Jensen Huang, the boss of Nvidia, imagines a future woven of factories of a new type, kinds of unique counters devoted to AI, in order to facilitate the convenience of it while bringing industrial jobs to the United States.
The coal power plants, then electric, were at the heart of the industrial revolution, providing factories with the energy necessary to produce large -scale manufactured goods. Tomorrow, the “AI factories” (“AI Factories”) will be just as necessary to allow companies to take advantage of the new revolution that is coming: that of intelligence. This is in any case the opinion of Jensen Huang, the boss of Nvidia, whose graphics cards (GPU) have become essential to the training and operation of cutting -edge AIs.
In a speech delivered during the Hill and Valley Forum, an event organized in Washington to bring together Elites of Silicon Valley and political decision -makers, Jensen Huang said that each American company would soon need to use an AI factory to set up innovative solutions, which it operates in finance, health, industry or automotive. In the words of the boss of Nvidia: “Tomorrow, each car built, each product made will have its digital counter. An AI factory will therefore be necessary to design the models that will run in the vehicle.”
Jensen Huang’s vision
What is the difference between these future AI factories and today’s data centers? For Jensen Huang, where they are content to store and process information, IA factories will form complex IT infrastructure capable of generating intelligence in real time. They will combine servers, GPU, DPU (data processing units) and other specialized materials, associated with advanced storage and data management solutions.
Neither simple physical factories nor conventional data centers, these platforms will manufacture intelligence from the masses of data, by orchestrating the whole life cycle of AI, the ingestion of data to training, adjustment and inference in real time. Just as a traditional factory uses raw materials to transform them into finished products, an AI factory transforms raw data into intelligence: predictions, responses and lighting that feeds professional decisions.
“The right way to interpret all of this is to realize that AI will give enormous productivity increases. We will have whole data centers that will turn to make quantum physics, develop new materials, new energy sources, new batteries, help governments optimize their budgets … Data centers that will neither training nor inference, but a mixture of the two. Cons up Antoine Chkaiban, analyst at New Street Research, a market for market.
A new industrial revolution?
The vision of Jensen Huang translates a conception of the economy of tomorrow where AI will be a service and intelligence a convenience. “The message is clear: AI is no longer just a tool at the service of automation and efficiency-it is an essential infrastructure that will dictate economic growth, technological domination and competitive advantages in the coming years,” said Senthil Ravindran, of conceptvines, a consultant for start-ups. “Nvidia is not content to sell powerful fleas. It lays the foundations for a world where intelligence will be made on a large scale, such as electricity, steel and computer infrastructure has fueled previous industrial revolutions.”
For Jensen Huang’s company, the distinction is highly strategic. Beyond the power of its GPUs, for the moment unrivaled, Nvidia derives its strength from its ecosystem, which society has patiently developed for twenty years already. An ecosystem which includes communities of developers who help each other and design bookstores; companies that cause large models of AI and smaller models; And size clients like GAFAMs that set up the software bricks necessary to deploy NVIDIA GPUS in their data centers.
The vision of the factories of AI, where the Nvidia ecosystem will undoubtedly play a key role, illustrates a desire to go a notch further. The world leader in AI fleas is thus imposed as the spine of the future intelligence revolution, by building a complete nervous system mixing both software, hardware and industrial infrastructure to allow any business to “plug” on its ecosystem to draw all the benefits of this rupture technology. Jensen Huang is considering in short no more or less than being the Thomas Edison of AI.
“By 2035, the request for computer power for AI models should increase by a factor 100, which will require an estimated amount of $ 1,000 billion in investments in data centers specializing in AI by 2030. This tilting does not only aim to improve efficiency – it is a question of transitioning an economy of knowledge towards an intelligence economy, where it is decision -making in real time that will determine who will be the industrial leaders, “said Senthil Ravindran.
A way to wake Donald Trump
Finally, it is impossible to read the recent statements by Jensen Huang without taking into account the current political context in the United States. On the one hand, the American president claims the return of industrial jobs to American soil. On the other hand, he sparked an open trade war against China, and did everything to limit it to the best of American technology on AI. A policy that harms Nvidia, the company can no longer import its chips into the Middle Empire, including less powerful models specially designed to adapt to a previous wave of sanctions. The boss of Nvidia tries in this context of negotiating with Donald Trump to obtain a reduction in sanctions.
Note in this regard that the Hill and Valley Forum, where Jensen Huang recently exposed his vision around AI factories, was co -founded by Jacob Helberg, a member of the Trump administration. Jensen Huang took care to emphasize jobs in the construction and steel industry (one of the battle horses of Donald Trump) that a future Boom of IA factories will create in the United States. “Our country must recognize that manufactured activities constitute a work as respectable as necessary to build a country,” he said, in a declaration that the American president would not have denied.
“I am delighted to see this administration encourage and support industry on American soil,” said the boss of Nvidia recently. “If we do not catch up in this area, we risk dropping an immense economic sector.”
Earlier this year, NVIDIA also said they wanted to start making superordinators for AI fully in the United States. So many announcements that are undoubtedly intended to spare the good graces of the White House in order to have levers of negotiations in the trade war with China. Jensen Huang, who also declared in an interview that decision -makers “should recognize the importance of accelerating, supporting and promoting the US AI in the world,” said elected representatives to discuss export controls on AI fleas.
