How to describe Las Vegas to someone who has never been there? The first thing that comes to mind is certainly the Strip, literally a strip of asphalt overlooked by mega hotels have become so iconic. However, limiting the description of Las Vegas to the very visible glitter of hotels such as the Bellagio or the Venetian or the more recent Aria and Cosmopolitan is decidedly reductive, because it limits the image of the city only to its side linked to entertainment. Each mega hotel certainly offers countless recreational activities to the many tourists who visit the city of sin, but at the same time is equipped with infrastructures designed to accommodate very business-oriented events and meetings.
Las Vegas has become the capital of major events
It is no coincidence that an event like CES takes place in Las Vegas and not only at the actual convention center, but literally throughout the city, with spaces dedicated to the fair and its exhibitors distributed among many hotels, each of which has its own convention center, which in terms of size and quality of the structures would compete with many exhibition centers in our country. For years, therefore, Las Vegas has not only been the capital of entertainment, but also of most important corporate events: just to name another, there is AWS re:Invent which, between the end of November and the beginning of December, welcomes over 50,000 people every year, all hosted by the convention center of a single hotel, the Venetian, which is well exploited also from CES, given that Eureka Park is actually a pavilion of that structure. Eureka Park which hosts the diverse world of startups and which we talked about here. To put the scale of the events into context, CES hosted 141,000 visitors this year.
To get an idea of the type of structures that are part of the mega hotels in Las Vegas, one cannot fail to mention the Michelob ULTRA Arena, a 12,000 seat arena, which hosts the games of the Las Vegas women’s basketball team. The fact that a hotel like the Mandalay Bay can “hide” a venue of this size within its facilities makes it very clear how Las Vegas is different from any other city in the USA and why it is so successful in attracting large events. It is in this context that the keynote took place Jensen HuangFounder & CEO of NVIDIA, the most awaited moment of the entire CES, which attracted over 15,000 people who crowded the stands and the pitch of the arena, with kilometer-long queues, even for the press.
Jensen Huang’s vision of the world to come
The expectations that had been created for the event were amply rewarded, because what Jensen presented in a very tight hour and a half and without any breaks gave all the participants a vision of the world to come, in which, according to the CEO of NVIDIA, artificial intelligence will continue to evolve at an incessant pace and will make the decisive leap from the virtual to the physical world.
We have spoken extensively about the announcements related to the new GeForce RTX 5000 family graphics cards based on Blackwell architecture Who and we have described all the details of the architecture applied to graphics cards in this other article. Here we instead want to focus on the vision shared by Jensen because, perhaps never more than in this case, he proposes very broad scenarios that impact, specifically, the 360° evolution of technology and more generally the relationship between technology and any human activity. .
The first element to take into consideration to understand the vision of the CEO of NVIDIA is the evolution that large language models are having, which are the basis of any generative artificial intelligence application. We have entered the third phase of model development, called “test-time scaling” by Jensen, which must be inserted into a model “reasoning” context. In practice, the new models rely on the results achieved with the previous training of the current models to further improve future results. It is an important step, because up to now the improvements have been obtained by training the models on an ever-increasing amount of data, with incredible improvements that we can all notice using, simply, the new versions of the ChatGPT models. The accuracy and reliability of the results obtained, for example with GPT-4o compared to previous versions, is more than surprising, but several voices warn that we have practically run out of data on which to train the models. With “test-time scaling” this plateau can be overcome continue to evolve the modelscreating the conditions for further evolution of the entire generative artificial intelligence ecosystem.
In order for test-time scaling to work, models must be used and therefore tokens must be generated. More tokens are generatedthe more effective the new models will become and here we must comment on Jensen’s attitude during the keynote. Apart from the concepts expressed, his entire speech was characterized by a feeling of urgency, positive, and by the attitude of someone who has a very clear idea in his head about the final goal he wants to achieve, who also knows the road to take to get there, but then he can’t walk the whole way alone.
NVIDIA leads the way, the ecosystem must follow
Jensen “needs” the entire ecosystem to activate so that his vision can materialize. At the same time, in addition to stimulating all the actors involved, NVIDIA is bringing to the market all the tools necessary to make Jensen’s vision real. A very concrete example is NVIDIA Project DIGITIS, described as “a Grace Blackwell AI Supercomputer on your desk”. We give the definition of NVIDIA because it would be simplistic to call it a workstation and it is not a server that can be accessed in the cloud or installed in a data center, because it works locally, so it is really a new category of product, offered at a price, starting from $3,000, that is a fraction of what a solution based on this architecture cost when it was launched. It is a very concrete sign of how hard NVIDIA is pushing for the entire ecosystem to continue developing and experimenting in the AI field. Certainly to sell more products, given that it continues to be the ONLY player on the market to offer functional solutions for certain phases of the development of models and applications in the AI field, but the strong sensation felt at the keynote is that Jensen, having accumulated financial resources practically without limits, is more interested in seeing his vision materialize than in achieving short-term goals. So much so that the stock market, always too focused on immediate results, punished the stock after the keynote.
But what is the overall vision? Let’s take Jensen’s prediction that models will continue to evolve, even if there is no more data to train them on. For the CEO of NVIDIA we are therefore in the agentic AI phase systems designed to act autonomously and proactively to achieve specific objectivesinteracting with the environment in which they operate. Industrial automation systems are therefore being created with autonomous robots that can manage entire production lines, scientific research systems that conduct experiments and analyze the results and advanced personal assistants that, in addition to generating content, can understand complex contexts, plan and act , as in the past, in the virtual world, but also in the real one.
Artificial intelligence enters the real world
The transition to the real world is the next step hypothesized by Jensen and the decisive one to integrate artificial intelligence into every aspect of our lives and see its full potential expressed. We are not going into substantive or ethical considerations here, there is no doubt that AI is not just a new technology but a real paradigm shift. To always quote Jensen, It won’t be AI that steals our jobs, but someone who uses AI…
We are close, but not very close, to Physical AI, which will materialize with the real autonomous driving of vehicles and the spread of all-purpose robots. To speed up the process, Jensen introduced NVIDIA Cosmos, a platform designed to accelerate the development of Physical AI. Cosmos includes state-of-the-art generative models, called World Foundation Models (WFM), along with security systems and an accelerated pipeline for processing video data.
Too often it is said that great passion is necessary to obtain satisfactory results at work. This, which is a concept that has been abused too many times, is decidedly reductive when looking at Jensen on the CES stage. His is a visceral love for GPUs, from which NVIDIA was born and, thanks to his perseverance, he has demonstrated to the whole world the potential of this technology, which is the basis of all the latest developments in artificial intelligence. If the entire ecosystem follows the path traced by Jensen, it is very likely that artificial intelligence will reach its maximum potential. And we will have the opportunity to evaluate, at that point, whether it will have been a positive evolution. In the meantime I recommend everyone to watch the keynote which is available at this link on the NVIDIA YouTube channel.
Source: edge9.hwupgrade.it
