The AI revolution is here, and with it comes a new set of demands on the data centers, cloud, and hybrid operations that make it run.
In particular, the recent rise of generative AI, fueled by large language models (LLMs), has proved to be a strain on both storage capacity and the power needed to chew through large amounts of data at speed.
Currently, the data industry is responding well to these challenges but, across Europe, businesses are struggling to keep pace with their competitors in North America and Asia. Why?
To date, according to Pitchbook, $122 billion has been invested in Generative AI businesses headquartered in the US, compared to a modest $3.8 billion invested in the Generative AI businesses based in Europe.
This stark gap in investment between US and European-headquartered companies can be put down to two challenges faced by European players in the generative AI market: a lack of GPU investment in the region, and a regulatory system that has created an atmosphere in which organizations fear using their data.
European GPU investment is lagging
The rise in AI has led to much larger requirements for compute power, which in turn has seen a huge spike in demand for GPU chips.
The issue, particularly in Europe, is the lack of GPU infrastructure to drive the AI revolution. For example, the UK government has set aside £100 million ($127m) to buy Nvidia GPU chips in order to power a new national AI supercomputer.
To put this in perspective, though, Elon Musk has announced an order for GPUs worth almost double the pledge from the UK government to help power a new AI project at X (formerly Twitter), whilst it has been reported that China has put in an order worth around $5 billion to acquire high-end Nvidia GPU Chips. Whichever way you look at it, the investment in GPU chips in Europe is behind the rest of the world.
EU lawmakers have set out that they wish to double the bloc’s share of global chip output, from 10 percent to 20 percent, in the next seven years.
It is hoped that investment from governments will help to meet the growing demand for GPUs in the region and ease the cost of chips such as the Nvidia H100.
However, the AI revolution is already here and, if they want to stay competitive, European businesses simply can’t afford to wait for nearly a decade until GPU availability and investment increases.
The red-tape hurdle
A lack of investment isn’t the only reason European businesses are falling behind in the AI race: stringent regulations are also providing a tough hurdle for organizations to overcome.
Generative AI is built on large language models, and the data used to train these models falls under various regulations such as GDPR. This presents a problem for European companies, who must comply with some of the world’s most restrictive data privacy regulations.
In fact, there is some evidence to suggest that AI operations could put data at a greater risk than normal: Gartner revealed last year that 40 percent of organizations had an AI privacy breach and that, of those breaches, only one in four was malicious.
This points to problems with mismatched data privacy expectations due, in many cases, to misaligned data regulations.
So, European businesses currently face two huge challenges as they struggle to keep pace with the international race to innovate with AI: a lack of investment in GPUs and strict regulations. What is the solution?
Cloud sovereignty for AI
The alternative could lie in the cloud. GPU Cloud providers offer an accessible way for businesses to unlock the power of chips such as NVIDIA H100s without the prohibitive start-up costs by giving European companies access on-demand.
This solves the scarcity dilemma, but what about the regulatory issues? This is where sovereign cloud solutions really come into their own.
Microsoft led the way with the concept of cloud data sovereignty at the beginning of this year with the launch of a new solution for public sector cloud users who needed to be able to guarantee that users’ data is stored and processed in a specific region.
The ability to select a data center region to meet regulatory requirements is an important step forward, and could hold the key to unlocking cloud-based GPU for European organizations.
The standards for sovereign clouds can, of course, vary widely depending on their location.
Whilst clouds physically based in datacenters in America or parts of Asia may have relatively relaxed security measures, those based within the European Economic Area enforce some of the world’s strongest protections.
For European organizations looking to get the most out of the AI revolution, cloud GPU solutions based in Europe are the natural gold standard.
By choosing a provider that complies with data protection laws within the European Economic Area, businesses will be using a solution which will ensure that each subscriber’s data and metadata is protected from foreign access and stored in compliance with strong EU regulations.
Europe's AI competitiveness is at risk if access to the highest-powered GPUs comes with the risk of non-compliance with key data privacy laws.
Cloud GPU providers need to ensure that, in granting access to the compute power necessary to meet the demand of a burgeoning AI industry in Europe, they are not compromising data protection.
Sovereign cloud solutions are the necessary next step in the AI revolution; without them, European businesses cannot compete on the international stage.
More in AI & Analytics
-
-
Sponsored Together in Electric Dreams
-