US semiconductor company which on Tuesday reached – even if only for a short time – a market value of one trillion dollars. Since last October, its shares have gained about 200 percent, and yesterday they traded at 401.1 dollars. All thanks to the strong interest of investors in artificial intelligence, which NVIDIA technologies enable: ChatGPT, the famous chatbot developed by OpenAI, for example, uses thousands of graphics processing units created by the company.
THE SUPERCOMPUTER FOR ISRAEL
NVIDIA said Monday that it is building Israel 's most powerful artificial intelligence supercomputer, called Israel-1. It will be one of the world's best-performing computers, with performance up to eight exaflops (one exaflop is 1,000,000,000,000,000,000 calculations per second), capable of "training" algorithms with larger amounts of data in less time.
THE NVIDIA DGX GH200 SUPERCOMPUTER
On the same day, the company announced a new series of supercomputers for artificial intelligence with large memory, the DGX GH200, designed to allow the development of new and more advanced models of generative artificial intelligence (similar to ChatGPT, so to speak).
Jensen Huang, chief executive officer of NVIDIA, said that generative artificial intelligence and large language models "are the digital engines of the modern economy", and that NVIDIA's supercomputers will allow "to expand the frontier of artificial intelligence" . The DGX GH200 are the first computers to combine Grace Hooper chips with the NVLink Switch System, both NVIDIA technologies, to ensure that all the graphics processing units in the computer work together as if they were one, for the benefit of programming simplicity and performance.
GOOGLE, META AND MICROSOFT LEAD THE ROW
According to the company, the first technology companies that will probably have access to the DGX GH200 – available for the end of 2023, it seems – will be the US Google Cloud, Meta and Microsoft.
THE HELIOS SUPERCOMPUTER
NVIDIA is also building a supercomputer for internal use, to be allocated to its research and development teams. Called Helios, it will consist of four DGX GH200 systems and over a thousand Grace Hopper chips. It should be up and running by the end of this year.
ASML AND TSMC HOLD ON TO NVIDIA'S SUCCESS
As Finimize points out, NVIDIA's valuation is currently double that of its main rival, Taiwan's TSMC, and close to the numbers of big tech companies like Apple, Amazon and Alphabet. According to the portal, NVIDIA's shares could even exceed the value of those of Amazon and Alphabet.
NVIDIA, however, isn't the only company benefiting from this explosion in interest in artificial intelligence: TSMC and ASML , a Dutch company that makes advanced microchip machinery, are also gaining. Indeed, after developing the chips for artificial intelligence, NVIDIA has to rely on TSMC for their manufacturing. TSMC, in turn, needs the ASML machines. “When NVIDIA wins,” Finimize writes, so do TSMC and ASML.
HOW IS THE STOCK ON THE STOCK EXCHANGE
On Tuesday, the stock of NVIDIA on the stock market closed at 401.11 dollars, 3 percent more, touching by a whisker – as reported by Reuters – the symbolic value of one trillion dollars. Only four US companies are valued at more than a trillion: Alphabet, Amazon, Apple and Microsoft.
NVIDIA could become the fifth. As Angelo Zini, an analyst at CFRA Research, explained to the agency, "we consider NVIDIA the most important company on the planet in an era that is rapidly changing towards an era emphasized by greater artificial intelligence capabilities". Big tech companies today are very focused on artificial intelligence. Generative models need computers containing very powerful chips called graphics processing units, or GPUs: NVIDIA makes 80 percent of them themselves.
The higher price target for NVIDIA values it at around $1.6 trillion, like Alphabet. Despite its already sky-high valuation, analysts believe the company has room to grow because generative AI technology is only in its infancy, and large-scale adoption will occur in the coming years.
This is a machine translation from Italian language of a post published on Start Magazine at the URL https://www.startmag.it/innovazione/nvidia-intelligenza-artificiale-crescita-azioni/ on Wed, 31 May 2023 08:13:38 +0000.