Europe sets its sights on multi-billion-euro gigawatt AI factories
Data storage tapes are kept at the National Energy Research Scientific Computer Center (NERSC) facility at the Lawrence Berkeley National Laboratory, which will be supported by Nvidia’s upcoming Vera Rubin Chips.
Manuel Orbegozo | Reuters
Europe sets its views on Gigawatt factories to support the delayed artificial intelligence industry and meet the challenges of a rapidly changing sector.
Production artificial intelligence around the concept of industrialized factories, especially in recent months, especially in the last months Nvidia CEO Jensen Huang stressed the importance of infrastructure at an event in June. Huang received a new “industrial revolution” at the GTC Conference in Paris, France. in question The company was trying to help countries to establish AI factories that create income through partnerships in France, Italy and the UK.
The European Union defines factories as a “dynamic ecosystem” that brings together calculation power, data and abilities to create AI models and applications.
The block was a delay behind the US and China to scale artificial intelligence. With 27 members in the union, the region moves more slowly when it comes to understanding the new legislation. A grille in the need for higher energy costs, permission delays and terrible modernization can prevent developments.
The European Commission’s Deputy General Manager of Technology Sovereignty, Henna Virkkuen, told CNBC that Bloc’s goal is to bring high quality data sets, information processing capacity and researchers together.
“For example, there are 30% more researchers than we have from the United States, focused on AI. We also have approximately 7,000 initiatives [that] They develop artificial intelligence, but the most obstacle for them is that they have very limited information processing capacity. Therefore, we have decided that we have invested in this very important infrastructure with our member states. ”
These are huge investments because they are four times stronger than the largest AI factories when it comes to calculation capacity.
Henna Virkkuen
European Commission Vice President of Technology Sovereignty
“We have everything necessary to be competitive in this sector, but at the same time we want to create our technological sovereignty and competitiveness.”
So far, the EU has financed 10 billion euros ($ 11.8 billion) for establishing 13 AI factory and 20 billion euros as a starting point for investment in Gigafactories and marked “the largest public investment in the world”. Virkkuen, block 60 sites on the site of 16 member states Gigafactories’e 76 interest, he said.
Virkkuen said that the call for interest to Gigafactories was “overwhelming” and went far beyond the expectations of the Block. However, in order for factories to make a significant contribution to Europe’s information processing capacity, significant more investment from the private sector will need to be made to finance the expensive infrastructure.
‘Intelligence Revolution’
The EU defines facilities as “one point shop” for AI companies. It aims to reflect the process in industrial factories that convert raw materials into goods and services. With a AI factory, raw data enters the entrance and advanced AI products are expected.
According to UBS stock research analyst Andre Kukhnin, it is a data center with additional infrastructure about how to adopt technology.
“Idea is to create a GPU [graphics processing units] Basically, to create data centers with GPUs that can train and inferences models, and to create a infrastructure that allows you to make it accessible for SMEs and parties that cannot only go to them. “He said.
Citi Research Analyst Martin Wilkie is the key to the appointment of the facility as an AI factory.
“You create a platform with these chips with crazy calculation capacity,” he said. “And if you have connected them to a grill that can really get the power to use them to full capacity, then the world is on your feet. You have this tremendous ability to do something, but it will be defined by what you use for your success.”
Telecommunication company Telbavar In November last year, the AI factory in Norway was released, and it is already investigating the possible use for such facilities. The company currently has a small GPU cluster because it tested the market before the scaling.
Telenor’s chief innovation manager and AI factory Kaaren Hilsen and EVP infrastructure Jannicke Hilland, in front of a nvidia shelf in the company’s AI factory
Telbavar
“The journey began with a belief – Nvidia believed that each country should produce its own intelligence,” Telenor’s chief innovation officer and AI factory Kaaren Hilsen President CNBC. He said.
Hilsen stressed that data sovereignty is the key. “If you want to use AI to make innovation and make the job more efficient, you potentially demonstrate critical and job -sensitive information to these AI models,” he said.
The company works with Babelspeak, which Hilsen describes as a Norwegian version of Chatgpt. Technology translates precise dialogues such as border police and pilot, which cannot use public translation services due to security problems.
Hilsen, “Egemen AI factories can really help to advance the society” a “intelligence revolution” we are experiencing.
Billion-euro investment
Virkkuen said that the region’s first AI factory will be operational in the coming weeks and one of the biggest projects will be launched in Munich, Germany in the first days of September. A different story for Gigafactories.
“These are huge investments because when it comes to information processing capacities, they are four times stronger than the largest AI factories and it means billions of investments. Each of which needs three to five billion. [euros] In the investment,
Bruegel Senior Research Assistant Bertin Martens questioned why such investments have to be subsidized by government funds.
“We don’t know how much special investment is proposed as a complement of the taxpayer subsidy and how big these factories are.
Power consumption is also a key issue. Martens said that building an AI Gigafactory can take one to two years – but it requires much more time to build an energy production of this size.
“If you want to build a state -of -the -art gigafactory with hundreds of thousands of Nvidia Cips, you should rely on at least one Gigawatt power consumption for one of these factories. In these countries, the fact that Europe’s electricity network is enough will require great investment in these factories.”
UBS Estimates The current board will double the global data center capacity and rising demand. Based on the EU’s 20 billion euros investment and each factory’s 100,000-speed processor business plan, UBS estimates that each factory can be 100-150 MW with a total capacity of approximately 1.5-2 GW.
This can add approximately 15% to Europe’s total capacity – according to data, a huge increase even compared to the US, which is currently about one -third of the global capacity.
Following the announcement of the EU-US trade framework, EU Chief Ursula von der leyen in question That the US AI chips will help AI Gigafactories to strengthen AI Gigafactories to help states “protecting technological edges”.
“It can be claimed to be relatively easy, provided that you have money. It is relatively easy to buy chips from Nvidia and create these hardware factories, but it is a completely different question to be operated and make economically applicable.”
The authority said that the EU should probably start on a smaller scale, as it could not build its own border models immediately due to the expenses of the region.
“I think Europe can slowly build its infrastructure and business models around AI to reach this stage, but this will not happen immediately.” He said.
