As expected, the event Microsoft Ignite has been full of news related to artificial intelligence. From the introduction of the Maia 100 chip for data centers, to the facelift for Bing Chat, those from Redmond did not miss the opportunity that they are fully committed to the new technology. Although perhaps the loudest comment was made by Jensen HuangCEO of NVIDIAwho assured that Generative AI will be even more disruptive than the Internet.
The leader of the Californian company took advantage of his presence on stage to talk about the new collaborations between his company and Microsoft. The most notable announcement was the launch of the new AI Foundry Service from NVIDIA, which will run on Azure and allow customers to create their own massive language models and customize them according to their needs.
As its name indicates, what the green giant will offer will be an artificial intelligence foundry service. The company wants to be to generative AI what TSMC is to the world of semiconductors, basically. And at NVIDIA they are convinced that the furor that technology is experiencing these days is just the tip of the iceberg.
“Generative AI is the most important platform transition in the history of computing,” Huang said. The executive director of NVIDIA also maintained that no other technological leap that has occurred in recent decades is similar to what we are about to experience. “In the last 40 years, nothing has been so big. It is bigger than the PC, it is bigger than the mobile phone and it will be, by far, bigger than the Internet”he remarked.
NVIDIA puts its strongest chips in generative AI
That NVIDIA turns a large part of its business strategy to generative AI is far from surprising. Those from Santa Clara, California, have been the main beneficiaries because of the furor over ChatGPT, DALL-E, Midjourney, Stable Diffusion, and countless tools that take advantage of this technology.
Your accelerator plates H100, which cost around $40,000 each and are used to train artificial intelligence models, have become an extremely prized commodity in Silicon Valley and around the world. OpenAI, for example, acknowledged that this year it suffered from the inability to obtain the number of units it expected of said hardware. Although I was hoping the story would improve in 2024. It will be interesting to see what happens with the new and more powerful H200.
Not even the veto on the export of artificial intelligence chips to China has had a negative impact on NVIDIA. In fact, Tencent recently said it has secured enough stock of H800 GPUs to continue development of its Hunyuan generative AI model for at least “a couple more generations.”
It is also worth mentioning that this year NVIDIA has managed to get into the select group of companies valued at more than 1 billion dollars. In fact, today it is sixth among the most valuable companies in the world, behind Apple, Microsoft, Saudi Aramco, Google and Amazon. And although generative AI is not its only business unit, it is the one that has given it the most push this year.
Last August, the company reported revenue of $13.5 billion during the second quarter of fiscal year 2024. Of this figure, $10.32 billion corresponded to the sale of data center hardware, double compared to the previous quarter. But that was not all. The most brutal thing was that it signed profits of 6,188 million dollars in that quarter alone; that is, a year-on-year increase of more than 840%. How can we not want generative AI to continue growing, then?